paper_id
stringlengths 9
16
| version
stringclasses 26
values | yymm
stringclasses 311
values | created
timestamp[s] | title
stringlengths 6
335
| secondary_subfield
sequencelengths 1
8
| abstract
stringlengths 25
3.93k
| primary_subfield
stringclasses 124
values | field
stringclasses 20
values | fulltext
stringlengths 0
2.84M
|
|---|---|---|---|---|---|---|---|---|---|
astro-ph/0210057 | 1 | 0210 | 2002-10-02T08:07:16 | Binary evolution and neutron stars in globular clusters | [
"astro-ph"
] | Improved observations of globular clusters are uncovering a large number of radio pulsars and of X-ray sources. The latter include binaries in which a neutron star or a white dwarf accretes matter from a companion, recycled pulsars, and magnetically active binaries. Most of these sources originate from close encounters between stars in the cluster core; magnetically active binaries and some cataclysmic variables have evolved from primordial binaries. The formation rate through close encounters scales differently with the central density and core radius of the cluster than the probability for a single binary to be perturbed by an encounter. This is exploited in some preliminary observational tests of the close encounter hypothesis.
Accreting black holes have been found in globular clusters with other galaxies; the absence of such holes in the Milky Way clusters is compatible with the small number expected. | astro-ph | astro-ph |
New horizons in globular cluster astronomy
ASP Conference Series, Vol. **, 2002
Eds. G. Piotto, G. Meylan, G. Djorgovski, M. Riello
Binary evolution and neutron stars in globular clusters
Frank Verbunt
Astronomical Institute, Utrecht University, Postbox 80.000, 3508 TA
Utrecht, The Netherlands; email [email protected]
Improved observations of globular clusters are uncovering a
Abstract.
large number of radio pulsars and of X-ray sources. The latter include
binaries in which a neutron star or a white dwarf accretes matter from
a companion, recycled pulsars, and magnetically active binaries. Most of
these sources originate from close encounters between stars in the cluster
core; magnetically active binaries and some cataclysmic variables have
evolved from primordial binaries. The formation rate through close en-
counters scales differently with the central density and core radius of the
cluster than the probability for a single binary to be perturbed by an
encounter. This is exploited in some preliminary observational tests of
the close encounter hypothesis.
Accreting black holes have been found in globular clusters with other
galaxies; the absence of such holes in the Milky Way clusters is compati-
ble with the small number expected.
1.
Introduction
Interest in binaries in globular clusters revived when the first X-ray maps of
>
the whole sky showed an overabundance in globular clusters of bright (Lx ∼
1036 erg/s) sources: whereas globular clusters contain ∼0.1% of the stars of
our galaxy, they contain about 10% of the bright X-ray sources. It was soon
suggested that close stellar encounters in globular cluster cores were responsible
< 1035 erg/s) sources
(Clark 1975). With more sensitive instruments faint (Lx ∼
were detected in globular clusters: 8 in 8 clusters with the Einstein satellite in
the 1980s (Hertz & Grindlay 1983), and 57 in 23 clusters with ROSAT in the
1990s (Verbunt 2001). Of the latter 57, 17 are more than two core radii from
the cluster core. In the new millenium, Chandra detected about 100 faint X-
ray sources in 47 Tuc alone, and 25-40 sources in each of NGC6752, NGC6397,
and NGC6440 (Grindlay et al. 2001a,b, Pooley et al. 2002a,b; Fig. 1). First
XMM results are now being published (Webb et al. 2002; Webb, Gendre in
these proceedings).
Radio pulsars have also been detected in large numbers: 20 in 47 Tuc
alone, 8 in M15, 5 in NGC6752, and some 20 in a dozen other clusters (Freire
et al. 2001, D'Amico et al. 2002, and lists in e.g. Phinney 1992, Lyne 1994, and
on the Web: Freire 2002). Almost all of these are recycled radio pulsars, i.e.
have obtained their rapid rotation (and probably also their low magnetic field)
through accretion of matter and angular momentum from a companion star.
1
2
Verbunt
10 arcsec
12
21
2
4
22
17
23
10
3
18
1
6
11
9
14
15
24
19
716
5
20
13
8
N
E
Figure 1.
X-ray sources
in NGC6440.
Sources 1
certainly and 2, 3 and 5
probably are neutron stars
accreting from a compan-
ion star at a low rate; most
of the other sources are
probably white dwarfs ac-
creting from a companion,
i.e. cataclysmic variables.
Note that the source distri-
bution extends beyond the
core radius (indicated with
a dashed circle), but re-
mains well within the half-
mass radius (solid-line cir-
cle). From Pooley et al.
(2002b).
At the moment 13 bright X-ray sources are known in a total of 12 globular
clusters. The only binaries producing such luminosities are low-mass X-ray
binaries, in which a neutron star or black hole accretes mass from a companion.
Because of the occurrence of X-ray bursts, 12 of the 13 bright sources in clusters
are known to be neutron stars; the 13th one is -- on the basis of its X-ray spectrum
-- probably a neutron star as well (in 't Zand et al. 1999, White & Angelini 2001).
Thus there is no evidence for accreting black holes in any globular cluster with
< 1035 are most likely
our galaxy. The dim sources with 1032.5
∼
< 1032.5) are
neutron stars accreting at a low rate; most of the dimmer sources (
∼
probably cataclysmic variables, but some of them are recycled radio pulsars, or
magnetically active close binaries ('RS CVns') -- see e.g. Fig.8 in Verbunt et al.
(1997); and the articles by Cool and Edmonds in these proceedings.
< Lx(erg/s)
∼
In this review I discuss the formation and evolution of these X-ray sources
and recycled radio pulsars. Formation through evolution from a primordial bi-
nary is compared with formation via close stellar encounters in Section 2. The
evolution of binaries with a compact star (neutron star or white dwarf) is de-
scribed in Section 3. Various tests of our theoretical ideas are made by com-
parison with observation in Section 4. A brief discussion in Section 5 of X-ray
sources in globular clusters of other galaxies, precedes the Summary.
2. Formation: evolution versus encounters
All binaries that we observe in the disk of the galaxy have evolved from pri-
mordial binaries. Binaries which are very common in the disk, apparently are
commonly formed via ordinary binary evolution. If we find such binaries in a
globular cluster, it is likely that they have evolved from primordial binaries as
well. Examples of such binaries are RS CVn's and contact binaries. On the
other hand, binaries with neutron stars (or black holes) are extremely rare in
Neutron stars in globular clusters
3
the disk; such a binary in a globular cluster is formed almost certainly from a
close encounter of a neutron star with a single star or with a binary. Cataclysmic
variables are in between these extremes: in globular clusters with dense cores,
such as 47 Tuc, most may have formed via stellar encounters; but in more open
clusters, as ω Cen, they may have evolved from primordial binaries (Verbunt &
Meylan 1988).
We consider tidal capture first. The encounter rate is proportional to the
encounter cross section A, to the relative velocity v, and to the numbers nc and n
per unit volume, of compact stars and ordinary stars respectively. To obtain the
encounter rate Γ for the cluster as a whole, one integrates over the cluster volume;
because of the high densities in the core this roughly corresponds to multiplying
the central rate with the core volume. Due to gravitational focussing, the cross
section A of close encounters is proportional to the radius of the star R and
inversely proportional to the square of the velocities v (see also Davies, these
proceedings). Thus (Hut & Verbunt 1983):
Γ ∝ Z ncnAvdV ∝ Z ncnR
v
dV ∝
or3
ρ2
c
v
R
(1)
Where ρo is the central mass density and rc the core radius. An analogous
reasoning gives the exchange encounter rate
Γe ∝ Z ncnbAbvdV ∝ Z ncnba
v
dV ∝
ρ2
or3
c
v
a
(2)
where nb is the number of binaries per unit volume, and a the semi-major axis
of the binary. The ratio of tidal capture to exchange encounters is roughly
Γ
Γe ∼
R
a
n
nb
(3)
The susceptibility of large binaries to close encounters with other cluster
stars implies that many such binaries are dissolved by passing stars. As a result,
the formation of cataclysmic variables from evolution of primordial binaries,
which passes through a stage in which the binary is very wide, is suppressed in
globular clusters (Davies 1997).
The importance of tidal capture is under debate, because it leads to an
initially highly eccentric orbit: the circularization of this orbit is accompanied
by dissipation of an amount of energy ∆E which is comparable to the binding
energy Eb of the ordinary star:
∆E
Eb ≃ (cid:18)−GM m
2ac (cid:19) / 3GM 2
5R ! ≃
5
6
m
M
R
ac
(4)
where M and R are the mass and radius of the tidally distorted star, m the
mass of the other star and ac the radius of the circularized orbit. If the energy is
dissipated more rapidly than it can be radiated or convected away, the ordinary
star is destroyed, and no binary remains (Ray et al. 1987).
Tidal capture, if successful, tends to lead to orbits with a ≃ 3R, i.e. to orbits
with short periods, less than 1 day, say. In contrast, exchange encounters favour
4
Verbunt
wide orbits, and in addition will cause a recoil velocity of the binary which may
take it out of the cluster core -- perhaps even out of the cluster! Tidal capture
can also occur during an encounter of a single star with a binary, when the
three-body intraction brings two stars close to one another.
A much debated point of uncertainty is the number of neutron stars that
remain in a cluster after their formation: young pulsars obtain a kick velocity
at birth, as witnessed by the observed velocities of radio pulsars. It has been
argued that these velocities are higher than the escape velocities of globular
clusters, implying that no neutron stars would be retained. In my opinion, the
velocities of pulsars have been rather overestimated, due to underestimates of
the errors in proper motions and distances (see Hartman 1997). A recent list of
16 accurate velocities obtained with VLBI (Brisken et al. 2002) indicates that
as much as a third of neutron stars is born with velocities less than ∼ 50 km/s.
Thus as many as 30% of the neutron stars born in globular clusters are retained.
Loss of neutron stars from globular clusters is counteracted by mass segregation,
which concentrates the (relatively heavy) neutron stars in the cores, where the
encounter rates are highest (Verbunt & Meylan 1988).
The presence of pulsars in globular clusters is interesting, as they constrain
the amount of non-luminous matter (to which they themselves do not contribute
significantly, as the mass in white dwarfs is always much higher). A pulsar which
has a radial velocity vr with respect to us, will be observed at a period shifted by
∆P = (vr/c)P . If the pulsar is accelerated, a period derivative ∆ P = (ar/c)P
is measured on top of the intrinsic period derivative. Since the latter is always
positive, measurement of a negative period derivative indicates that acceleration
dominates, and thus provides an estimate of the mass density. Phinney (1992)
has shown that the pulsars in M15 put a tight constraint on the mass of any
black hole in the cluster core -- the most likely mass of the cluster core can
in fact be explained by the sum of visible stars and white dwarfs, if the mass
distribution initially followed the Salpeter function, i.e. no black hole is required.
3. Binary evolution
The evolution of a binary with a neutron star (or black hole, or white dwarf) is
a complex topic, which is described only briefly here. For more detail, see e.g.
Verbunt (1993). The mass transfer in a close binary in which a companion to
a neutron star fills its Roche lobe, can be driven by loss of angular momentum
from the orbit.
If a compact star of mass m receives matter from a Roche-
M/M ≃
filling star with mass M , the mass transfer rate is roughly given by −
J/Jb, where Jb is the binary angular momentum. For a main sequence
m/m ∼
star donor, the orbital period Pb is roughly given by Pb(hr) ≃ 8M (M⊙), and
gravitational radiation leads to a mass transfer rate of order 10−10M⊙ yr−1.
When accreted onto a neutron star this leads to an X-ray luminosity ∼ 1036
erg s−1. The observation of higher X-ray luminosities in such binaries have lead
to the suggestion of additional mechanisms of loss of angular momentum, such
as 'magnetic braking', in which a stellar wind carries away angular momentum
efficiently because it is tied to the stellar surface out to large radii by magnetic
field lines. The efficiency of magnetic braking is unknown; recent studies of
stellar rotation in young star clusters suggests that its importance may have
Neutron stars in globular clusters
5
been over-estimated (e.g. Tinker et al. 2002). The mass transfer rate may also
be affected if the donor is irradiated by X-rays from the accreting star.
If
this irradiation causes the star to expand, it may enhance the mass transfer
rate; if it causes an enhanced stellar wind and mass loss from the binary, it may
reduce the mass transfer rate. Whether irradition is important is not known; the
absence of strong heating effects in most well-observed low-mass X-ray binaries
suggests that the accretion disk shield the secondary from X-rays emitted near
the accreting star.
If the mass transfer stops, the neutron star may spin rapidly enough to
switch on as a radio pulsar. The standard description of evolution through loss
of angular momentum does not foresee such an end however, but rather predicts
an everlasting, albeit ever lower, mass transfer. The observation of close binaries
with a pulsar shows that the mass transfer does, in fact, stop. Perhaps if the
mass transfer is variable, it may reach a state sufficiently low that the pulsar
switches on, after which the pulsar wind itself can prevent further mass transfer.
Once more, the details of such a process are not understood.
M /M ≃ m/m ∼
The star that donates matter to a neutron star can also be a white dwarf;
in that case the orbital period is roughly Pb(s) ≃ 50/M (M⊙). In such a binary,
stable mass tranfer can only occur if the white dwarf has a mass less than
∼ 0.66M⊙; a more massive white dwarf will expand so rapidly with mass loss
that it is disrupted within one binary revolution (Verbunt & Rappaport 1988).
If the initial binary is too wide, loss of angular momentum is not important,
and mass transfer will only start when the donor expands as it ascends the
R/R, where R is the stellar
(sub)giant branch. In that case −
R increases with R, and thus the mass
radius. According to stellar evolution,
transfer rate is expected to be higher in wide binaries. Conservation of angular
momentum widens the orbit as mass is transferred from the (sub)giant to the
neutron star; typically, the orbital period has increased by a factor ∼ 7 by the
time that the whole envelope has been transferred. From that point on, the
binary will consist of a neutron star and a white dwarf. The mass transfer
puts an early end to the growth of the giant core, and the emerging white
dwarf is undermassive; typically Mwd ∼ 0.25M⊙. Tidal interaction during the
mass transfer leads to an orbit which is almost but not quite circular. The
observed white dwarf masses and the relation between orbital period and (small)
eccentricity provide strong support for this evolution scenario for recycled pulsars
with white dwarf companions (Phinney 1992).
It is observed that rather more millisecond pulsars are found in globular
clusters than can be formed from the currently observed X-ray binaries. A
solution could be that most recycled pulsars were spun up in binaries with
donors of intermediate mass, 1-3 M⊙ (Davies & Hansen 1998).
4. Formation through close encounters: testing the theory
With a growing number of binaries with neutron stars observed in globular
clusters, we can start comparing their observed properties with the theory of
formation. To do so, we refer to Eqs. 1 and 2, and note that the virial theo-
rem relates the velocity dispersion to the central density and core radius of the
6
Verbunt
Figure 2.
Central density as a function of core radius for globular
clusters in our Galaxy. Data from Harris (1996, revision of June 22
1999). Dotted lines indicate loci of constant formation rate Γ in the
cluster, according to Eq. 6; approximate numerical factors are indicated
on the right for R = R⊙ (applicable to tidal capture) and for a =
100R⊙, nb ≃ n (exchange encounters) where x indicates one encounter
per 10x yr. Dashed lines indicate lines of constant encounter rate γ for
one single system, according to Eq. 7; approximate numerical factors
are indicated on top.
cluster:
Therefore
v ∝ √ρo rc
1.5rc
2R
and
Γ ∝ ρo
1.5rc
2a
(5)
(6)
Γe ∝ ρo
In Figure 2 we plot lines of constant Γ in a graph showing the central density as
function of core radius for the globular clusters of our Milky Way. The numbers
indicated use the King model for the numerical factor in Eq. 5. The formation
rate of binaries with neutron stars can be high because the central density is
high, or because the cluster has a large core.
The situation is different for the probability that a binary, once formed, will
undergo a subsequent encounter which may change or destroy it. The rate at
which a binary undergoes close encounters is given by
γ = nAv ∝
ρo
v
a ∝
0.5
ρo
rc
a
(7)
Neutron stars in globular clusters
7
Figure 3.
Various -- preliminary -- tests on neutron stars in binaries
in globular clusters. a) Number of binaries with neutron stars, b) Slope
of the luminosity function of the X-ray sources, c) Orbital periods of
pulsar binaries, d) Pulsar pulse periods. For details see text.
Lines of constant γ are also shown in Figure 2. We see that in the clusters with
the highest central density, both close binaries (formed from tidal capture) and
wider binaries (formed from an exchange encounter) are affected by subsequent
encounters.
The exact encounter rates in a cluster depend on the mass function in the
core, hence on the mass segregation, on the fraction of stars in binaries, and
on the period distribution of the binaries. For encounters involving neutron
stars, the rates depend on the retention fraction of neutron stars. Also, it is not
clear how one should treat a collapsed cluster. All these factors are different in
different clusters. In the absence of detailed information on most clusters, we
can only perform preliminary tests, in which we ignore these factors.
One such test is that the probability that a cluster contains a bright X-ray
source scales with Γ, i.e. with the collision number Σ ≡ ρ2
c /v. The bright X-ray
sources pass this test (Verbunt & Hut 1987). Johnston et al. (1992) claim to find
that the probability that a cluster contains a recycled pulsar is less dependent
on ρc, viz. ∝ ρ1.5
; however, this lower dependence may be an artefact of their
assumption that v is the same in all clusters (compare eqs.1 and 6).
c r3
c
8
Verbunt
Some further tests are illustrated in Figure 3. Numbers used in these tests
for the X-ray sources / radio pulsars were taken from: ω Cen Rutledge et al. 2002
/ -- ; 47 Tuc Grindlay et al. 2001a / Freire et al. 2001; NGC6440 Pooley et al.
2002b / Lyne et al. 1996; M15 White & Angelini 2001 / Phinney 1992; NGC6752
Pooley et al. 2001a / D'Amico et al. 2002; NGC6397 Grindlay et al. 2002b / -- .
Numbers for other pulsars from: M53 Kulkarni et al. 1991; M5 Anderson et al.
1997; M4 Thorsett et al. 1999; NGC6342 Lyne et al. 1993; NGC6624 Biggs et
al. 1994.
a) ω Cen contains one neutron star X-ray binary and no (known) pulsar,
whereas 47 Tuc, NGC6440 and M 15 contain respectively 2, 4 and 2 neutron star
X-ray binaries and 20, 1 and 8 (known) pulsars. ω Cen indeed has a lower value
of Γ than the other three clusters.
b) The slope p of the X-ray luminosity function dN (Lx) ∝ Lx
−pd ln Lx
(including neutron star binaries, cataclysmic variables, pulsars, and magnetically
active binaries) is steep in 47 Tuc (p = 0.8), intermediate in NGC6752 and
NGC6440 (p = 0.5), and shallow in NGC6397 (p = 0.3). This may be related to
the high value of γ in NGC6397, which prevents a binary from evolving without
being interfered with; thus ordinary binaries in this cluster have largely been
destroyed (Pooley et al. 2002b).
c) The clusters M4 and M53 contain pulsars in binaries with long orbital
periods of 191 and 256 days; these clusters indeed have a low value of γ, i.e.
wide binaries are not affected by encounters. 47 Tuc, with a higher value for γ,
does not contain pulsars in binaries with periods longer than 2.3 d. (The binary
in M4 in fact has a companion third star at about 6000 R⊙; the outer binary
has only an expected life time of 108 yr.)
d) It has been an outstanding puzzle that all pulsars in 47 Tuc and NGC6752
have very short pulse periods (< 7.6 ms and < 9.0 ms), whereas those in M15
have pulse periods ranging up to 111 ms. Long periods also are found in NGC6342,
NGC6440 and NGC6624 (1.0 s, 289 ms and 379 ms). There is no obvious corre-
lation of the pulse period range with Γ or γ.
The encounter hypothesis for the formation of binaries with a neutron star
survives the preliminary tests.
5. Globular clusters in other galaxies
With the Einstein and Rosat satellites, X-ray sources in globular clusters of
M31 have been discovered. On the whole it appears that the properties of these
sources are not significantly different from those in our galaxy (Supper et al.
1997). Thanks to Chandra it has become possible to observe X-ray sources in
other galaxies, with fairly accurate positions. As a result, it has been found
that globular clusters in other galaxies contain X-ray sources with luminosities
above 1039 erg/s (Angelini et al. 2001, Sarazin et al. 2001, Kundu et al. 2002)
At such luminosities, the X-ray sources are probably black holes accreting from
a binary companion; this is confirmed by the soft X-ray spectrum when it can
be measured. Apparently the clusters in those galaxies do contain black holes in
binaries. This is not necessarily a significant difference with the clusters in our
galaxy:
if 20 % of the bright X-ray sources in globular clusters would contain
black holes, the expectation value for such binaries in our globular cluster system
Neutron stars in globular clusters
9
would be 2 or 3. For this expectation value the probability of finding zero is
appreciable.
A first contribution by XMM is the detection of a faint X-ray source in
Mayall II, the largest cluster of M31 (Verbunt, Meylan & Mendez, in prepara-
tion).
6. Summary
The overabundance in globular clusters of binaries with neutron stars is due to
formation of such binaries in close encounters. When the neutron star accretes
matter from its companion, it is an X-ray source. Chandra observations show
that there are 5 to 10 times as many neutron stars accreting at low rates than
at high rates. Whether this is related to their formation mechanism is not clear,
since a similar overabundance is found in the galactic disk where binaries with
neutron stars evolve from primordial binaries (Cornelisse et al. 2002). When the
neutron star stops accreting, it can become a radio pulsar. Most cataclysmic
variables in globular clusters are also formed in close encounters.
Tidal capture leads to a short orbital period, and has no associated recoil
velocity; exchange encounters favour wide binaries and do involve recoil. Most
neutron star binaries are found in or close to the cores of globular clusters. Most
binaries with a pulsar have relatively short orbital periods, and mass functions
that indicate a mass of ≃ 0.25M⊙ for their white dwarf companion.
In my
view, this indicates that these binaries were formed via tidal capture. Some
binaries have rather longer orbital periods, and some are far from the cluster
core. These binaries were probably formed via exchange encounters. From the
X-ray luminosity function it appears that magnetically active binaries -- that
make up the low -- luminosity end of the distribution -- are destroyed in dense
clusters.
Observations of globular clusters in other galaxies show that black holes
are present in them. It would be interesting to investigate whether any of the
low -- luminosity X-ray binaries in the globular clusters of our galaxy contains a
black hole. The small number of X-ray sources in clusters of our Galaxy hampers
comparison of their properties with those in globular clusters systems of other
galaxies.
References
Anderson, S., Wolszczan, A., Kulkarni, S., & Prince, T. 1997, ApJ, 482, 870
Angelini, L., Loewenstein, M., & Mushotzky, R. 2001, ApJ, 557, L35
Biggs, J., Bailes, M., Lyne, A., et al. 1994, MNRAS, 267, 125
Brisken, W., Benson, J., Goss, W., & Thorsett, S. 2002, ApJ, 571, 906
Clark, G. 1975, ApJ, 199, L143
Cornelisse, R., Verbunt, F., in 't Zand, J., et al. 2002, A&A, 392, 931
D'Amico, N., Possenti, A., Fici, L., et al. 2002, ApJ, 570, L89
Davies, M. 1997, MNRAS, 288, 117
Davies, M. & Hansen, B. 1998, MNRAS, 301, 15
10
Verbunt
Freire, P. 2002 http://www.naic.edu/∼pfreire/GCpsr.html
Freire, P., Camilo, F., Lorimer, D., et al. 2001, MNRAS, 326, 901
Grindlay, J., Heinke, C., Edmonds, P., , & Murray, S. 2001a, Science, 292, 2290
Grindlay, J., Heinke, C., Edmonds, P., et al. 2001b, ApJ, 563, L53
Harris, W. 1996, AJ, 112, 1487
Hartman, J. 1997, A&A, 322, 127
Hertz, P. & Grindlay, J. 1983, ApJ, 275, 105
Hut, P. & Verbunt, F. 1983, Nat, 301, 587
in 't Zand, J., Verbunt, F., Strohmayer, T., & et al. 1999, A&A, 345, 100
Johnston, H., Kulkarni, S., & Phinney, E. 1992, in X-ray binaries and the forma-
tion of binary and millisecond pulsars, ed. E. van den Heuvel & S. Rap-
paport (Dordrecht: Kluwer), 349 -- 364
Kulkarni, S., Anderson, S., Prince, T., & Wolszczan, A. 1991, Nat, 349, 47
Kundu, A., Maccarone, Th., & Zepf, S. 2002, ApJ, 574, L5
Lyne, A. 1994, in The evolution of X-ray binaries, ed. S. Holt & C. Day (New
York: AIP), 331 -- 338
Lyne, A., Biggs, J., Harrison, P., & Bailes, M. 1993, Nat, 361, 47
Lyne, A., Manchester, R., & D'Amico, N. 1996, ApJ (Letters), 460, L41
Phinney, E. 1992, Phil. Trans. R. Soc. London A, 341, 39
Pooley, D., Lewin, W., Homer, L., et al. 2002a, ApJ, 569, 405
Pooley, D., Lewin, W., Verbunt, F., et al. 2002b, ApJ, 573, 184
Ray, A., Kembhavi, A., & Antia, H. 1987, A&A, 184, 164
Rutledge, R., Bildsten, L., Brown, E., et al. 2002, ApJ in press
Sarazin, C., Irwin, J., & Bregman, J. 2001, ApJ, 556, 533
Supper, R., Hasinger, G., Pietsch, W., & et al. 1997, ApJ, 317, 328
Thorsett, S., Arzoumanian, Z., Camilo, F., & Lyne, A. 1999, ApJ, 523, 763
Tinker, J., Pinsonneault, M., & Terndrup, D. 2002, ApJ, 564, 877
Verbunt, F. 1993, ARA&A, 31, 93
-- . 2001, A&A, 368, 137
Verbunt, F., Bunk, W., Ritter, H., & Pfeffermann, E. 1997, A&A, 327, 602
Verbunt, F. & Hut, P. 1987, in The Origin and Evolution of Neutron Stars, IAU
Symposium No. 125, ed. D. Helfand & J.-H. Huang (Dordrecht: Reidel),
187 -- 197
Verbunt, F. & Meylan, G. 1988, A&A, 203, 297
Verbunt, F. & Rappaport, S. 1988, ApJ, 332, 193
Webb, N., Gendre, B. & Barret, D. 2002, A&A, 381, 481
White, N. & Angelini, L. 2001, ApJ, 561, L101
|
astro-ph/0611475 | 1 | 0611 | 2006-11-15T11:21:52 | Evidence for bipolar jets from the optical spectra of the prototypical symbiotic star Z Andromedae | [
"astro-ph"
] | We have studied optical spectra of the symbiotic star Z And, obtained during its latest outburst started in April 2006, with the aim of finding changes in the spectrum yielding clues to the nature of the hot component and its outbursts. The spectroscopic observations of Z And have been made using the 1.5-meter telescope at the Tartu Observatory, Estonia, and processed in a standard way. We have found high velocity satellites to the hydrogen Balmer emission lines. Starting from July 30, 2006, weak additional emission components at velocities of about +/-1150 km/s were detected. Their appearance near the outburst maximum and similarity to the emission features in another symbiotic star Hen 3-1341 imply fast collimated outflows from the hot component of Z And. This finding is consistent with the earlier results by several authors that symbiotic stars can emit bipolar jets at certain stages of their outbursts. A significant decrease in the temperature of the hot component in initial stages of the outburst was detected by the disappearance of the high excitation emission line from the spectrum. | astro-ph | astro-ph |
Astronomy&Astrophysicsmanuscript no. 6630
October 22, 2018
c(cid:13) ESO 2018
Letter to the Editor
Evidence for bipolar jets from the optical spectra of the
prototypical symbiotic star Z Andromedae ⋆
M. Burmeister1,2 and L. Leedjarv1
1 Tartu Observatory, 61602 Toravere, Estonia
2 Institute of Theoretical Physics, University of Tartu, Tahe 4, 51010 Tartu, Estonia
Received 24 October 2006 / Accepted 3 November 2006
ABSTRACT
Aims. We have studied optical spectra of the symbiotic star Z And, obtained during its latest outburst started in April 2006, with the
aim of finding changes in the spectrum yielding clues to the nature of the hot component and its outbursts.
Methods. The spectroscopic observations of Z And have been made using the 1.5-meter telescope at the Tartu Observatory, Estonia,
and processed in a standard way.
Results. We have found high velocity satellites to the hydrogen Balmer emission lines. Starting from July 30, 2006, weak additional
emission components at velocities of about ± 1150 km s−1 were detected. Their appearance near the outburst maximum and similarity
to the emission features in another symbiotic star Hen 3-1341 imply fast collimated outflows from the hot component of Z And.
This finding is consistent with the earlier results by several authors that symbiotic stars can emit bipolar jets at certain stages of
their outbursts. A significant decrease in the temperature of the hot component in initial stages of the outburst was detected by the
disappearance of the high excitation emission lines from the spectrum.
Key words. stars: outflows -- binaries: symbiotic -- stars: individual: Z And
1. Introduction
Z Andromedae is considered to be the prototype for the class of
symbiotic stars (Kenyon 1986; Corradi et al. 2003) -- interacting
binary stars consisting of a red giant and a hot compact com-
panion, mostly a white dwarf. The hot component ionises part
of the wind of the red giant, thus giving rise to the characteris-
tic "combination" spectrum where high excitation emission lines
are superimposed on the cool giant's spectrum. Z And has an or-
bital period of 757.5 days (Mikołajewska & Kenyon 1996) and
an orbital inclination of 47◦ ± 12◦ (Schmid & Schild 1997). Its
light curve represents classical symbiotic star outbursts, demon-
strating brightening of the star by 2 -- 3 magnitudes in the visual
region in every few years. Z And is also the only symbiotic star
for which coherent optical oscillations with the period of 28 min-
utes have been detected, most likely indicating the presence of
a strongly magnetic WD (Sokoloski & Bildsten 1999). Finally,
Z And is among about 10 known symbiotic stars, producing col-
limated outflows or jets (Brocksopp et al. 2004; Leedjarv 2004).
Extended collimated outflows perpendicular to the orbital plane
were detected from radio images during the 2000 -- 2002 outburst
(Brocksopp et al. 2004).
Behaviour of Z And during the outburst and the following
quiescence in 2000 -- 2003 was extensively studied and analysed
by Sokoloski et al. (2006). They proposed to call Z And and
possibly other classical symbiotic stars 'combination novae', in-
dicating that both dwarf nova-like accretion disk instability and
nova-like nuclear shell burning are involved in the outbursts
of classical symbiotic stars. According to the visual photomet-
ric observations from the AAVSO database, a new outburst of
Send offprint requests to: M. Burmeister, e-mail: [email protected]
⋆ Based on observations collected at Tartu Observatory, Estonia
Z And began in early June 2004. Having hardly recovered from
this outburst, the star started a new activity cycle in April 2006.
In the present paper we describe and analyse the optical spectra
of Z And, obtained mostly during decline from the optical maxi-
mum in 2006. Section 2 describes the observations and in Sect. 3
we present and analyse evidence for fast bipolar outflows during
the outburst of Z And. Section 4 contains conclusions.
2. Observations
Spectroscopic observations of Z And have been carried out at
the Tartu Observatory, Estonia, using the 1.5-meter telescope
equipped with the Cassegrain grating spectrograph. Two dif-
ferent CCD cameras from the SpectraSource Instruments were
used until March 2006. Later on the spectra are registered with
the Andor Technologies CCD camera Newton USB-207 with
400 × 1600 chip, pixel size 16 × 16 µm, Peltier cooled. A few oc-
casional spectra of Z And have been obtained since September
1997, mostly in the region of Hα. A more systematic monitoring
started on July 30, 2006, after detecting the signs for bipolar jets.
Most of the red spectra cover the wavelength region
6480−6730 Å, with linear dispersion about 0.2 Å pixel−1. Blue
spectra are mostly taken in the 4600−5060 Å region with
∼ 0.3 Å pixel−1. Some of the lower dispersion spectra (about
0.5 Å pixel−1 and 0.6 Å pixel−1, respectively) are also used. The
spectra were reduced using the software package MIDAS pro-
vided by ESO. After standard procedures of subtracting the bias
and sky background, the spectra were reduced to the wavelength
scale using the Th-Ar hollow cathode lamp spectrum for cali-
bration. Thereafter, the spectra were normalised to the contin-
uum, and positions, peak intensities, and equivalent widths of
the emission lines were measured.
2
M. Burmeister and L. Leedjarv: Bipolar jets from Z And
y
t
i
s
n
e
t
n
i
e
v
i
t
a
l
e
R
25
20
15
10
5
0
-1000
0
Radial velocity (km s-1)
1000
Fig. 1. Hα line profiles from August 7 (solid line) and September
10 (dotted line). The former shows an absorption component.
Table 1. Radial velocities of the jet components of Hα and Hβ
line in km s−1
Date
2006
Jul 30
Aug 7
Aug 9
Aug 10
Sep 1
Sep 10
Sep 12
Sep 13
Sep 15
Sep 21
Sep 24
Sep 27
Oct 6
Oct 12
Mean
Hα
JD -
2 400 000 Blue
-1114
53947.5
-1115
53955.5
53957.4
-1106
-1134
53958.5
-1243
53980.4
-1150
53989.4
-1193
53991.4
53992.5
-1180
-1156
53994.4
-1081
54000.4
-1114
54003.5
-1115
54006.4
54015.3
-1106
-1124
54021.4
-1138±41
Red
1391
1271
1275
1280
1165
1157
1145
1067
1059
1100
1087
1107
1110
1108
1166±96
Blue
-969
-1288
-1180
-1217
-1191
-1176
-1070
-1162
-1150
Hβ
Red
1376
1160
1084
1127
1061
1039
1098
1137
1134
-1196
-1160±82
1134
1135±88
3. Results and discussion
Both Hα and Hβ lines have a central emission component at
radial velocities of about 30 km s−1. The shape of the profiles
is asymmetric with the blue wing distorted by absorption. On
August 7 and 9, the blue step of the Hα line deepens into an ab-
sorption component (Fig. 1). A similar absorption is seen in Hβ
on September 1.
The most striking features of these Balmer lines, however,
are the satellite emission components (see Fig. 2), detected for
the first time on July 30, 2006 in the Hα line. We are not aware
of observations of similar features in the spectrum of Z And be-
fore. On September 14, 2006, a telegram on the detection of jets
in Z And was published by Skopal & Pribulla (2006). Very sim-
ilar additional emission features were discovered in the spectra
of the symbiotic stars Hen 3-1341 by Tomov et al. (2000), and
StHα 190 by Munari et al. (2001), who interpreted them to be
emitted by the collimated bipolar jets.
We fitted these emission components with a Gaussian to find
their radial velocities. The results are given in Table 1. On av-
erage the RVs are about ±1150 km s−1. Given the asymmetric
Hα
Feb 26, 2006
Jul 30, 2006
Aug 7, 2006
Aug 9, 2006
Sep 1, 2006
Sep 15, 2006
Sep 27, 2006
Oct 12, 2006
-1000
0
y
t
i
s
n
e
t
n
i
e
v
i
t
a
l
e
R
1.4
1.2
1
1.4
1.2
1
1.4
1.2
1
1.4
1.2
1
1.4
1.2
1
1.4
1.2
1
1.4
1.2
1
1.4
1.2
1
Hβ
1.4
1.2
1
1.4
1.2
1
1.4
1.2
1
1.4
1.2
1
1.4
1.2
1
1000
-1000
1000
Radial velocity (km s-1)
0
Fig. 2. Time evolution of the jet components around the Hα and
Hβ lines. Our last spectrum before the outburst from February
26, although more noisy, does not show additional emission
components.
shape of the additional components, the accuracy of those ve-
locities is no better than about 40−50 km s−1. If we assume that
the jets are perpendicular to the orbital plane and the inclina-
tion of the orbit is 47◦, the true velocities of the jets are close to
±1700 km s−1.
The equivalent widths of the supposed jet components form
about 2% of that of the central emission, but these estimates are
uncertain, as the small emissions are partly blended in the central
component. The measured values of EWs are plotted in Fig. 3.
At least in some of the red spectra, weak [N II] 6548 and
6584 Å emission lines can be distinguished at the low-velocity
sides of Hα jet components. Where measurable, the velocities of
the [N II] lines occur at about −45 km s−1.
According to the AAVSO light curve, the optical maximum
of Z And took place in early July 2006 (around JD 2 453 920 −
930). Our last optical spectrum before the outburst was made on
February 26, 2006. No signs of jets are seen in this spectrum, so
the jets must have been formed at some time between February
and July. For most of the time of our observations, the jet features
in the spectrum have been rather stable. Smaller EWs on July 30
possibly indicate the early phase of jet development.
Such an appearance of the jets at the time of the optical max-
imum fits the behaviour of another symbiotic star Hen 3-1341
as described by Munari et al. (2005) well. In general, collimated
bipolar jets are observed from several types of stars, such as, e.g.,
X-ray binaries, supersoft X-ray sources, young stellar objects,
etc., and of course, from active galactic nuclei. The standard sce-
M. Burmeister and L. Leedjarv: Bipolar jets from Z And
3
He I 4713
He II 4686
3
2.5
2
W
E
1.5
1
0.5
0
53960
53980
54000
JD - 2 400 000
54020
Fig. 3. EWs of the blue (filled circles) and red (empty circles) jet
component of the Hα line and of the blue (filled triangles) and
red (empty triangles) jet component of the Hβ line.
nario for the production of jets (e.g., Livio 1997; Lynden-Bell
2003, and many later references) includes an accretion disc that
is threaded by a vertical magnetic field. In addition, an energy
or wind source associated with the central accreting object is
needed. This requirement explains, for instance, the absence of
jets from magnetic cataclysmic variables, in spite of the presence
of an accreting white dwarf and magnetic field. The behaviour
of Hen 3-1341 well demonstrates that the jets are emitted only
in the early and the brightest phase of the outburst, and that they
are fed by the wind from the outbursting component (Munari et
al. 2005). The outburst of Hen 3-1341, however, lasted longer
(from 1998 to 2004) than those of Z And usually do.
The wings of the Hα main emission component extend to
about ±1000 km s−1, which may indicate fast wind from the out-
bursting hot component. On the other hand, indications of low-
velocity outflow can be seen in the lines of He I 4471, 4713, and
5016, which present P Cygni profiles with deep absorption com-
ponents. The development of the P Cygni profile of He I 4713
is seen in Fig. 4. Velocity of the outflow measured from these
lines is between −50 and −100 km s−1. This is comparable to
the velocity −90 kms−1 found by Skopal et al. (2006) during the
2000 -- 2003 outburst. At the same time, the He I 4922 and 6678
lines are fully in emission or have only a very weak absorption
component. Figure 5 presents an example.
The present outburst of Z And is similar to that of Hen 3-
1341 also by the behaviour of other emission lines. It is known
that high excitation emission lines tend to weaken during the
outbursts of symbiotic stars. In Hen 3-1341, the He II 4686 and
the Raman scattered O VI line at 6825 disappeared completely.
Similar changes in the spectrum of Z And during the 2000 -- 2003
outburst were reported by Skopal et al. (2006) and Sokoloski
et al. (2006). We also an detected almost complete disappear-
ance of the Raman scattered line at 6825 and of the He II 4686
line in the spectrum of Z And during the present outburst in
2006. Figure 6 presents the comparison of those spectral lines
from the quiescence in September 1997 and from the outburst in
September 2006. In Fig. 4, the gradual reappearance of the He II
4686 line is seen while the brightness of the star decreases.
We made an attempt to estimate the effective temperature of
the hot component during our observations by using the method
Aug 9, 2006
Sep 1, 2006
Sep 15, 2006
Sep 27, 2006
Oct 12, 2006
y
t
i
s
n
e
t
n
i
e
v
i
t
a
l
e
R
2
1.6
1.2
0.8
2
1.6
1.2
0.8
2
1.6
1.2
0.8
2
1.6
1.2
0.8
2
1.6
1.2
0.8
2
1.6
1.2
0.8
2
1.6
1.2
0.8
2
1.6
1.2
0.8
2
1.6
1.2
0.8
2
1.6
1.2
0.8
-200
0
200
-200
0
200
Radial velocity (km s-1)
Fig. 4. Time evolution of the He I 4713 and He II 4686 lines,
representing matter outflow and outburst state, respectively. The
profiles are given on the same scale for comparison.
He I 4713
He I 6678
3
2.5
2
1.5
1
y
t
i
s
n
e
t
n
i
e
v
i
t
a
l
e
R
-300
-200
-100
100
Radial velocity (km s-1)
0
200
300
Fig. 5. Example of an He I 4713 line and He I 6678 line from
September 27, 2006.
proposed by Iijima (1981). Like Sokoloski et al. (2006) we used
equivalent widths of Hβ and He II 4686 instead of fluxes and
neglected the He I 4471 line. The temperatures in August and
September tend to be low, falling in the range 75 000 -- 90 000 K,
and begin to rise at the end of September. The results are given in
Table 2. A similar decline of Thot at the brightness maximum and
its gradual increase after the maximum was found by Sokoloski
4
M. Burmeister and L. Leedjarv: Bipolar jets from Z And
Table 2. Effective temperatures of the hot star estimated by the
Iijima method
Date
2006
Aug 9
Sep 10
Sep 12
Sep 13
Sep 15
Sep 21
Sep 24
Sep 27
Oct 12
JD -
2 400 000
53957.438
53989.398
53991.346
53992.549
53994.394
54000.363
54003.519
54006.388
54021.368
Temperature
(K)
76 000
91 000
80 000
75 000
81 000
87 000
91 000
100 000
114 000
He II 4686
O VI 6825
20
15
10
5
y
t
i
s
n
e
t
n
i
e
v
i
t
a
l
e
R
2.5
2
1.5
1
0
4670
4680
4690
0.5
4700
6800
6820
6840
6860
Wavelength (A)
Fig. 6. Left: He II 4686 line of Z And from September 27, 1997
(solid line) and September 27, 2006 (dotted line). Right: Raman
scattered line O VI 6825 from September 2, 1997 (solid line)
and September 27, 2006 (dotted line). Disappearance of the high
ionization emission lines refers to an outburst state.
et al. (2006) during the 2000 -- 2003 outburst. On the contrary,
Thot increased from the quiescence value of <∼ 150 000 K to
about 180 000 K during the smaller scale short outburst in 1997.
Those facts imply similarity with another classical symbiotic
star AG Dra in which Gonz´alez-Riestra et al. (1999) have dis-
tinguished between cool and hot outbursts, based on the changes
in the He II Zanstra temperature. The outbursts of Z And in 2000
and 2006 can be considered as cool ones and that in 1997 as hot
outburst. It might be of interest to note that AG Dra has also been
in outburst since July 2006. We will describe spectroscopic be-
haviour of AG Dra in another forthcoming paper. Here we only
note that during the present outburst the He II 4686 line and the
Raman scattered O VI line 6825 also became very weak in the
spectrum of AG Dra. This is different from smaller scale out-
bursts of AG Dra in the late 1990s -- early 2000s when all the
emission lines became stronger (e.g., Tomov & Tomova 2002;
Leedjarv et al. 2004).
4. Conclusions
From our observations of the prototypical symbiotic star Z And
we can conclude the following:
(1) The additional emission components of the hydrogen Hα
and Hβ lines at ±1150 km s−1 indicate ejection of bipolar jets
starting from late July 2006, and persisting for at about three
months, at least. Taking into account the orbital inclination 47◦,
ejection velocity of the jets would be about 1700 km s−1.
(2) Although Z And has been extensively studied over the
past decades, no such jets were detected from the optical spectra
before. We have some spectra in the Hα region near the bright-
ness maximum of the 2004 outburst, but no additional emission
components can be found from these. According to Skopal et al.
(2006), no signs of jets could be seen during the maximum of the
2000 -- 2003 outburst. Spectra by Sokoloski et al. (2006) from the
same time have too low resolution to confirm the existence or ab-
sence of additional emissions. We also have one spectrum from
August 25, 2001, which was taken almost at the same time when
Brocksopp et al. (2004) discovered the radio jets from Z And.
In our spectrum, the jets are not seen. This most likely means
that the jets are transient phenomenon, emitted only for a short
time in the brightest phase of the outbursts, thus confirming the
standard scenario for jet ejection in which an additional source
of energy or wind is needed besides the accreting central body
and magnetic field.
(3) The high excitation lines He II 4686 and Raman scattered
O VI 6825 almost disappeared from the spectrum of Z And in the
early phase of the outburst, confirming that the temperature of
the hot component decreased to about 75 000 K, as was found by
the Iijima method. In September 2006, about 3 -- 4 months after
the start of the outburst, the He II 4686 line started to become
stronger again.
The ongoing outburst of Z And has shown that new aspects
can be found in the behaviour of such well studied symbiotic
stars. Together with the combination nova model proposed by
Sokoloski et al. (2006), new clues for understanding the out-
bursts of AG Dra, Hen 3-1341, and other classical symbiotic
stars can be obtained. Continuing monitoring of the ongoing out-
burst of Z And is strongly encouraged.
Acknowledgements. The authors thank Dr. Kalju Annuk for taking some of the
spectra of Z And and for useful comments on the manuscript of the paper. We
also thank our referee, Dr. Michael Bode, for his suggestions. The present study
was supported by the Estonian Ministry of Education and Research under the
target financed project 0062464S03 "Structure, chemical composition, and evo-
lution of stars" and by the Estonian Science Foundation grant No 6810.
References
Brocksopp, C., Sokoloski, J.L., Kaiser, C., et al. 2004, MNRAS, 347, 430
Corradi, R.L.M., Mikołajewska, J., & Mahoney, T.J. (eds.) 2003, Symbiotic Stars
Probing Stellar Evolution, ASP Conf. Series, Vol. 303
Gonz´alez-Riestra, R., Viotti, R., Iijima, T., & Greiner, J. 1999, A&A, 347, 478
Iijima, T. 1981, In: Photometric and Spectroscopic Binary Systems, eds. E.B.
Carling & Z. Kopal, Dordrecht, Kluwer, 517
Kenyon, S.J. 1986, The Symbiotic Stars, Cambridge University Press
Leedjarv, L. 2004, Baltic Astronomy, 13, 109
Leedjarv, L., Burmeister, M., Mikołajewski, M., et al. 2004, A&A, 415, 273
Livio, M. 1997, In: Wickramasinghe, D.T., Bicknell, G.V., Ferrario, L. (eds.),
Accretion Phenomena and Related Outflows, ASP Conf. Series, Vol. 121,
845
Lynden-Bell, D. 2003, MNRAS, 341, 1360
Mikołajewska, J., & Kenyon, S.J. 1996, AJ, 112, 1659
Munari, U., Tomov, T., Yudin, B. F., et al. 2001, A&A, 369, L1
Munari, U., Siviero, A., & Henden, A. 2005, MNRAS, 360, 1257
Schmid, H.M., & Schild, H. 1997, A&A, 327, 219
Skopal, A., & Pribulla, T. 2006, Astronomers Telegram, 882
Skopal, A., Vittone, A.A., Errico, L. et al. 2006, A&A, 453, 279
Sokoloski, J.L., & Bildsten, L. 1999, ApJ, 517, 919
Sokoloski, J.L., Kenyon, S.J., Espey, B.R., et al. 2006, ApJ, 636, 1002
Tomov, N. A., & Tomova, M. T. 2002, A&A, 388, 202
Tomov, T., Munari, U., & Marrese, P. M. 2000, A&A, 354, L25
List of Objects
'Z And' on page 1
M. Burmeister and L. Leedjarv: Bipolar jets from Z And
5
'Hen 3-1341' on page 2
'StHα 190' on page 2
'Hen 3-1341' on page 2
'AG Dra' on page 4
|
astro-ph/0010145 | 1 | 0010 | 2000-10-07T04:02:46 | Dynamics of Stellar Collisions | [
"astro-ph"
] | I compare gas-dynamical and stellar-dynamical models of collisions. These two models have distinctly different physics; for example, shocks introduce irreversibility in gas systems, while stellar systems evolve in a completely reversible fashion. Nonetheless, both models yield broadly similar results, suggesting that analogies between gas and stellar dynamics have some heuristic validity even applied to collisions. | astro-ph | astro-ph |
Stellar Collisions, Mergers, and Their Consequences
ASP Conference Series
M. Shara, ed.
Dynamics of Stellar Collisions
J. E. Barnes
Institute for Astronomy, University of Hawaii, 2680 Woodlawn Drive,
Honolulu HI, 96822
I compare gas-dynamical and stellar-dynamical models of
Abstract.
collisions. These two models have distinctly different physics; for exam-
ple, shocks introduce irreversibility in gas systems, while stellar systems
evolve in a completely reversible fashion. Nonetheless, both models yield
broadly similar results, suggesting that analogies between gas and stellar
dynamics have some heuristic validity even applied to collisions.
1.
Introduction
Stars and galaxies are prototypical examples of self-gravitating systems. While
both are held together by gravity, they differ in size, makeup, and structure.
From a dynamical point of view, the basic difference is that stars are made up of
particles which often undergo collisions, while galaxies are made up of particles
which almost never collide (this conference notwithstanding).
Analogies between stellar and gaseous systems run deep. For example, there
is a close relationship between spherical stellar systems with isotropic distribu-
tion functions f = f (E) and gaseous systems with barotropic equations of state
P = P (ρ). The stability properties of spherical stellar and gaseous systems
provide further analogies; an isotropic stellar system with df /dE < 0 is stable if
the barotropic gas-sphere with the same density profile is stable (Antonov 1962;
Lynden-Bell 1962). This is a sufficient but not necessary condition; on the whole,
spherical stellar systems seem to be more stable than their gaseous counterparts.
For flattened systems the situation is more complex. Infinite, uniformly rotating
sheets of gas and stars have very similar stability criteria (Toomre 1964). On
the other hand, certain finite, uniformly-rotating stellar disks (Kalnajs 1972) are
much less stable than gaseous disks with the same density profile.
Questions of existence and stability exploit analogies between stellar and
gaseous systems at or near equilibrium. In this paper I use numerical simulations
to examine how such analogies work in situations which are far from equilibrium:
collisions. Colliding stars and colliding spherical galaxies evolve in superficially
similar ways, as seen in Fig. 1. Here the top row of frames shows a collision
of two gas-spheres, while the bottom row shows the analogous collision of two
spherical systems of collisionless particles, hereafter labeled "grit"1. A more
1In my original presentation I used "dust" instead of "grit", borrowing this term from cosmology.
However, S. Shapro pointed out that "dust" is only applicable to a cold, pressureless medium.
Casting about for a suitable term, I chose "grit" since this conveys some resilience to pressure.
1
2
Barnes
Figure 1.
Gas (top) and grit (bottom) versions of the same collision. Ini-
tially the two bodies approach on a parabolic relative orbit, passing each
other in a counter-clockwise direction. The mass ratio is M1/M2 = 2 and the
targeted pericentric separation rp = 0.5 is half the radius of the larger body.
detailed comparison of the two collisions in Fig. 1 reveals significant differences.
For example, it's evident that orbit decay doesn't work in quite the same way;
contrasting the third pair of frames, the two bodies are much closer in the gas
model than they are in the grit model. Moreover, grit evolves reversibly, while
gas evolves irreversibly; by integrating the merged grit model backward in time,
one can unscramble the wreckage and recover the initial conditions (van Albada
& van Gorkom 1977), but the presence of shocks in the gas model precludes such
reversals.
1.1. Gas model
Technically, the interior of a normal star is a partly-ionized plasma in local
equilibrium with a black-body radiation field; support against gravity is provided
by some combination of thermal and radiation pressure. But for low-mass stars
the latter is insignificant, and to a good approximation the pressure is given by
the equation of state for an ideal gas.
Gas dynamics unfolds in a space of three dimensions; the dynamical vari-
ables are functions of position r and time t. Two such variables are the density
ρ = ρ(r, t) and the velocity field v = v(r, t). One more variable is needed to
represent the thermodynamic state of the gas; in these calculations I use the
entropy function a = a(S), which enters directly in the equation of state:
P = a(S)ργ ,
(1)
where P is the pressure, and γ is the ratio of specific heats, here assigned the
value γ = 5/3 appropriate for a monatomic gas.
Dynamics of Stellar Collisions
I adopt the following dynamical equations:
∂ρ
∂t
+
∂
∂r
· (ρv) = 0 ,
∂v
∂t
∂a
∂t
+ (cid:18)v ·
+ (cid:18)v ·
∂
∂
1
ρ
∂P
∂r
−
∂Φ
∂r
∂r(cid:19) v = −
∂r(cid:19) a = (γ − 1)ρ1−γ uv .
,
3
(2)
(3)
(4)
Here eq. 2 represents conservation of mass and eq. 3 represents conservation of
momentum; the gravitational potential Φ is calculated from Poisson's equation.
Eq. 4 describes the evolution of the entropy function; the flow is adiabatic,
conserving the entropy of each fluid element, except where the viscous heating
function uv > 0. A standard form of artificial viscosity (Monaghan & Gingold
1983) is used to implement the increase in gas entropy due to shocks.
I solve these dynamical equations using Smoothed Particle Hydrodynamics
or SPH (eg. Monaghan 1992). The code is similar to "TREESPH" (Hernquist &
Katz 1989); it includes a hierarchical algorithm to compute gravitational forces,
individual smoothing radii hi set so that that each particle i interacts with a fixed
number of neighbors, and individual time-steps adjusted to satisfy a Courant
condition. The code doesn't include terms proportional to ∇hi which arise
when the SPH equations are derived from a Hamiltonian (Nelson & Papaloizou
1993, 1994). Since the code integrates Eq. 4 instead of the equivalent energy
equation, the neglect of these ∇hi terms leads to imperfect energy conservation
(Hernquist 1993). Most of the calculations presented below conserve energy to
∼ 1% or better, so the neglect of these terms is probably not critical in this
application.
1.2. Grit model
The dominant mass components in typical galaxies are stars and dark matter. In
the absence of conflicting evidence the latter is often assumed to be composed
of particles with masses much less than 106M⊙ (cf. Lacey & Ostriker 1985).
The gravitational potential of a galaxy is quite smooth and individual stars or
dark matter particles follow smooth trajectories, undisturbed by collisions, close
encounters, or any effect due to the discrete nature of the mass.
Stellar dynamics unfolds in a space of six dimensions; the phase-space dis-
tribution f = f (r, v, t) is a function of position r, velocity v, and time t. In the
limit appropriate for galaxies, f evolves according to the Collisionless Boltzmann
Equation or CBE:
∂f
∂t
+ v ·
∂f
∂r
−
∂Φ
∂r
·
∂f
∂v
= 0
(5)
Like eq. 2, eq. 5 is a continuity equation, conserving the mass of the system.
But eq. 5 describes an incompressible flow in six dimensions; dynamics moves
elements of phase-fluid around but conserves the value of f associated with each.
I solve the CBE using standard N-body techniques (eg. Barnes 1998). The
code uses a hierarchical algorithm to compute gravitational forces and a simple
leap-frog integrator to advance particle coordinates.
4
Barnes
1.3. Polytropes
I chose to collide simple polytropes, instead of using stellar models. While ac-
curate simulations of merging stars require realistic stellar models (Sills & Lom-
bardi 1997), polytropes are still appropriate for comparing the basic physics of
stellar and galactic collisions. Polytropes are easily constructed, and their prop-
erties are well-understood (eg. Chandrasekhar 1939). Moreover, grit models of
polytropes have simple distribution functions (Eddington 1916) and the stabil-
ity of these systems has been studied using N-body simulations (H´enon 1973;
Barnes, Goodman, & Hut 1986).
Both gas and grit polytropes obey the following relationship between mass
density ρ and gravitational potential Φ:
ρ = ρ(Φ) = ρ1(1 − Φ/Φ1)n .
(6)
Here Φ1 is the value of the potential on the surface of the polytrope and ρ1 is a
constant with units of density. By construction, gas polytropes are marginally
stable to convection, and this implies that the index n is related to the ratio of
specific heats by n = 1/(γ − 1). For grit polytropes, n is a free parameter, and
the distribution function takes the form
f = f (E) = (cid:26) f1(1 − E/Φ1)n−3/2
0
if E < Φ1,
otherwise,
(7)
where f1 is a constant with units of phase-space density.
For a monatomic gas, γ = 5/3, and the appropriate index n = 3/2. The cor-
responding grit polytrope has an unusually simple structure; f has the constant
value f1 within the six-dimensional volume defined by E = 1
2 v2 + Φ(r) < Φ1,
and vanishes everywhere else.
To construct realizations of polytropes I first tabulated the density profile
ρ(r), potential Φ(r), and pressure P (r). Grit models were generated by chosing
initial positions ri and velocities vi for each particle according to the distribu-
tion function f (r, v) = f ( 1
2 v2 + Φ(r)) (H´enon 1973). Likewise, gas models were
generated by chosing initial positions ri according to the density profile ρ(r),
setting initial velocities vi = 0, and assigning initial entropy function values
ai = P (ri)/ργ (ri). In both cases all particles had equal masses. Since particle
coordinates were assigned independently, these initial configuration have Pois-
sonian fluctuations. Such fluctuations are inevitable in N-body simulations, but
they are much larger than the fluctuations normally present in SPH calculations.
Therefore, the gas realizations were relaxed by evolving them with the SPH code
using a velocity-damping term. This has the effect of "ironing out" the initial
density fluctuations without disturbing the overall density profile.
2. Collision Sample
The dynamics of a collision depend on the relative orbit and mass ratio of the
participants as well as their internal structures. The collisions reported here were
restricted to parabolic (zero-energy) orbits, which are appropriate for collisions
in globular clusters and similar environments. I used n = 3/2 polytropes and
assumed a linear relationship between mass and radius; these choices roughly
Dynamics of Stellar Collisions
5
Figure 2.
Energies vs. time for all collisions, each labeled by (M1/M2, rp).
The three solid curves in each panel are from the gas models; from top to
bottom, they show the thermal energy T (t), kinetic energy K(t), and potential
energy U (t). Likewise, the two dashed curves are from the grit models; they
show the kinetic and potential energies.
caricature the properties of low-mass main-sequence stars. It may ultimately be
worth relaxing these restrictions, but even with them in place, two parameters
must still be chosen for each collision: the mass ratio and the pericentric sepa-
ration of the initial relative orbit. I adopted mass ratios M1/M2 = 1 and 2, and
studied collisions with pericentric separations rp ranging from zero (head on)
to R1 + R2 (grazing). To simplify comparisons between collisions with different
mass ratios, I stipulated that all cases have the same total mass.
Since G is the only dimensional constant in these calculations, it's con-
venient to adopt units with G = 1. The polytropes used for the equal-mass
collisions (M1/M2 = 1) have masses M1 = M2 = 0.75 mass units and radii
R1 = R2 = 0.75 length units; for the gas models, the initial entropy func-
tion value is a(S1) = a(S2) = 0.2889. The polytropes used for the unequal-
mass collisions (M1/M2 = 2) have masses M1 = 1.0 and M2 = 0.5 and radii
R1 = 1.0 and R2 = 0.5; the initial entropy function values are a(S1) = 0.4240
and a(S2) = 0.1682.
For both mass ratios I considered four collisions with pericentric separations
rp = 0, 0.5, 1, and 1.5 (= R1 + R2); each collision was run with both gas and grit
models, using 24576 particles for each calculation. I ran all collisions with rp ≤ 1
until the participants merged and relaxed to near-equilibrium configurations.
The grazing collisions were not run to completion; orbit decay is very gradual
in such wide passages and the time required to reach merger seemed excessive.
Fig. 2 summarizes the evolution of the collisions. In each case the partici-
pants have a first pericentric passage at t ≃ 2.5, marked by sharp minima in the
6
Barnes
Figure 3.
First passages for equal-mass collisions. Each row shows a dif-
ferent collision at times equally spaced between t = 2 and t = 3; rp increases
from top to bottom. Contours indicate density on the orbital plane in steps
of a factor of 4; dashed lines show grit, solid lines show gas, with a heavier
contour for ρ = 1. Half-tones indicate shocks.
curves of potential energies U (t). In the grit models (dashed lines) this potential
energy is entirely invested in kinetic energy K(t), which has a maxima opposing
the minima of U (t). In the gas models (solid lines) the gravitational energy is
shared between K(t) and the thermal energy T (t), with the relative apportion-
ment depending on the collision parameters. Head-on collisions produce sharp
peaks in T (t) as gravitational energy heats the gas; the stored thermal energy
drives a subsequent re-expansion and bounce before the system settles down.
On the other hand, the collisions with rp ≥ 1 put most of the available gravi-
tational energy into bulk motion, producing modest peaks in K(t) while barely
disturbing T (t).
Dynamics of Stellar Collisions
7
Figure 4.
First passages for unequal-mass collisions. Each row shows a
different collision at times equally spaced between t = 2 and t = 3; rp increases
from top to bottom. Contours indicate density on the orbital plane in steps
of a factor of 4; dashed lines show grit, solid lines show gas, with a heavier
contour for ρ = 1. Half-tones indicate shocks.
2.1. First passages
Figs. 3 and 4 report on the first passages of all collisions. Here the contours
represent density in the orbital plane, evaluated using SPH-style interpolation;
dashed contours show grit density, while solid contours show gas density. Shaded
areas show where uv > 0; these track the shocks propagating through the gas.
The most obvious differences between the gas and grit models occur in the
head-on collisions (top rows of Figs. 3 and 4). In the equal-mass head-on collision
the two grit polytropes pass directly through each other. As they do so they
are briefly compressed by their mutual gravity; at later times they rebound,
spraying grit particles onto loosely-bound orbits, and fall back to form a single
merged object (van Albada & van Gorkom 1977). The gas polytropes obviously
8
Barnes
can't interpenetrate; a shock develops when they first touch. Subsequently the
shocked gas forms a disk perpendicular to the collision axis and bounded on
the left and right by strong shocks. In the final image of this collision (upper
right in Fig. 3) no trace of the two gas polytropes remains; the shocked gas
forms a single object which expands due to its large thermal energy content. On
the other hand, the grit polytropes, having passed through each other, are still
distinct.
The unequal-mass head-on collision produces a more complex morphology.
As the small grit polytrope plunges through its larger partner it generates a
"wake" -- a region of higher density where gravitational focusing concentrates
grit from the large polytrope towards the collision axis. This wake is similar to
the mutual compression of the two grit polytropes in the equal-mass collision;
in both cases the energy needed is extracted from orbital motion, resulting in
rapid orbit decay. The gas model produces a very different result; as the small
polytrope plows into its partner it creates a strong bow shock. This shock pushes
gas away from the collision axis, evacuating a cavity in the wake of the small
gas polytrope. Meanwhile, a weaker shock travels backward through the small
polytrope. By the last image (upper right in Fig. 4) the bow shock has reached
the surface of the large polytrope, while the cavity has been filled in; the small
gas polytrope has lost most of its forward momentum and lies, still largely intact,
near center of the combined object. In contrast, at this time the grit polytropes
have temporarily separated, though they will soon fall back together.
Many of the dynamical effects just described also play a role in the off-
axis collisions. In the equal-mass rp = 0.5 collision the bodies of the two gas
polytropes are traversed by parallel shocks which slow their relative motion and
increase their entropy. The unequal-mass version of this collision produces an
asymmetric bow shock as the small polytrope moves through the outer layers
of its companion, leaving a temporarily furrow behind.
In contrast, the grit
versions of these collisions again show regions of enhanced density in the wakes
of the interpenetrating polytropes. Less dramatic effects are seen in the wide
collisions; the rp = 1 collisions are basically "kinder and gentler" versions of
their closer counterparts, with weaker shocks and more subtle wakes. Finally,
the grazing collisions (rp = 1.5), shown at the bottom of Figs. 3 and 4, don't
develop significant shocks; these passages involve only gravitational interactions.
2.2. Orbit decay
Although their initial orbits are parabolic, all of the off-center collisions leave
the participants on bound orbits after their first passage. This orbit decay is
generally due to the transfer of orbital energy to internal degrees of freedom, but
the way this transfer takes place may be quite different in gas and grit models.
From Fig. 2 it's already clear that the interplay between gravity and gas-
dynamics in orbit decay is complex. The plots of U (t) for the collisions with rp =
0.5 and 1.0 show two or more minima, each representing a different pericentric
passage. In three of these collisions the grit models return to pericenter before the
corresponding gas models, but the opposite order is seen in the (M1/M2, rp) =
(2, 0.5) collision. Thus compared to the grit models, gas can either accelerate or
delay orbit decay.
Dynamics of Stellar Collisions
9
Figure 5.
Orbital trajectories for all off-center collisions. Solid lines are
gas results, while dashed lines are grit results. Each is labeled by (M1/M2, rp);
the scale bar under each label is 1 unit long.
Fig. 5 shows orbital trajectories for all off-center collisions. These trajec-
tories confirm that in the (M1/M2, rp) = (1, 0.5), (1, 1.0), and (2, 1.0) collisions
the grit models undergo more rapid orbit decay than their gas counterparts, and
that the opposite is true for the (2, 0.5) collision. Moreover, they reveal a further
puzzle; in grazing collisions the outgoing orbits of the gas polytropes are more
tightly bound than those of their grit doppelgangers.
Closer inspection of Fig. 5 suggests a mechanism which might delay orbit
decay in some gas models. Collision (M1/M2, rp) = (1, 0.5), in which the differ-
ence between the gas and grit models is quite dramatic, shows it most clearly:
on their first passage the gas polytropes "bounce" off each other as if elastic. Of
course, this is not too surprising; an ideal gas is perfectly elastic in the absence
of shocks. Moreover, the corresponding panel of Fig. 2 shows a sharp peak in
the thermal energy T (t) during this passage, and Fig. 3 shows that the gas is
compressed to roughly twice its initial density at the moment of closest approach
(t ≃ 2.5). Similar bouncing trajectories are seen in the other off-axis collisions.
On the other hand, the rapid orbit decay of the (M1/M2, rp) = (2, 0.5)
collision points to a different effect. Here as in the head-on cases the orbits
of the gas polytropes decay faster than the orbits of their grit counterparts.
For the head-on collisions this is no great mystery; ram pressure brings the
10
Barnes
gas polytropes to a screeching halt on their first and only passage. A similar
explanation probably applies in this deeply-penetrating off-axis collision; the
small polytrope does so much work plowing through the body of its large partner
that it loses most of its orbital momentum.
The puzzle of the grazing collisions can't be explained in this way as the
gas polytropes barely touch each other. These decays are governed by tides,
and it may seem paradoxical that the gas and grit models with rp = 1.5 di-
verge since the same tidal gradients exist in both cases. However, the bottom
rows of Figs. 3 and 4 show that gas polytropes suffer stronger tidal distortions
in grazing collisions than do their grit counterparts. This make sense; in the
initial polytropes, gas particles are nearly stationary, while grit particles are in
constant motion. Thus during a tidal encounter, a well-placed gas particle can
accumulate more momentum than a wandering grit particle. Now the stronger
tidal response of the gas polytropes is the key; recall that there would be no
orbit decay whatsoever if both participants remained exactly spherical. By be-
ing more responsive, gas polytropes couple orbital motion to internal degrees of
freedom more effectively, and hence in grazing collisions their orbits decay faster
than those of grit polytropes.
2.3. Later passages
Thanks to orbit decay, later passages are slower and closer than first ones. Closer
passages are always more violent, as Figs. 3 and 4 amply illustrate. On the other
hand, the speed of passage affects gas and grit models in different ways. For grit
models, slower passages are generally more disruptive; galaxies collide in exactly
the way that cars don't, because low speeds give tides more time to act. Gas
models, though also sensitive to tides, suffer less from shocks in slow collisions;
the coalescence of two gas polytropes is more like a gentle swirling together than
a head-long plunge.
Every passage pumps orbital energy into internal degrees of freedom, dis-
tending both gas and grit polytropes. After each passages, gas systems relax
into convectively stable configurations; low-entropy material remains centrally
concentrated, while high-entropy material forms an extended envelope. Grit sys-
tems also acquire distended outer envelopes; eq. 5 conserves phase-space density,
so in phase-space such an envelope is really a thin "hyper-ribbon", tidally ex-
tracted from the main body, which still has the high phase-space density of
its source. Nonetheless, this ribbon comes to resemble a smooth envelope as
phase-mixing winds it into a tight spiral.
In basic outline, final encounters resemble mild versions of head-on colli-
sions; either both partners are subsumed into a single object, or one partner
burrows into the center of the other.
3. Remnant Structure
In the aftermath of a merger the wreckage undergoes a last episode of dynamical
relaxation as it evolves towards an equilibrium configuration. This final convul-
sion generally begins at the center and travels outward. In a grit system po-
tential fluctuations during the final merger scatter particles onto loosely-bound
radial orbits; as these outgoing particles reach apocenter they create a caustic or
Dynamics of Stellar Collisions
11
Figure 6.
Density profiles derived by spherical averaging. Solid lines are
gas results, while dashed lines are grit results. Left: remnants of equal-mass
collisions. Right: remnants of unequal-mass collisions; thin lines show profiles
for particles from the small polytropes. Top to bottom: remnants of collisions
with rp = 0, 0.5, and 1, displaced downward by successive factors of 100 for
clarity.
"shell". In a gas system the final merger may create a weak outgoing shock, but
much of the readjustment takes place at subsonic velocities, with no attendant
increase in entropy.
3.1. Density profiles
Fig. 6 presents density profiles for all six merging collisions (those with rp ≤ 1).
These profiles were calculated by evaluating the mass in nested spherical shells.
Despite the differing dynamics of gas and grit mergers, the resulting remnants
have similar profiles over a range of five decades in density. At large radii both
grit and gas densities fall off roughly as ρ ∝ r−4. For grit models this asymptotic
slope results as phase-mixing spreads out a population of particles with a wide
range of energies including both bound and unbound orbits (Jaffe 1987; White
1987).
The unequal-mass mergers yield density profiles with "shoulders" where the
slope briefly becomes shallower as r increases. These compound profiles arise
because the small polytropes resist disruption during the merger process. The
thin lines in the right-hand panel of Fig. 6 show density profiles derived using
only particles originating in from the small polytropes. In every case, the small
12
Barnes
Figure 7.
Initial vs. final binding energies for particles from remnants of
(M1/M2, rp) = (1, 0.0) collision. Left: gas model; right: grit model. Note
that the two plots have different scales. In both plots, E = 0 is escape energy.
polytropes have settled, virtually intact, in the centers of the merger remnants.
For the gas models, this outcome is explained by the lower entropy which the
small polytrope retains throughout the collision and merger; the remnant can't
be convectively stable unless this low-entropy material winds up at the center.
If both partners have the same entropy, how much rearrangement takes
place during a merger? Fig. 7 presents scatter-plots of initial binding energy Ei
vs. final binding energy Ef for particles in both versions of the (M1/M2, rp) =
(1, 0.0) collision. Here Ei was measured with the polytropes at infinite separa-
tion, while Ef was measured from the merger remnant; internal energies were
included in calculating binding energies of gas particles. These plots show that
initial binding energy is a useful, albeit imperfect, predictor of final binding
energy. For the gas model, the correlation between Ei and Ef is fairly strong;
moreover, there is a one-to-one relationship between binding energy and radius,
so the initial and final radii of gas particles are also correlated.
For the grit model things are more complex. As the right-hand panel of
Fig. 7 shows, grit particles have a bimodal distribution of final binding energies.
This arises because the orbital phases of individual grit particles have a strong
bearing on their response to tidal interactions. For example, consider a particle
which reaches an apocenter of its orbit just before the polytropes interpenetrate,
and the subsequent pericenter just after they separate. Such a particle sees a
deeper potential well while falling in than it does while climbing out, so it gains
energy; conversely, a particle on a similar orbit with a different phase may lose
energy. The resulting bimodal energy distribution explains the shoulder in the
density profile of this remnant (Fig. 6, left-hand panel, upper dashed curve).
Tightly-bound particles form the core, while the rest form the ρ ∝ r−4 envelope;
the deficit of particles with Ef ≃ −1.2 accounts for the dip in ρ(r) at r ≃ 0.8.
Dynamics of Stellar Collisions
13
Figure 8. Merger remnants. Contours indicate densities in steps of a factor
of 4; solid lines show gas, while dashed lines show grit. Each remnant is
shown twice, once on a slice through the orbital plane (above), and once on
a perpendicular slice (below).
3.2. Shapes
Fig. 8 presents density contours for the full sample of gas and grit merger rem-
nants. Two slices through each remnant are shown; one slice in the orbital
plane, and one slice at right angles to this plane. As in Figs. 3 and 4, here too
the head-on collisions produce the most striking discrepancies between gas and
grit models. The head-on gas remnants are nearly spherical, as expected for a
self-gravitating gas configuration with no angular momentum. The head-on grit
remnants, on the other hand, are prolate bars aligned with the collision axis.
This is a "long-term" memory of the original collision velocities. Without the
14
Barnes
Figure 9.
Rotation velocities vφ vs. cylindrical radii R for remnants of
(M1/M2, rp) = (1, 1.0) collision. Left: gas model; middle: gas atoms; right:
grit model.
short-range interactions which randomize gas motions, grit systems maintain
anisotropic velocity dispersions which can support such aspherical shapes.
The remnants produced in off-axis collisions have significant amounts of
angular momentum. Most such collisions yield fairly oblate remnants; sliced
perpendicular to the orbital plane, the gas remnants are sometimes "disky"
while the grit remnants seem more "boxy", but in general the density contours
are very similar. The remnants formed by the (M1/M2, rp) = (1, 0.5) collisions
are somewhat unusual in this regard. Both are distinctly non-axisymmetric;
sliced along the orbital plane, the grit model has elliptical density contours at
all radii, while the gas model seems to have a doubled core and a slightly elon-
gated envelope. The grit remnant may be stable, since anisotropy and rotation
together can support a tumbling triaxial structure in equilibrium. The same
may not be true of the non-axisymmetric gas remnant; more experiments could
help settle this issue.
3.3. Kinematics
Fig. 9 compares the kinematics of gas and grit versions of remnants from collision
(M1/M2, rp) = (1, 1.0). This remnant has a relatively large amount of angular
momentum and a major-to-minor axis ratio of roughly 2:1. Here the velocity in
the direction of rotation, vφ, has been plotted against cylindrical radius in the
rotation plane, R. Between the plots of particle velocities for gas (left) and grit
(right), an additional plot shows velocities of gas atoms taking thermal motion
into account (middle). The gas has a very regular velocity field, as reported
in previous studies (eg. Lombardi, Rasio, & Shapiro 1996).
In contrast, the
grit has a broad distribution of velocities at each point; while the overall sense
of rotation is the same as in the gas case, some grit particles even counter-
rotate. But when the grit is compared to the gas atoms, it's evident that these
velocity distributions are very much alike. The similarity of these distributions
is consistent with the good match between the density contours of the gas and
grit versions of this remnant (Fig. 8, upper-right).
Kinematics for the remnants of the head-on collision (M1/M2, rp) = (1, 0.0)
are presented in Fig. 10. Here the radial velocity, vr, has been plotted against the
Dynamics of Stellar Collisions
15
Figure 10.
(1, 0.0) collision. Left: gas model; middle: gas atoms; right: grit model.
Radial velocities vr vs. radii r for remnants of (M1/M2, rp) =
spherical radius, r. This remnant has no angular momentum, so gas velocities
should vanish; the small velocities seen in the left-hand panel indicate that the
remnant is not yet fully relaxed. The grit version also is not yet completely
relaxed, since in perfect equilibrium the distribution should be symmetric with
respect to the line vr = 0. Moreover, the grit plot shows that in phase-space
the envelope of this remnant is not smooth, but consists of narrow ribbons of
particles. These particles populate the upper peak in the final binding energy
distribution for this remnant (Fig. 7, right-hand panel); launched as the two
polytropes interpenetrated, they fell back in coherent streams and have since
been wound up by ongoing phase-mixing. With time this fine structure will
become harder to see, and eventually this plot will look something like the plot
of atomic velocities shown in the middle. But even in the limit t → ∞ some
differences will remain; these include the radial anisotropy of the grit velocity
distribution and bimodal nature of its binding energy distribution.
4. Conclusions
Comparison of gas-dynamical and stellar-dynamical models of colliding systems
reveals many specific differences and hints at some underlying similarities. While
stellar systems interact via gravity alone, gas systems interact via a complex
mixture of gravity, pressure forces, and shocks. This mix greatly increases the
range of dynamical behavior in stellar collisions. Even when interactions between
gas systems are limited to tides, the resulting deformations are stronger than
those seen in equivalent encounters of stellar systems.
Remnants of gas-dynamical and stellar-dynamical mergers also exhibit sig-
nificant differences. Stellar systems have anisotropic velocity distributions, sup-
porting a wide range of remnant morphologies, while gas systems settle into a
small range of equilibria. But at the same time, the stellar and gaseous remnants
studied here have very similar density profiles, and often similar shapes as well.
This seems remarkable, since the physics behind these density profiles is rather
different. The extended envelopes of gaseous remnants contain high-entropy ma-
terial produced by shocks, which are violent, well-localized events resulting from
physical encounters. On the other hand, the envelopes of stellar remnants con-
16
Barnes
tain material extracted from the participants by tides; this material retains the
high fine-grain phase-space density it starts with, and only when coarse-grained
does it approximate the atomic velocity distribution of a gaseous envelope. Two
puzzles remain: first, why do such disparate mechanisms -- shocks and tides --
produce such similar remnant envelopes? And second, how is coarse-graining
-- an operation applied to a stellar system by an observer -- analogous to the
irreversible physical processes which occur in gaseous systems?
Acknowledgments.
I thank Piet Hut, James Lombardi, Frederic Rasio,
and Stuart Shapiro for helpful suggestions. Support for this work was provided
by NASA grant NAG 5-8393 and by Space Telescope Science Institute grant
GO-06430.03-95A.
References
Antonov, V. A. 1962, Vestn. Leningrad Univ., 19, 96 [translated in de Zeeuw, T., ed.
1987, Structure and Dynamics of Elliptical Galaxies (Dordrecht: Reidel), p. 531]
Barnes, J. E. 1998, Galaxies: Interactions and Induced Star Formation, eds. D. Friedli,
L. Martinet, D. Pfenniger (Berlin: Springer-Verlag), 275
Barnes, J., Goodman, J., Hut, P. 1986, ApJ, 300, 112
Chandrasekhar, S. 1939, An Introduction to the Study of Stellar Structure (Chicago:
Univ. of Chicago Press)
Eddington, A. S. 1916, MNRAS, 76, 572
H´enon, M. 1973, A&A, 24, 229
Hernquist, L. 1993, ApJ, 404, 717
Hernquist, L. & Katz, N. 1989, ApJS, 70, 419
Jaffe, W. 1987, Structure and Dynamics of Elliptical Galaxies, ed. T. de Zeeuw (Dor-
drecht: Reidel), p. 511
Kalnajs, A. 1972, ApJ, 175, 63
Lacey, C. G., Ostriker, J. P. 1985, ApJ, 299, 633
Lombardi, J. C., Rasio, F. A., Shapiro, S. L. 1996, ApJ, 468, 797
Lynden-Bell, D. 1962, MNRAS, 124, 279
Monaghan, J. J. 1992, ARA&A, 30, 543
Monaghan, J. J., Gingold, R. A. 1983, J. Comp. Phys., 52, 374
Nelson, R. P., Papaloizou, J. C. B. 1993, MNRAS, 265, 905
Nelson, R. P., Papaloizou, J. C. B. 1994, MNRAS, 270, 1
Sills, A., Lombardi, J. C. 1997, ApJ, 484, L51
Toomre, A. 1964, ApJ, 139, 1217
van Albada, T. S., van Gorkom, J. H. 1977, A&A, 54, 121
White, S. D. M. 1987, Structure and Dynamics of Elliptical Galaxies, ed. T. de Zeeuw
(Dordrecht: Reidel), p. 339
|
0709.1216 | 1 | 0709 | 2007-09-08T15:10:35 | The Persian-Toledan Astronomical Connection and the European Renaissance | [
"astro-ph"
] | This paper aims at presenting a brief overview of astronomical exchanges between the Eastern and Western parts of the Islamic world from the 8th to 14th century. These cultural interactions were in fact vaster involving Persian, Indian, Greek, and Chinese traditions. I will particularly focus on some interesting relations between the Persian astronomical heritage and the Andalusian (Spanish) achievements in that period. After a brief introduction dealing mainly with a couple of terminological remarks, I will present a glimpse of the historical context in which Muslim science developed. In Section 3, the origins of Muslim astronomy will be briefly examined. Section 4 will be concerned with Khwarizmi, the Persian astronomer/mathematician who wrote the first major astronomical work in the Muslim world. His influence on later Andalusian astronomy will be looked into in Section 5. Andalusian astronomy flourished in the 11th century, as will be studied in Section 6. Among its major achievements were the Toledan Tables and the Alfonsine Tables, which will be presented in Section 7. The Tables had a major position in European astronomy until the advent of Copernicus in the 16th century. Since Ptolemy's models were not satisfactory, Muslim astronomers tried to improve them, as we will see in Section 8. This Section also shows how Andalusian astronomers took part in this effort, which was necessary in the path to the Scientific Revolution. Finally, Section 9 presents the Spanish influence on the eve of the Renaissance. | astro-ph | astro-ph | Academia Europaea 19th Annual Conference
in cooperation with:
Sociedad Estatal de Conmemoraciones Culturales,
Ministerio de Cultura (Spain)
“The Dialogue of Three Cultures and our European Heritage”
(Toledo Crucible of the Culture and the Dawn of the Renaissance)
2 - 5 September 2007, Toledo, Spain
Chair, Organizing Committee: Prof. Manuel G. Velarde
The Persian-Toledan Astronomical Connection and the European Renaissance
M. Heydari-Malayeri
Paris Observatory
Summary
This paper aims at presenting a brief overview of astronomical exchanges between the
Eastern and Western parts of the Islamic world from the 8th to 14th century. These cultural
interactions were in fact vaster involving Persian, Indian, Greek, and Chinese traditions. I will
particularly focus on some interesting relations between the Persian astronomical heritage and
the Andalusian (Spanish) achievements in that period. After a brief introduction dealing
mainly with a couple of terminological remarks, I will present a glimpse of the historical
context in which Muslim science developed. In Section 3, the origins of Muslim astronomy
will be briefly examined. Section 4 will be concerned with Khwârizmi, the Persian
astronomer/mathematician who wrote the first major astronomical work in the Muslim world.
His influence on later Andalusian astronomy will be looked into in Section 5. Andalusian
astronomy flourished in the 11th century, as will be studied in Section 6. Among its major
achievements were the Toledan Tables and the Alfonsine Tables, which will be presented in
Section 7. The Tables had a major position in European astronomy until the advent of
Copernicus in the 16th century. Since Ptolemy’s models were not satisfactory, Muslim
astronomers tried to improve them, as we will see in Section 8. This Section also shows how
Andalusian astronomers took part in this effort, which was necessary in the path to the
Scientific Revolution. Finally, Section 9 presents the Spanish influence on the eve of the
Renaissance.
1. Introduction
Before dealing with the main topics of this paper, it seems necessary to comment first on
three widely used terms in this field: Arab/Arabic astronomy, Islamic astronomy, and zij.
Arab/Arabic is not meant as an ethnic but rather a linguistic term. In fact a large
number of Non-Arab scholars, mainly Persians, Turks, and Spanish people, wrote their works
in Arabic. Even so, many astronomical works were also produced in other languages of this
civilization, especially Persian and in the later centuries Turkish. For example, the main zijs
were originally written in Persian, a notable example being the Ulugh Beg’s (c. A.D. 1394-
1449) zij, a landmark in precise observations before the Renaissance. We also note a disparity
with respect to Western scholars who wrote in Latin. As far as these scholars are concerned,
the Latin adjective is not specified (e.g., the expressions like “the Latin astronomer
Copernicus”, “the Latin physicist Newton”, or “the Latin philosopher Leibnitz” are not used).
As for the term Islamic, it should be taken in the sense of the civilization rather than
the religion, because much of the astronomy was secular. Moreover, many non-Muslims
within the Islamic civilization contributed to this science and must be acknowledged. Once
again, we find the above-mentioned disparity, since the term Christian, which refers also to a
civilization, is not used either (e.g. Galileo and Newton are not usually referred to as
“Christian scientists”).
Zij is the generic name applied to books in Arabic and Persian that tabulate parameters
used for astronomical calculations of positions of the Sun, the Moon, and the five planets of
antiquity. The word is derived from Middle Persian zig, variant zih, meaning “cord, string”
(Modern Persian zeh “cord, string”), from Avestan jiiā- “bow-string”, cognate with Sanskrit
jiyā- “bow-string”, Proto-Indo-European base *gwhi- “thread, tendon” (from which derive
also Greek bios “bow”, Latin filum “thread”, Russian žica “thread”). The term zig originally
referred to the threads in weaving, but because of the similarity between the rows and
columns of astronomical tables and the parallel threads, it came to be used for an
astronomical table, and subsequently a set of tables.
2. A glimpse of the historical background
Cultural developments in the course of history are not detached from underlying
social/political events. In order to better understand the advent of Islamic science, it would be
interesting to have a fast glimpse of the historical background.
The first Islamic state, established by the Umayyad dynasty (661-750), lasted for 89
years. There were social turmoil in various parts of the vast conquered territories and in
particular Iranian resistance movements opposed the Arab domination, especially since the
Umayyads considered non-Arabs as mawali, people of lowly status.
The Umayyads were overthrown by the leader of a revolutionary movement Abu
Moslem Khorâsâni (Persian name Behzâdân), who enthroned Abu al-Abbâs as-Saffâh, a
member of the prophet’s lineage. This was the beginning of the Abbasid dynasty (750-1258),
which lasted over five centuries, although Saffâh himself reigned for only four years.
The only Umayyad survivor, Abd ar-Rahmân I, escaped to Andalusia where he
established himself as an independent Emir (756-788). More than a century later one of his
successors, Abd ar-Rahmân III (912-961), assumed the title of Caliph, establishing Cordoba
as a rival to the Abbasid capital.
The Abbasid era was substantially influenced by the Persian culture and tradition of
government. A new position, that of vizier (Arabic from Middle Persian vicīr “decree,
decision”, from vicīrītan “to decide, to judge, to distinguish”; vicīrtār “judge, arbitrator,
decider”), was created and many Abbasid caliphs were eventually relegated to a more
ceremonial role than under the Umayyads. Several prominent viziers, serving Hârun ar-
2
Rashid (786-809) and his son al-Ma’mun (813-833), were members of the Persian Barmakid
family of Buddhist faith. Moreover, al-Ma’mun’s mother was Iranian, and he himself had
grown up in Khorâsân, the Eastern province of Iran.
During the second caliph al-Mansur (754-775) the capital was moved from Damascus
to Baghdad, not far from Ctesiphon, the ancient capital of Iranian Sassanids. The designers
hired by al-Mansur to lay the city’s plan were two Persians: Naubakht, a former Zoroastrian,
and Mâshâ’allah, a Jew from Khorâsân. The two men also determined an astrologically
auspicious date for the foundation of the city: 30 July 762.
It is also notable that the city name Baghdad is Persian, meaning “god-given” or
“God’s gift”, from bagh “god, lord” + dâd “given”. The first component derives from Old
Persian baga-, Avestan baγa- “lord, divider” (from bag- “to allot, share”), cognate with
Sanskrit bhága- “part, portion”, Proto-Indo-European base *bhag- “to divide”; cf.
Slavic/Russian bog “god”, Greek phagein “to eat” (originally “to have a share of food”). The
second component dâd, from dâdan “to give”, Old Persian/Avestan dā- “to give, grant”,
Proto-Indo-European base *do- “to give”; cf. Sanskrit dadáti “he gives”, Greek didonai
“to give”, Latin datum “given”, Russian dat’ “to give”.
The reign of Hârun ar-Rashid and his successors fostered an age of intellectual
activities. A cultural center, the House of Wisdom (Bait al-Hikmah), was set up, which was
reminiscent of the Persian Sassanid academy of Gundishapur. An intense translation activity
was undertaken and all sorts of books were translated from Middle Persian, Sanskrit, Syriac,
and Greek into Arabic. In just a few decades the major scientific works of antiquity, including
those of Galen, Aristotle, Euclid, Ptolemy, Archemides, and Apollonius, were translated into
Arabic. The main translators were Hunayn ibn Ishâq (c. 809-873), a Nestorian Christian with
an excellent command of Greek, and Thâbit ibn Qurra (c. 836-901), a Helenized pagan from
Harran, a town in northern Mesopotamia (today Turkey), that was the center of a cult of “star
worshippers”.
The paper needed for books was abundant, since Muslims had learned the techniques
of paper making from the Chinese. In fact the Chinese prisoners of war captured in the battle
of Talas (731) were ordered to produce paper in Samarkand and by the year 794 a paper mill
was installed in Baghdad.
3. Origins of Muslim astronomy
At the advent of Islam Arabs did not have an elaborated/documented astronomy. The first
astronomical documents translated into Arabic were of Indian and Persian origins. These
translations set the basis for the first Muslim astronomical works. Greek astronomy,
represented by Ptolemy, was introduced later, but it gained a fully dominant position owing to
its predictive capacity. Here are the main founding sources:
The Persian work was Zij-e shâh or Zij-e shahryâr, originally composed for the
Sassanid emperor Khosrow I (Anushirwân) about the year 550. It was translated into Arabic
by Abu al-Hasan al-Tamimi and commented on by Abu Ma’shar (Albumasar) of Balkh. It
3
became the basis for example for the work of the previously mentioned astronomers
Naubakht and Masha’allah. Zij-e shâh contained some elements of Indian and Greek
traditions. It also had its specific Persian particularities, mainly the basic year for the tables,
which was the coronation date of the last Sassanid emperor Yazdegerd III (16 June 632), and
the Solar year based on Nowruz, or spring equinox. The Yazdegerd III’s era was in use in
Muslim astronomy during several centuries, before being replaced with the Hijra. As an
interesting particularity of the Zij-e Shâh, the day started from midnight.
As to the Indian sources, several works have been mentioned in early Muslim
astronomy, the main one Siddhānta (Sanskrit meaning “established end, final aim, doctrine,
concept”) attributed to Brahmagupta (598-670). This work was brought to Baghdad sometime
around 770 by an Indian political delegation, which had an astronomer named Kanka. The
book was translated into Arabic by al-Fazâri and Ya’qub ibn Târiq, who were assisted by the
Indian astronomer. The Sanskrit term was later corrupted to Sindhind in Arabic.
The Greek source was Ptolemy’s Almagest, dating from about A.D. 150. The
Ptolemy’s Mathematike Syntaxis “Mathematical Compilation”, in later antiquity known
informally as Megale Syntaxis or Megiste Syntaxis “The Great Compilation”, was translated
from Greek into Arabic in the 9th century during the translation campaign launched by al-
Ma’mun. It was translated several times, during which the title word Megiste was transformed
into al-majisti. The earliest translators were the above-mentioned Hunayn ibn Ishâq and
Thâbit ibn Qurra. By this time the work was lost in Europe.
The Persian astronomer Ahmad Farghâni (Alfraganus) presented a thorough
descriptive summary of Almagest in his textbook Jawâmi’, known as Elements, written
between 833 and 857. It was translated into Latin in the 12th century and was widely studied
in Europe until the time of Regiomontanus (1436-1476). Farghâni also composed a very
important treatise on the astrolabe around 856. Although the astrolabe was a Greek invention,
the earliest dated instrument that has been preserved comes from the Islamic period. The only
extant ancient treatise on the astrolabe is due to Johannes Philoponos, written in the first part
of the 6th century.
The first Muslim astronomer who based his work principally on Ptolemy was al-
Battâni (c. 853-929, born in Harran), who made his observations at al-Raqqa in Syria.
Ptolemy’s work re-entered Europe from its Arabic versions with the transformed name
Almagest in the medieval Latin translations.
4. Khwârizmi’s zij
Zij of Sindhind, written about 820 by Khwârizmi, was the first major astronomical work in the
Muslim world. It was mainly based on Indian/Persian astronomy. Interestingly, Khwârizmi’s
zij had its greatest long-term influence in Muslim Spain and Western Europe through the
incorporation of some of its material in the Toledan Tables.
Abu Ja’far Muhammad ibn Musâ Khwârizmi (c. 780- c. 850) was a key figure in the
history of algebra, an astronomer and geographer. The epithet Khwârizmi indicates that he or
4
his forebears came from the Persian region of Khwârizm, the present-day Khiva in
Uzbakistan. The historian Tabari (c. 838-923) gives him the additional epithet Majusi (related
to magus), meaning Zoroastrian. This would have been possible at that time for a man of
Persian origin. However, the preface of his Algebra (if effectively written by himself) shows
him a pious Muslim. Anyhow, Tabari’s designation could mean also that his ancestors, and
perhaps himself in his youth, had been Zoroastrian.
Zij of Sindhind was in particular based on the Iranian solar year with the starting era
that of Yazdegerd III, as previously indicated. The Sun, the Moon, and each of the five
planets known in antiquity had a table of mean motion and a table of equations (variations
with respect to mean values). In addition, there were tables for computing eclipses, solar
declination and right ascension, and various trigonometric tables. The form of a set of tables
closely resembled that made standard by Ptolemy. But most of the basic parameters in the Zij
(the mean motions, the mean positions at epoch, positions of apogee and the node) were
derived from Indian astronomy. The maximum equations were taken from Zij of Shâh. The
fundamental meridian was that of Arin, lying 70° east of Baghdad. Arin was a corruption of
Ujjayni (present-day Ujjain), a city situated in central India, which was the “Greenwich” of
the ancient Indian astronomy.
The original Arabic version is lost. A Latin translation exists carried out by the
English scholar Adelard of Bath (c. 1080-1152) in the early 12th century. This translation was
made not from the original, but from a revision executed by a Spanish astronomer, al-Majriti
(c. 950- c. 1007).
The Zij continued to be used in classrooms and commented on even after al-Battâni,
the aforementioned Mesopotamian astronomer produced his great work (Zij al-Sâbi), based
principally on the Almagest and his own observations. Battâni is considered as the first
Muslim astronomer to carry out new, systematic observations since the time of Ptolemy.
Khwârizmi is also recognized for his book on algebra: al-kitâb al-mukhtasar fi hisâb
al-jabr w’al-muqâbala “the Compendious Book on Calculation by Completion and
Balancing”. The term algebra in the European languages derives from al-jabr “completion,
restoration” used by Khwârizmi. It should be underlined that at the early days of Algebra
mathematical symbols were not used and one had to resort to phrases instead.
He is also remembered for being at the origin of the transfer of Hindu numerals
(commonly called Arabic numerals) to Europe. Likewise, the term algorithm derives from his
name, from French algorithme, refashioned under mistaken connection with Greek arithmos
“number”, from Old French algorisme, from Middle Latin algorismus “the Arabic system of
arithmetical notation”, a distorted transliteration of al-Khwârizmi.
Judging from what has remained from Khwârizmi, his importance is mainly due to the
fact that he was the first to scholarly introduce the Indian/Persian traditions in
astronomy/mathematics into the Muslim world.
5. Khwârizmi’s influence on Andalusian astronomy
Abd ar-Rahman II, the Umayyad Emir of Cordoba (822-852), was an amateur of books on
philosophy, medicine, and music. He also loved astrology and his trustworthy astrologer
5
occupied a high rank in the court. It is possible that his liking for astrology stemmed from two
notable astronomical events. The almost total solar eclipse of 17 September 833 drove the
striken mob to the Cordoba’s grand mosque to carry out the ritual fright prayer. The other
event was an intense meteor shower between 2 April and 18 May 839. It was during Abd ar-
Rahman II’s reign that a version of Khwârizmi’s zij was introduced.
The real development of the Andalusian science took place in the second half of the
10th century. The first prominent Andalusian astronomer was Maslama ibn Ahmad al-Majriti
(c. 950- c. 1007) of Madrid. He is particularly renowned for his version of Khwârizmi’s zij, in
which he changed the chronology from the Persian epoch (as explained before) to the Muslim
Hijri calendar. Further, Majriti transferred the standard meridian from Arin (explained earlier)
to Cordoba. Majriti’s version of the Zij was translated into Latin by Adelard of Bath and was
commented upon by Ibn al-Muthannâ, a work being extant in a Hebrew version. It seems
moreover that Majriti’s work improved the calculation methods of Khwârizmi.
Although Andalusian astronomers knew Ptolemy’s work at least from the 10th
century, they never fully abandoned the Sindhind tradition. Even about a century after Majriti,
the Zij of Jayyân, set up by Ibn Mu’âdh al-Jayyani (d. 1093), was an adaptation of
Khwârizmi’s zij to the coordinates of the city of Jaén in south-central Spain.
6. Flourishing of Andalusian astronomy
Andalusian astronomy flourished in the 11th century with Abu Ishâq Ibrâhim an-Naqqâsh (d.
1100), surnamed az-Zarqâli (Spanish transcription Azarquiel), from zarqâ’ (“blue”; “the blue-
eyed one”). He was the first in the history of astronomy to unambiguously state the proper
motion of the solar apogee with respect to stars and distinguish it from the precession of the
equinoxes. The solar apogee was the farthest distance of the Sun from the Earth, when the
Sun had its smallest apparent size. According to Zarqâli’s observations, the motion of the
solar apogee amounted to 12″.04 per year. Resulting from 25 years of solar observations, first
at Toledo and then at Cordoba, this measurement is highly remarkable compared to modern
observations, which yield a displacement of 11″.6 annually. Today we know that the motion
of the solar apogee results from the rotation of the line of apsides, or major axis, of the
Earth’s orbit due to gravitational perturbation by other planets.
Zarqâli’s great manual skill allowed him to construct precision instruments for
astronomical use. Perhaps inspired by the Toledan astronomer Ali ibn Khalaf, he constructed
a “universal astrolabe” that could be used at any latitude. This innovation by Andalusian
astronomers removed the inconvenience of having to change the plate (safiha) for each
latitude. He also built the water clocks of Toledo, which were capable of determining the
hours of the day and night and indicating the days of the lunar month. The clocks were in use
until 1133, when Hamis ibn Zabara (during Alfonso VII) tried to discover how they worked;
they were dismounted but could not be reassembled.
Zarqâli’s results had far-reaching consequences. Up till now Andalusian astronomy
was subject to the pre-eminence of Eastern astronomy. Zaqâli starts a new period in which the
Andalusian astronomers confirm the excellence of their research. The achievements of the
11th century Andalusian astronomers were transmitted and praised in the Middle East,
announcing the beginning of a historical shift in the geographical focus.
6
It would be interesting to recall other notable features of Zarqâli’s epoch in a broader
scope. Zarqâli was contemporary with the Persian mathematician, astronomer, and poet Omar
Khayyâm (1048-1131). He was the first in the history of mathematics to undertake a
systematic study and classifiaction of equations of degree ≤ 3 and to elaborate a geometrical
solution for them. He is also known for his outstanding reform of the Iranian solar calendar,
and his poems (rubaiyat). Interestingly, when Zarqâli was 25 years old one of the rarest
astronomical events happened: the explosion of the famous supernova. On July 4, 1054,
Chinese astronomers noted a "guest star" in the constellation Taurus. This star became about
four times brighter than Venus in its brightest light, or about mag –6; it was visible in daylight
for 23 days and in the night sky for 653 days. It is unlikely that Muslim astronomers did not
see this phenomenon, but the fact that they did not care about this extraordinary event
suggests that they were exclusively concerned with planetary motions.
7. Toledan Tables, Alfonsine Tables
Zarqâli is above all renowned for his contribution to the Toledan Tables which were probably
compiled after 1068. This work, which represents the first original development of
Andalusian astronomy, was extremely influential in Europe for three centuries until the
advent of the Alfonsine Tables. It was adapted to local meridians almost all-over Europe (e.g.
Pisa, London, Toulouse, Paris, Marseille) until the 13th century. The fact that a large number
of copies were made of the Toledan Tables in the 14th and even 15th century implies that
they were in use even after the Alfonsine Tables were introduced.
The main sources for the bulk of the table collections were those of Khwârizmi
(mainly planetary latitudes), Battâni (planetary equations), and Ptolemy. In fact the oldest
version of the Toledan Tables was mainly modeled on Khwârizmi’s Sindhind, but had
admixture from Battâni. In addition, the oldest versions of the Toledan Tables preserve some
tables of Khwârizmi that are rare or absent elsewhere.
The Toledan Tables also incorporated the thory of trepidation attributed to Thâbit ibn
Qurra (c. 830-901). Trepidation was a spurious oscillation of the equinoxes thought to have a
period of about 7000 years. In order to explain trepidation, Thâbit was said to have added a
new sphere to the eight Ptolemaic spheres beyond the sphere of fixed stars. However, Al-
Battâni rejected trepidation. In fact trepidation dominated the medieval astronomy. The
original Arabic version of the Toledan Tables has been lost, but two Latin versions have
survived, one by Gerard of Cremona (12th century) and one by an unknown author.
The Alfonsine Tables were drawn up in Toledo to correct the anomalies in the
Toledan Tables, which they rapidly superceded sometime after 1320. Alfonso X, el sabio,
king of Castile and Léon from 1252 to 1284, encouraged his astronomers and translators to
prepare a major corpus of astronomical text in Castilian. The starting point of the Alfonsine
Tables is January 1, 1252, the year of king’s coronation (1 June). The original Spanish
version of the tables is lost, but a set of canons (introductory instructions) for planetary tables
are extant. They are written by Isaac ben Sid and Judah ben Moses ha-Cohen, two of the most
active collaborators of Alfonso X. The Alfonsine Tables were the most widely used
astronomical tables in the Middle Ages and had an enormous impact on the development of
European astronomy from the 13th to 16th century. They were replaced by Erasmus
Reinhold’s Prutenic Tables, based on Copernican models, that were first published in 1551.
7
The Latin version of the Alfonsine Tables first appeared in Paris around 1320, where a
revision was undertaken by John of Lignères and John of Murs, accompanied by a number of
canons for their use written by John of Saxony. There is a controversy as to the exact
relationship of these tables with the work commissioned by the Spanish king. Surprisingly,
the earliest evidence of their use in Spain dates back to about 1460 in Salamanca when a
Polish astronomer, Nicholaus Polonius, arrived and brought them with him. It is quite
possible that the astronomers of the early 14th century working in Paris did a fine job in
adapting previous astronomical material. However, if they created something quite new, why
did they keep calling it Alfonsine Tables? The fact that the authors of the Parisian tables did
not choose a new title for their work suggests that they did not consider it to be fundamentally
different from the original Alfonsine Tables. In 1483 the Parisian Alfonsine Tables were
printed in Venice for the first time, followed shortly afterwards by other editions, all of them
by printers in that city.
8. Andalusian criticism of Ptolemy
Ptolemy’s Almagest, with its 13 chapters, dominated medieval astronomy. Notwithstanding,
Ptolemy’s model was an inadequate representation of planetary motions. As an extreme
example, according to Ptolemy’s model for the Moon, our satellite should appear to be almost
twice as large when it is full than it is at quadrature, which is an absurdity since it is not seen
as such.
The Muslim scholars did not follow Ptolemy blindly but disagreed on a number of
points. For instance, the Egyptian Ibn al-Haytham (965-1040), the greatest authority on optics
in the Middle Ages, wrote a treatise, Al-shukuk ‘alâ Batlamyus (Doubts about Ptolemy),
criticizing the Ptolemaic system for its complexity and incompatibility with the Aristotelian
physics. Similarly, the Persian polymath Abu Rayhân Biruni (973-1048), author of Qânun al-
Mas’udi, considered as the most important Muslim astronomical text, rejected Ptolemy’s
opinion about the immobility of the solar apogee on the basis of many astronomical
observations. Mostly, the criticisms of Ptolemy were mainly philosophically motivated.
Aristotle was a great authority who dominated every field of knowledge, including physics,
whereas Ptolemy did not have such a standing. The objections had three main reasons:
1) In the Aristotelian cosmology the Earth was situated at the center of the Universe. In
contrast, for Ptolemy the Earth did not occupy the central place. The main sphere (orb), called
deferent, was centered on a point halfway between the Earth and another point invented by
Ptolemy himself, termed equant. This point was meant to bring Ptolemy’s planetary theory
closer to observations. The epicycle rotated on the deferent with uniform motion, not with
respect to the Earth, nor around the center of the deferent, but with respect to the off-center
point equant. However, this does not work since one might think in terms of a sphere moving
at a uniform speed around an axis that does not pass through its center.
2) Aristotle believed that circular motion with uniform speed was the natural property of
celestial objects. The equant point introduced by Ptolemy caused the epicycle center to move
with variable speed on the deferent circle.
8
3) The Ptolemaic epicycles and deferents were abstract, geometrical concepts, while in the
Aristotelian cosmology the spheres were physical entities to which celestial bodies were
attached.
In Andalusia, after Zarqâli the Ptolemaic model was much debated and its criticism
gained momentum, mainly by philosophers. Ibn Rushd (Averroes, 1126-1198) and Ibn Bâjja
observed spots on the solar disk which they ascribed to the transits of Mercury and Venus.
The idea of spots on the Sun was not admissible since Aristotle’s doctrine maintained that the
heavens were incorruptible. Anyhow, putting forward the transit argument meant a criticism
of Ptolemy, since he had argued that the lines of sight joining these planets to Earth never
cross the Sun. These transits were a matter of debate among Andalusian thinkers. Further,
Averroes rejected Ptolemy’s eccentric deferents and argued for a strictly concentric model of
the Universe.
The 12th century Andalusian astronomers were divided between those who followed
Zarqâli’s tradition, like Abu as-Salt, Ibn al-Kammâd (c. 1100), Ibn al-Haïm (c. 1205), and
Jâber ibn Aflah (1100-1150) and tried to improve the “orthodox” astronomy, and those who
were critical of Ptolemy. Ibn Rushd, Maïmonid, Ibn Bâjja, and Ibn Tufayl dreamed of an
astronomy that would be in agreement with the Aristotelian physics, based on three kinds of
motions: centrifugal, centripetal, and circular about a center that should be identified with
Earth.
Al-Bitruji (Alpetragius, d. c. 1190) is the only Aristotelian philosopher to formulate an
embryo of a strictly geocentric system. However, there was nothing really new in this since
Democritus had allegedly stated similar ideas. Anyhow, Bitruji’s system was quite qualitative
and he was unable to calculate the planetary tables with his model. For example, in Bitruji’s
system Saturn could on occasion deviate from the ecliptic by as much as 26 degrees (instead
of the required 3 degrees).
The most significant improvement of Ptolemy’s model to comply with absolute
circularity was obtained in the 13th century by the Persian astronomer, mathematician,
founder of the Marâgha observatory, and political figure Nasireddin Tusi (1201-1274). He
devised a theorem on the combination of regular circular motions (now called the Tusi
couple) that generated linear motion. A circle of radius R rotates inside a circle of radius 2R.
The smaller circle rotates at twice the speed of the larger one and in opposite direction. The
initial tangent point will travel in linear motion back and forth along the diameter of the larger
circle. Today we know that the linear motion is a particular case of the family of hypocycloid
curves. A change in relative sizes and speeds of the circles will produce multi-cusped curves.
This innovation allowed Tusi to account for non-uniform motions of the planets. It explained
how the epicycle could move uniformly around the equant, and still oscillate back and forth
toward the center of the deferent. In fact the Tusi couple was the only new mathematical
model for planetary motions from the time of Ptolemy. Tusi used the theorem successfully in
his Tadhkira to reconstruct models for the Moon and the upper planets. However, as he
himself points out, he could not generalize the solution to Mercury. Tusi’s work was
elaborated by other astronomers at the Maragha observatory, Mu’ayyad ad-Din al-‘Urdi (d.
1266) and Qutbeddin Shirâzi (d. 1311). Finally, following Tusi’s work, a completely
concentric rearrangement of the planetary mechanisms was achieved by the Syrian
9
astronomer Ibn ash-Shâtir (1304-1376), who succeeded in eliminating not only the equant but
also certain other objectionable circles from Ptolemy’s models and thus placing the Earth at
the center of the Universe.
9. The eve of Scientific Revolution
On the eve of the Renaissance Spain was the major interface between Muslim science
and the rising European scientific activities. Many of the treatises written by Muslim scholars
reached Europe via Spain, where they were translated into Latin. The astrolabe was re-
introduced in the West through Spain in the 11th century. In about 1390 the English poet
Geoffrey Chaucer (c. 1340-1400), inspired by such translations, wrote an essay of the
astrolabe. It is possible that scientific activity centered at Oxford at the time contributed to the
surge of interest in the device. On the astrolabes of that period one finds typical sets of Arabic
star names written in Gothic Latin letters, for example Vega, Altair, Rigel, Alpheratz, and so
on. Thus, as a result of the astrolabe tradition of Andalusian astronomy, most navigational
stars today have Arabic names.
The Muslim attempts to criticize Ptolemy could not spare the Ptolemaic model.
Nonetheless, they were necessary steps in the path leading to the Scientific Revolution. They
certainly helped Nicolaus Copernicus (1473-1543) to come up with his revolutionary
heliocentric model, more especially since he included the concepts of epicycle and deferent to
explain the planetary motions. He was courageous and smart enough to replace the Earth with
the Sun and thus open a new era in the history of human thought. Nonetheless, he preserved
the geometrical/mathematical tools used by Ptolemy, since he still viewed celestial motions as
circular rather than elliptical and so still required the equant to describe elliptical motions.
The three Keplerian laws of planetary motions involving elliptical orbits (around 1610) and
Newton’s universal theory of gravitation (Principia 1687) were not yet discovered.
Copernicus knew of the results by Muslim scholars, and cites Battâni as well as the
Andalusians Zarqâli and Ibn Rushd. He made use of the Tusi couple in his models in De
revolutionibus orbium coelestium (On the Revolutions of the Celestial Spheres). There is
growing consensus that Copernicus, living some three centuries after Tusi, became aware of
Tusi’s result in some way, probably through Byzantine intermediaries, although an exact
chain of transmission has not yet been identified. In brief, Copernicus put an end to a long
period in the Middle Ages in which Muslim astronomers prevailed. Once the human thought
was freed from the shackles of geocentric model, unprecedented discoveries and revolutions
occurred owing to great scientific figures like Kepler, Galileo, Newton, and others.
Copernicus closed a parenthesis between the Golden Greek/Roman epoch and the
Renaissance. Europe started to excel in other fields than art and architecture.
The Alfonsine Tables created by Toledan astronomers in the 13th century were the
last major astronomical work by Spanish astronomers before the Renaissance. After the fall of
the Almohad dynasty (1121–1269), the Muslim Spain was reduced to the Nasrid kingdom of
Grenada (1232-1492). In spite of Alfonso X’s policy to attract Muslim scholars, they
preferred to settle in Grenada or emigrate to North Africa or the Middle East. The Christian
Spain gradually established itself as a European power. The discovery of America under
10
Spanish patronage was also a serious strike on Ptolemy’s worldview, since his Geographia
ignored the new continent. The discovery proved Ptolemy wrong as far as the form and
contents of the terrestrial surface were concerned, and this likewise had repercussions for his
celestial ideas. Further, the discovery boosted navigation and the need for exact time
determinations at different longitudes. More generally, it opened a new world with a wealth
of plants, animals, commodities, and peoples previously unknown to Europeans.
Another parallel which can be drawn between Persia and Andalusia is of linguistic
nature. In the wind of the Renaissance “national” languages started to step out of the
occultation by the lingua franca. In particular, the creation of a standardized Spanish
language based on the Castilian dialect began in the 13th century with King Alfonso X, who
encouraged scholars to write original works in Castilian and translate histories, chronicles,
scientific, legal, administrative, and literary works from other languages (principally Latin,
Greek, and Arabic). This was a tremendous job since Castilian had never been used
previously for writing on technical matters in particular astronomy. The Alfonsine scholars
had to exert much effort to build a new specialized language and create the necessary
terminology. As for Persian, although it had resisted Arabic and had created immense literary
masterpieces like Shâh-nâmeh by Ferdowsi (935-1020) as well as scientific and philosophical
works by scholars like Ibn Sinâ (Avicenna, c. 980-1037) and Biruni, the 13th century
witnessed a kind of surge in producing science and philosophy in Persian. This was mainly
encouraged by the above-mentioned astronomer/mathematician/politician Tusi, who himself
wrote several works in Persian: Zij-e Ilkhâni (based on observations at the Marâgha
Observatory), an astronomical treatise entitled Moiniyyé, and a work on ethics, Akhlâq-e
Nâseri. He even found time to translate into Persian the Book of Fixed Stars, by the Persian
astronomer Abd ar-Rahmân Sufi (903-986), originally written in Arabic.
References
Bagheri, M. 1998, The Persian Version of Zij-i Jâmi’ by Kušyâr Gilâni, in « La Science dans le Monde Iranien à
l’Epoque Islamique », Institut Français de Recherche en Iran, Téhéran
Breggren, L. 1997, Islamic Astronomers and the Equant, JHA, xxviii, 270
Calvo, E. 2005, Science in al-Andalus in the Lifetime of Khayyâm, Farhang (Institute for Humanities and
Cultural Studies, Tehran), Vol. 18, No 53-54, p. 329
Chabás, J.; Goldstein, B.R. 2000, Astronomy in the Iberian Peninsula: Abraham Zacut and the Transition from
Manuscript to Print, American Philosophical Society
Chabás, J.; Goldstein, B.R. 2003, The Alfonsine Tables of Toledo, Kluwer, Dordrecht
Copernic, N. Des révolutions des orbes célestes, traduction by Alexandre Koyré, 1998, Diderot Editeur
Dictionary of Scientific Biographies [DSB]. 1978, New York
Gingerich, O. 1986, Islamic Astronomy, Scientific American, 256, 74
Gingerich, O. 2002, An Annotated Census of Copernicus’ De Revolutionibus (Nuremberg, 1543 and Basel,
1566), Brille
Hartner, W. 1977, The Role of Observations in Ancient and Medieval Astronomy, JHA, viii, p. 1
Heydari-Malayeri, M., The Iranian Calendar, http://aramis.obspm.fr/~heydari/divers/calendar.html
Hugonnard-Roche, H. 1997, Influence de l’astronomie arabe en Occident médiéval, in Rashed 1997, p. 309
Kennedy, E.S. 1994, Tusi’s Cosmology, JHA, xxv, p. 321
King, D.A. 1987, Ninth-Century Islamic Astronomy, JHA, xviii, p. 284
Kremer, R.L. 2007, “Abbreviating” the Alfonsine Tables in Cracow, JHA, xxxviii, p. 283
Kokowski, M. 2004, Copernicus's Originality: Towards Integration of Contemporary Copernican Studies,
Polish Academy of Sciences, Warsaw
Morelon, R. (Edited and translated by). 1987, Thâbit ibn Qurra: OEuvres d’Astronomie, Les Belles Lettres, Paris
11
Neugebauer, O. 1957, The Exact Sciences in Antiquity, Dover Publications, Inc., New York
Pedersen, F. 2002, The Toledan Tables (A review of the manuscripts and textual versions with an edition), Royal
Danish Academy of Sciences and Letters
Ragep, F.J. 1992, Thâbit’s Astronomical Works, JHA, xxiii, p. 61
Ragep, F.J. 1993, Nasir al-Din al-Tusi’s Memoir on Astronomy, Springer Verlag New York (in two volumes)
Rashed, R. (sous la direction de). 1997, Histoire des sciences arabes, 1. Astronomie, théorie et appliquée.
Editions du Seuil, Paris
Rashed, R.; Vahabzadeh, B. 1999, Al-Khayyâm mathématicien, Librairie Scientifique et Technique Albert
Blanchard, Paris
Rosenfeld, B. 1974, Biruni, JHA, v, p. 134
Saliba, G. 1987, Theory and Observation in Islamic Astronomy: The Work of Ibn al-Shâter of Damascus, JHA,
xiii, p.35
Saliba, G. 1985, Solar Observations at the Maraghah Observatory before 1275: A New Set of Parameters, JHA,
xvi, p.113
Saliba, G. 1994, A Sixteenth-Century Arabic Critique of Ptolemaic Astronomy: The Work of Shams al-Din al-
Khafri, JHA, xxxv, p. 15
Saliba, G. 1997, Les théories planétaires en astronomie arabe après le XIe siècle, in Rashed 1997, p. 71
Sigismondi,, C.; Fraschetti, F. 2001, Measurement of the Solar Diameter in Kepler’s Time, The Observatory
121, 380
Stephenson, F.R.; Said, D.D. 1991, Precision of Medieval Islamic Eclipse Measurements, JHA, xxii, p. 197
Thurston, H. 1994, Early Astronomy, Springer Verlag
Veselovsky, I.N. 1973, Copernicus and Nasir al-Din al-Tusi, JHA, iv, p. 128
Vernet, J.; Samsó, J. 1997, Les développements de la science arabe en Andalousie, in Rashed 1997, p. 271
12
|
astro-ph/0109105 | 2 | 0109 | 2002-08-30T22:35:37 | Overproduction of primordial helium-4 in the presence of neutrino oscillations | [
"astro-ph",
"hep-ph"
] | The maximum overproduction of helium-4 in cosmological nucleosynthesis with active--sterile neutrino oscillations, nu_e -> nu_s, efficient after decoupling of electron neutrino, is analyzed. The kinetic effects on primordial nucleosynthesis due to neutrino spectrum distortion, caused by oscillations, are precisely taken into account.
The maximum overproduction of primordial He-4 as a function of oscillation parameters is obtained from the analysis of the kinetics of the nucleons and the oscillating neutrinos, for the full range of parameters of the discussed oscillation model. A maximum relative increase of He-4, up to 14% for non-resonant oscillations and up to 32% for resonant ones is registered. Cosmological constraints on oscillation parameters are also discussed. | astro-ph | astro-ph | CERN-TH/2002-217
Overproduction of primordial helium-4
in the presence of neutrino oscillations
D. P. Kirilova1
Theory Division, CERN, Geneva, Switzerland and
Institute of Astronomy, Bulgarian Academy of Sciences,
blvd. Tsarigradsko Shosse 72, Sofia, Bulgaria
Abstract
The maximum overproduction of helium-4 in cosmological nucle-
osynthesis with active -- sterile neutrino oscillations, νe ↔ νs, efficient
after decoupling of electron neutrino, is analyzed. The kinetic effects
on primordial nucleosynthesis due to neutrino spectrum distortion,
caused by oscillations, are precisely taken into account.
The maximum overproduction of primordial 4He as a function of oscil-
lation parameters is obtained from the analysis of the kinetics of the
nucleons and the oscillating neutrinos, for the full range of parame-
ters of the discussed oscillation model. A maximum relative increase
of 4He, up to 14% for non-resonant oscillations and up to 32% for
resonant ones is registered. Cosmological constraints on oscillation
parameters are also discussed.
2
0
0
2
g
u
A
0
3
2
v
5
0
1
9
0
1
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
1Regular Associate of the Abdus Salam ICTP, Trieste, Italy
1
1
Introduction
In this work I study the maximum overproduction of 4He in Big Bang Nucle-
osynthesis (BBN) with electron -- sterile neutrino oscillations νe ↔ νs.
The positive indications for oscillations, obtained by the neutrino exper-
iments (SuperKamiokande, SNO, Soudan 2, LSND, etc.) turned the subject
of neutrino oscillations into one of the hottest points of astrophysics and neu-
trino physics. The solar neutrino problem, the atmospheric neutrino anomaly
and the positive results of LSND experiments may be naturally resolved by
the phenomenon of neutrino oscillations, implying nonzero neutrino mass
and mixing. Although sterile neutrino impact in oscillations explaining at-
mospheric and solar neutrino anomalies, is strongly constrained from the
analysis of experimental oscillations data, still some small fraction of νs may
participate in oscillations. Hence, it is interesting to study cosmological ef-
fects of such oscillations.
Cosmological nucleosynthesis with neutrino oscillations was studied in
numerous publications, discussed in detail in [1, 2]. In these publications the
central goal was to obtain cosmological constraints on oscillation parameters.
The overproduction of 4He itself was not studied in detail until now. Such
a study may be of interest for constructing of alternative cosmological mod-
els, for constraining galactic chemical evolution, it may be useful for non-
standard models predicting active-sterile oscillations, like models with extra
dimensions, mirror world particles, etc.
Here I address that question of the possible maximal overproduction
of 4He due to oscillations. For that purpose the case of νe ↔ νs, oscilla-
tions, effective after the electron neutrino decoupling from the plasma (i.e
for δm2 sin2 2ϑ ≤ 10−7eV2), is the most suitable. In that case due to the fact
that sterile neutrino state is usually less populated than the active one at
the start of oscillations, the oscillations may cause strong distortion of the
electron neutrino spectrum, which affects the kinetics of the nucleons freezing
before nucleosynthesis, and hence, the primordial production of 4He. And,
as will be shown by the numerical analysis, this kinetic effect of oscillations,
may be much bigger than the one corresponding to an increase in the total
energy density due to an additional neutrino flavor δNρ, mainly considered
in literature.
We analyze 4He primordial production, taking into account all known ef-
fects of νe ↔ νs oscillations on the primordial synthesis of 4He. The produc-
2
tion of 4He was calculated in the non-resonant and resonant oscillation cases.
In both cases strong overproduction of 4He was found possible - up to 14%
and 32%, correspondingly. Thus in the discussed model of non-equilibrium
oscillations the maximum overproduction of 4He corresponds to an increase
of the neutrino effective degrees of freedom δN max
The oscillation effects on BBN and their description are briefly reviewed
in the next section. The kinetic approach, the results on 4He primordially
produced abundance in the presence of oscillations, and the cosmological
constraints on oscillation parameters are discussed in the last section.
kin ∼ 6.
2 Cosmological Nucleosynthesis with Neutrino
Oscillations
2.1 Standard Cosmological Nucleosynthesis
According to the standard BBN, during the early hot and dense epoch of the
Universe, the light elements D, 3He, 4He, 7Li were synthesized successfully.
The most reliable and abundant data are now available for 4He. This fact
and its relatively simple chemical evolution make 4He the preferred element
for the analysis of the oscillations effect on BBN.
The contemporary values for the mass fraction of 4He, Yp, inferred from
observational data, are in the range 0.238 -- 0.245 (the systematic errors are
supposed to be around 0.007) [3].
Primordially produced 4He abundance Yp, is calculated with great preci-
sion within the standard BBN model [4]. According to the standard BBN
4He is a result of a complex network of nuclear reactions, proceeding after
the neutron-to-protons freezing. It essentially depends on the freezing ratio
(n/p)f . The latter is a result of the freezing of the weak processes:
νe + n ↔ p + e−
e+ + n ↔ p + νe,
(1)
which maintained the equilibrium of nucleons at high temperature (T > 1
MeV). Their freeze-out occurred when in the process of Universe cooling
the rates of these weak processes, Γw, became comparable and less than the
expansion rate H(t):
3
Γw ∼ G2
F E 2
ν Nν ≤ H(t) ∼ √gef f T 2
Thus, the produced 4He is a strong function of the number of the effective
degrees of freedom at BBN epoch, gef f = 10.75 + 7/4δNρ.
Yp depends also on the electron neutrino spectrum and number density,
In
and on the neutrino-antineutrino asymmetry, which enter through Γw.
the standard BBN model three neutrino flavors, zero lepton asymmetry and
equilibrium neutrino number densities and spectrum distribution are postu-
lated:
nνe(E) = (1 + exp(E/T ))−1.
Almost all neutrons, present at the beginning of nuclear reactions, are
sucked into 4He. So, the primordially produced mass fraction of 4He can be
approximated by
Yp ∼ 2(n/p)f /(1 + n/p)f exp(−t/τn).
where τn is the neutron mean lifetime.
The theoretical uncertainty of the precisely calculated Yp is less than 0.1%
(δYp < 0.0002) within a wide range of η. So, in the standard BBN scenario,
where τn, and gef f are fixed, Yp, as well as the rest cosmologically produced
light elements are functions of only one parameter - the baryon-to-photon
ratio η.
Deuterium measurements in pristine environments towards low metal-
licity quasar absorption systems at very high z ∼ 3 provide us with the
most precision determination of the baryon density [5], giving the value:
η = (5.6 ± 0.5) × 10−10. Recently, the baryon density was also determined
from observations of the anisotropy of the cosmic microwave background
(CMB) by DASI [6], BOOMERANG [7, 8], CMB [9] and MAXIMA exper-
iments [10]. For the combined analysis of the data see also [11, 12]. The
CMB anisotropy data is in remarkable agreement with the baryon density
determined from deuterium measurements and BBN.
So, the predicted primordial 4He abundance Yp is in accordance with the
observational data and is consistent with deuterium measurements.
4
2.2 Effects of neutrino oscillations on nucleosynthesis
In case neutrino oscillations are present in the Universe primordial plasma,
they lead to changes in the Big Bang Nucleosynthesis, depending on the
oscillation channels and the way they proceed. The effect of flavor neutrino
oscillations on BBN is negligibly small because the temperatures and hence,
the densities of the neutrinos with different flavors are almost equal.
Active-sterile oscillations, however, are capable to shift neutrino num-
ber densities and spectrum from their equilibrium values. Besides, they
may change neutrino-antineutrino asymmetry and excite additional neutrino
types. Thus, the presence of neutrino oscillations invalidates the main BBN
assumptions about three neutrino flavors, zero lepton asymmetry and equi-
librium neutrino energy distribution.
Qualitatively, the oscillations effects on the nucleosynthesis of 4He may
be illustrated as follows:
• excitation of additional degrees of freedom
It is known that active-sterile oscillations may keep sterile neutrinos in ther-
mal equilibrium [13] or bring them into equilibrium in case they have already
decoupled. The presence of light steriles in equilibrium leads to an increase
of the effective degrees of freedom during BBN and to faster Universe ex-
pansion H(t) ∼ g1/2
ef f and earlier n/p-freezing, Tf ∼ (gef f )1/6, at times when
neutrons were more abundant [13, 14]:
n/p ∼ exp [−(mn − mp)/Tf ]
This effect leads to 5% 4He overproduction corresponding to one additional
neutrino type brought into equilibrium by oscillations.
• distortion of the neutrino spectrum
Much stronger effect of oscillations may be achieved in case of oscillations
between initially empty sterile neutrino state and electron neutrino. The
non-equilibrium initial condition leads to spectrum distortion of the active
neutrino due to oscillations.
Since oscillation rate is energy dependent Γ ∼ δm2/E the low energy
neutrinos start to oscillate first, and later the oscillations become noticeable
5
for the more energetic neutrinos. Due to that, the neutrino energy spectrum
nν(E) may strongly deviate from its equilibrium form [15]. This spectrum
distortion affects the kinetics of nucleons freezing - it leads to an earlier
n/p-freezing and an overproduction of 4He yield.
The effect can be easily understood having in mind that the distortion
leads both to a depletion of the active neutrino number densities Nν [16]:
Nν ∼ Z dEE 2nν(E)
and to a decrease of the mean neutrino energy, which reflects into a decrease
ν Nν, and hence, to an overproduction of 4He pri-
of the weak rates Γw ∼ E 2
mordial abundance.
The decrease of the electron neutrino energy due to oscillations into
low temperature sterile neutrinos, has also an additional effect: Due to the
threshold of the reaction converting protons into neutrons, when neutrinos
have lower energy, protons are preferably produced in reactions (1), which
may lead to an underproduction of 4He [17]. However, this turns to be a
minor effect.
The effect of spectrum distortion was numerically analyzed for hundreds
combinations of mass differences and mixing angles in previous studies of
active-sterile oscillations [15, 18, 19, 20]. It was proved important both in
the resonant [19] and in the non-resonant oscillations case [20].
In the discussed here scenario it is the dominant effect and leads to over-
production of the primordial 4He abundance.
• neutrino-antineutrino asymmetry growth
The idea of neutrino-antineutrino asymmetry generation during the resonant
transfer of neutrinos was first proposed in ref. [21]. Dynamically produced
asymmetry exerts back effect to oscillating neutrino and may change its oscil-
lation pattern [22, 23], it may suppresses oscillations at small mixing angles,
leading to weakening the oscillations effect on BBN, i.e. less overproduction
of
4He. 2
The effect of the oscillations generated asymmetry on 4He for the discussed
here model was analyzed for hundreds of δm2 − ϑ combinations in [19].
2There were independent studies of the asymmetry growth in active-sterile oscillations
for the case of large mass differences along the lines of the pioneer work [24].
6
3 Helium-4 overproduction due to νe ↔ νs
neutrino oscillations
We have provided an exact study of all the oscillation effects on the primordial
production of 4He.
3.1 The required kinetic approach
For the analysis of the non-equilibrium picture of active -- sterile neutrino os-
cillations, producing non-equilibrium neutrino number densities, distorting
neutrino spectrum and generating neutrino-antineutrino asymmetry a self-
consistent numerical analysis of the evolution of the nucleons and the oscil-
lating neutrinos in the high temperature Universe was provided.
Exact kinetic equations for the nucleons and for the neutrino density ma-
trix in momentum space [18] were used. This allowed to describe precisely the
kinetic effects of oscillations on helium production due to spectrum distortion
at each neutrino momentum.
The equation for the neutron number densities in momentum space nn
reads:
(∂nn/∂t) = Hpn (∂nn/∂pn) +
+Z dΩ(e−, p, ν)A(e−p → νn)2 [ne−np(1 − ρLL)−
−nnρLL(1 − ne−)]
−Z dΩ(e+, p, ν)A(e+n → pν)2 [ne+nn(1 − ¯ρLL)−
−np ¯ρLL(1 − ne+)] .
(2)
where dΩ(i, j, k) is a phase space factor and A is the amplitude of the cor-
responding process, neutrino ρLL and antineutrino number densities ¯ρLL at
each integration step of eq. (2) are taken from the simultaneously performed
integration of the set of equations for neutrino density matrix (see ref. [18]).
The equation provides a simultaneous account of the different compet-
ing processes, namely: neutrino oscillations (entering through ρLL and ¯ρLL),
Hubble expansion (first term) and weak interaction processes (next terms).
The numerical analysis was performed for the temperature interval [2 MeV,
0.3 MeV].
7
We have found that for a wide range of oscillation parameters the spec-
trum distortion of electron neutrino is considerable during the period of nu-
cleons freezing. Hence, usually the kinetic effect of oscillations due to electron
neutrino energy spectrum distortion plays the dominant role in the overpro-
duction of helium.
In Fig. 1 the evolution of the energy spectrum of the
electron neutrino through the period of nucleons freezing is illustrated. In
the Figs.1a,1b and 1c the energy spectra at different characteristic tempera-
tures, correspondingly T = 1 MeV, T = 0.7 MeV and T = 0.5 are presented.
By the dashed curve the equilibrium spectrum at the given temperature is
given for comparison. The spectra are calculated for oscillation parameters
δm2 = 10−7 eV2 and sin2 2ϑ = 0.1.
It is seen that oscillations affect non trivially the neutrino spectrum and
distort strongly its equilibrium form. So, it is not possible to describe cor-
rectly the spectrum distortion due to oscillations simply by shifting the effec-
tive temperature of the neutrino and accounting only for the depletion of its
total number density, considering its spectrum equilibrium as in refs. [26, 27].
For the proper description of the spectrum distortion in the case of non-
equilibrium electron-sterile oscillations, studied here, the evolution of the
neutrino ensembles should be explored using neutrino density matrix in mo-
mentum space, allowing to describe the evolution of neutrino at each mo-
mentum.
The neutrino-antineutrino asymmetry in the resonant case grows up to
5 orders of magnitude from its initial value (taken to be of the order of the
baryon asymmetry). So, this asymmetry influences BBN only indirectly -
through oscillations. This oscillations generated asymmetry leads to a de-
crease in the produced 4He compared with the case without asymmetry ac-
count. However, its effect comprises only up to about a 10% of the total
effect.
3.2 Maximum helium-4 overproduction
The overproduction of the primordial 4He, δYp = Y osc
p − Yp in the presence of
νe ↔ νs oscillations was calculated for mass differences δm2 ≤ 10−7 eV2 and
0 ≤ ϑ ≤ π/4. Several hundreds of δm2 − ϑ combinations were explored.
We have used the data from the precise calculations of the n/p-freezing
provided for the non-resonant case in [20] and for the resonant case in [19].
As far as it is the essential for the production of 4He. The neutron decay was
8
accounted adiabatically till the beginning of nuclear reactions at about 0.09
MeV.
We have found that the effect of oscillations becomes very small (less
than 1%) for small mixings: as small as sin22ϑ = 0.1 for δm2 = 10−7 eV2,
and for small mass differences: δm2 < 10−10 eV2 at maximal mixing. For
very small mass differences δm2 ≤ 10−11 eV2, or at very small mixing angles
sin2 2ϑ ≤ 10−3, the effect on nucleosynthesis becomes negligible.
In the non-resonant case the oscillation effect increases with the increase
of the oscillation parameters, hence it is maximal at maximal mixing and
greatest mass differences. In Fig. 2 (the lower curve) the maximal relative
increase in the primordial 4He as a function of neutrino mass differences at
maximal mixing: δY max
p /Yp(δm2)θ=π/4 is presented.
/Yp = δY osc
p
It is seen that for maximal mixing, the oscillation effect becomes greater
than 5% (the one corresponding to one additional neutrino type) already at
δm2 ≥ 3 × 10−9 eV2 (in the resonant case) and δm2 ≥ 6 × 10−9 eV2 (in
the non-resonant one). It continues to grow up with the increase of the mass
differences, and at δm2 ∼ 10−7 eV2 is several times bigger: δN max
kin ∼ 3 in
the non-resonant case and δN max
Further increase of the mass differences, however, will lead to oscillations
effective before electron neutrino decoupling, and therefore, to a smaller spec-
trum distortion effect, because the interactions with the plasma will lead to
faster thermalization of the sterile state. Hence, the effect on helium will
decrease with further increase of δm2 and finally reach an overproduction
of 5% again, corresponding to a full thermalization of the sterile state and
its equilibrium spectrum.
kin ∼ 6 for the resonant one.
In Fig. 3 we present a combined plot (for the resonant and the non-
resonant oscillation case) of δYp dependence on the mixing angle for δm2 =
10−7 eV2 and δm2 = 10−8 eV2. While in the non-resonant case the oscillations
effect increases with the increase of the mixing (see l.h.s of Fig.3.), in the
resonant case for a given δm2 there exists some resonant mixing angle, at
which the oscillations are enhanced by the medium (due to the MSW effect),
and hence, the overproduction of 4He is greater than that corresponding to
the vacuum maximal mixing angle. This behavior of the helium production
on the mixing angle is illustrated in the r.h.s. of the figure.
The upper curve in Fig.2 shows the maximal relative increase in the
resonant oscillations case as a function of mass differences. Each maximum
4He value corresponds to the resonant mixing angle for the concrete mass
9
(δm2, ϑres
p
p
/Yp = Y osc
difference: δY max
δm). As can be seen from Figs. 2 and 3,
a considerable overproduction can be achieved:
in the resonant case up to
32% and in the non-resonant one -- up to 14%. So, the net effect of spectrum
distortion of electron neutrino due to oscillations on the production of 4He
may be considerable and several times larger than the effect due to excitation
of one additional neutrino type.
Several words are due to cosmological constraints on neutrino oscillation
parameters, calculated in the discussed model of oscillations. Due to the
strong kinetic effect of neutrino spectrum distortion caused by active-sterile
oscillations discussed, the obtained constraints in that model of oscillations
are more stringent than those obtained in pioneer works, not accounting for
the kinetic effects of oscillations. Hence, according to these more precise
studies of spectrum distortion effects of oscillations, the δYp/Yp < 3% limit
excludes almost completely the LOW electron-sterile solution to the solar
neutrino problem [25], in addition to the excluded sterile LMA solution in
previous investigations.
it is more than an order of magnitude more
restrictive to the mass differences.
I.e.
This study has shown that considerable Yp overproduction may result
from the electron neutrino spectrum distortion due to νe ↔ νs oscillations.
The overproduction is maximal for the case of initially empty sterile neu-
trino state, considered here. The dependence of 4He overproduction on the
degree of population of the sterile neutrino state before νe ↔ νs oscillations
is considered in a separate paper [28]. Bigger initial population of νs leads to
a smaller spectrum distortion of νe, and hence smaller kinetic effect on pri-
mordial nucleosynthesis δNkin and higher energy density due to the increase
of the number of degrees of freedom δNρ. The interplay between the two
effects, however, is such that the overproduction of 4He is smaller than in the
case when initially δNρ = 0, discussed here.
4 Conclusions
The primordial production of 4He in the presence of νe ↔ νs oscillations, effec-
tive after electron neutrino decoupling was analyzed. A precise quantitative
study of the maximum overproduction of 4He, accounting for all oscillations
effects is provided. It was shown that the considerable spectrum distortion
10
of the electron neutrino caused by oscillations, which effects the kinetics of
the neutron-proton transitions during nucleons freezing, plays the dominant
role in the overproduction of 4He.
Enormous overproduction of 4He (up to 32% in the resonant case and
14% in the non-resonant case) was found possible in case the sterile neutrino
state was empty at the start of oscillations.
The results of this analysis can be useful for constraining nonstandard
physics, predicting active-sterile neutrino oscillations, like extra-dimensions,
producing oscillations, supernova bursts employing oscillations, etc.. It can
be of interest also for models of galactic chemical evolution.
Acknowledgments
I thank M. V. Chizhov for the help during the preparation of this paper.
This work has been completed during my visiting position at TH CERN. I
appreciate also the Regular Associateship at the Abdus Salam ICTP, Trieste.
References
[1] Kirilova, D. P., and Chizhov, M. V., 2001, CERN-TH/2001-020, astro-
ph/0108341
[2] Dolgov A., hep-ph/0202122.
[3] Izotov, Y. I., and Thuan, T. X., Ap. J. 500 (1998) 188.
[4] Lopez, R., and Turner, M. S., Phys. Rev. D 59 (1999) 103502.
[5] Meara J. et al., Ap.J. 552 (2001) 718.
[6] Pryke S. et al., ApJ 5681 (2002) 46.
[7] Netterfield C. et al., ApJ 571 (2002) 604.
[8] Bernardis P. et al., ApJ 564 (2002) 559.
[9] Padin S. et al., ApJ Letters 549 (2001) L1.
[10] Stompor B. et al., ApJ 561 (2001) L7.
11
[11] Wang X., Tegmark M., Zaldarriaga M., Phys.Rev. D65 (2002) 123001
[12] Douspis M. et al., Astr.Astrop. 379 (2001) 1.
[13] Dolgov, A. D., Sov. J. Nucl. Phys. 33 (1981) 700.
[14] Fargion, D., and Shepkin, M., Phys. Lett. B 146 (1984) 46.
[15] Kirilova, D. P., 1988, JINR E2-88-301, Dubna, Russia.
[16] Barbieri, R., and Dolgov, A., Phys. Lett. B 237 (1990) 440.
[17] Dolgov A.,Kirilova D., Int.J.Mod.Phys.A3 (1988) 267.
[18] Kirilova, D. P., and Chizhov, M. V., Phys. Lett. B 393 (1997) 375.
[19] Kirilova, D. P., and Chizhov, M. V., Nucl. Phys. B 591 (2000) 457.
[20] Kirilova, D. P., and Chizhov, M. V., Phys. Rev. D 58 (1998) 073004.
[21] Mikheyev, S., and Smirnov, A., 1986, in VI Moriond Meeting on Massive
Neutrinos in Particle Physics and Astrophysics, eds. O. Fackler and J.
Tran Thanh Van, Editions Frontiers, Tignes, p. 355.
[22] Kirilova, D., and Chizhov, M.,1996, in 17 International Conference on
Neutrino Physics and Astrophysics, NEUTRINO 96, eds. K. Enqvist,
K. Huitu, and J. Maalampi, World Scientific, Helsinki, p. 478.
[23] Kirilova, D. P., and Chizhov, M. V., 1999, hep-ph/9908525.
[24] R. Foot, M. J. Thomson and R. R. Volkas, Phys. Rev. D 53 (1996)
R5349
[25] Kirilova, D. P., and Chizhov, M. V., Nucl. Phys. B Proc. Suppl. 100
(2001) 360.
[26] Shi X., Schramm D. N., and Fields B. D., Phys. Rev. D 48, 2563 (1993).
[27] Shi X., and Fuller G. M., Phys. Rev. D 59, 063006 (1999).
[28] Kirilova D., CERN-TH-2002-209
12
Figure 1a: The figure illustrates the degree of distortion of the electron
neutrino energy spectrum x2ρLL(x), where x = E/T , caused by oscillations
with mass difference δm2 = 10−7 eV2 and sin2 2ϑ = 0.1 at a character-
istic temperature 1 MeV. The dashed curve gives the equilibrium neutrino
spectrum for comparison.
13
Figure 1b: The figure illustrates the degree of distortion of the electron
neutrino energy spectrum x2ρLL(x), where x = E/T , caused by oscillations
with mass difference δm2 = 10−7 eV2 and sin2 2ϑ = 0.1 at a characteristic
temperature 0.7 MeV.
14
Figure 1c: The figure illustrates the degree of spectrum distortion of
the electron neutrino caused by oscillations with mass difference δm2 = 10−7
eV2 and sin2 2ϑ = 0.1 at a characteristic temperature 0.5 MeV.
15
Figure 2: Maximum primordial 4He abundance for the resonant (upper
curve) and the non-resonant oscillation case (lower curve), as a function of the
neutrino mass differences. The non-resonant case is calculated at maximum
mixing, while in the resonant case the helium abundance is calculated at the
resonant mixing angle for the corresponding mass difference.
16
Figure 3: The dependence of the relative increase of primordial helium
on the mixing angle for the resonant (r.h.s.) and non-resonant (l.h.s.) oscil-
lation case. The upper curve corresponds to δm2 = 10−7 eV2, the lower one
to δm2 = 10−8 eV2.
17
|
astro-ph/0204159 | 1 | 0204 | 2002-04-09T15:52:48 | Is RX J185635-375 a Quark Star? | [
"astro-ph"
] | Deep Chandra LETG+HRC-S observations of the isolated neutron star candidate RX J1856.5-3754 have been analysed to search for metallic and resonance cyclotron spectral features and for pulsation behaviour. As found from earlier observations, the X-ray spectrum is well-represented by a ~ 60 eV (7e5 K) blackbody. No unequivocal evidence of spectral line or edge features has been found, arguing against metal-dominated models. The data contain no evidence for pulsation and we place a 99% confidence upper limit of 2.7% on the unaccelerated pulse fraction over a wide frequency range from 1e-4 to 100 Hz. We argue that the derived interstellar medium neutral hydrogen column density of 8e19 <= N_H <= 1.1e20 per sq. cm favours the larger distance from two recent HST parallax analyses, placing RX J1856.5-3754 at ~ 140 pc instead of ~ 60 pc, and in the outskirts of the R CrA dark molecular cloud. That such a comparatively rare region of high ISM density is precisely where an isolated neutron star re-heated by accretion of interstellar matter would be expected is either entirely coincidental, or current theoretical arguments excluding this scenario for RX J1856.5-3754 are premature. Taken at face value, the combined observational evidence -- a lack of spectral and temporal features and an implied radius at infinity of 3.8-8.2 km that is too small for current neutron star models -- points to a more compact object, such as allowed for quark matter equations of state. | astro-ph | astro-ph |
To appear in the Astrophysical Journal
Is RXJ1856.5-3754 a Quark Star?
Jeremy J. Drake1 , Herman L. Marshall2 , Stefan Dreizler3 , Peter E. Freeman1 ,
Antonella Fruscione1 , Michael Juda1 , Vinay Kashyap1 , Fabrizio Nicastro1 , Deron
O. Pease1 , Bradford J. Wargelin1 , Klaus Werner3
1Smithsonian Astrophysical Observatory, MS-3, 60 Garden Street, Cambridge, MA 02138
2MIT Center for Space Research, Cambridge, MA 02139
3 Institut fur Astronomie und Astrophysik, Universitat Tubingen, Sand 1, 72076 Tubingen,
Germany
ABSTRACT
Deep Chandra LETG+HRC-S observations of the isolated neutron star can-
didate RX J1856.5−3754 have been analysed to search for metallic and resonance
cyclotron spectral features and for pulsation behaviour. As found from earlier
observations, the X-ray spectrum is well-represented by a ∼ 60 eV (7 × 105 K)
blackbody. No unequivocal evidence of spectral line or edge features has been
found, arguing against metal-dominated models. The data contain no evidence
for pulsation and we place a 99% confidence upper limit of 2.7% on the unac-
celerated pulse fraction over a wide frequency range from 10−4 to 100 Hz. We
argue that the derived interstellar medium neutral hydrogen column density of
8 × 1019 ≤ NH ≤ 1.1 × 1020 cm−2 favours the larger distance from two recent
HST parallax analyses, placing RX J1856.5−3754 at ∼ 140 pc instead of ∼ 60 pc,
and in the outskirts of the R CrA dark molecular cloud. That such a compar-
atively rare region of high ISM density is precisely where an isolated neutron
star re-heated by accretion of interstellar matter would be expected is either en-
tirely coincidental, or current theoretical arguments excluding this scenario for
RX J1856.5−3754 are premature. Taken at face value, the combined observa-
tional evidence—a lack of spectral and temporal features, and an implied radius
R∞ = 3.8-8.2 km that is too small for current neutron star models—points to a
more compact ob ject, such as allowed for quark matter equations of state.
Subject headings: stars:
X-rays: stars
individual (RX J1856.5−3754) — stars: neutron —
– 2 –
1.
Introduction
The structure and evolution of neutron stars depends on the properties of matter at
nuclear and supranuclear densities. Such conditions are not achievable in terrestrial labora-
tories and the theoretical description of such superdense matter remains uncertain. It has
therefore been hoped that neutron stars, and in particular those that are isolated and not
complicated by strong accretion or magnetospheric signatures, might provide some empirical
insights: observations of their masses, radii and cooling characteristics could, in principle,
provide useful constraints for the equation of state (EOS) of dense matter (e.g. Lattimer &
Prakash 2001 and references therein).
In a relatively brief period of 106 -107 yr, a hot, isolated neutron star (INS) born in a
supernova explosion can cool, cease pulsar activity and become essentially inactive (see, e.g.,
the review of Treves et al. 2000). Of the estimated 108 -109 isolated neutron stars thought
to inhabit the Galaxy, only a tiny fraction are therefore expected to be sufficiently young
to remain hot and visible in X-rays. One possible mechanism capable of sustaining thermal
X-ray emission in an older INS is accretion of material from the ISM. To date, only a handful
of these older INS candidates have been found.
The soft X-ray source RX J1856.5−3754 discovered by Walter, Wolk & Neuhauser (1996)
is the brightest, and probably the closest (Kaplan, van Kerkwijk & Anderson 2002) of the
INS candidates. It was identified with a very faint (V ≃ 25.6) optical counterpart by Walter
& Matthews (1997), was found to have an optical flux about a factor of 2-3 higher than that
predicted by the Rayleigh-Jeans tail of the ∼ 55 eV blackbody spectrum that represents the
low resolution ROSAT PSPC spectrum (e.g. Walter et al. 1996; Campana et al. 1997, Pons et
al. 2001), and lies in the line-of-sight toward the dark molecular cloud R CrA. However, the
exact nature of RX J1856.5−3754 remains unknown—whether it is a fairly young, cooling
INS, perhaps undetected as a pulsar because of unfortunate beam alignment, or an older
ob ject reheated by ISM accretion.
Very Large Telescope (VLT) observations have recently revealed an Hα nebula round
RX J1856.5−3754 and a blackbody spectrum through the UV-optical range (van Kerkwijk
& Kulkarni 2001a,b). Hubble Space Telescope (HST) astrometry was used by Walter (2001)
and Kaplan et al. (2002) to estimate a parallax and proper motion, but with conflicting
results (16.5 ± 2.3 mas vs. 7 ± 2 mas). Walter (2001) argued that the proper motion points
to the Sco-Cen OB association and an age of ∼ 106 yr. However, Pons et al. (2001) failed
to detect the expected pulsation signature in ROSAT and ASCA data. Modelling X-ray,
EUV, UV and optical spectra using the Walter (2001) parallax and assuming a metal-
dominated atmosphere resulted in stellar radii too small for current EOS and smaller than
the Schwarzschild radius for a canonical 1.4M⊙ star, leading them to conclude that the
– 3 –
surface temperature distribution could be inhomogeneous. In contrast, the same atmospheric
analysis using the larger distance of Kaplan et al. (2002) would yield a radius consistent with
current theory. If the atmosphere is indeed metal-dominated, then line and edge features
should be visible in high resolution X-ray spectra.
RX J1856.5−3754 was observed in 2000 March for 55ks by the Chandra X-ray Observa-
tory Low Energy Transmission Grating (LETG) and the High Resolution Camera spectro-
scopic microchannel plate detector array (HRC-S) to look for spectral features and pulsar
activity. Within relatively large statistical uncertainties, Burwitz et al. (2001) failed to de-
tect either significant departures from a blackbody spectrum or pulsations in these data.
The prospect of important scientific gains from a longer observation prompted, at the re-
quest of different researchers, the undertaking of a very recent set of Chandra observations of
RX J1856.5−3754 under director’s discretionary time using LETG+HRC-S. A period search
using these data by Ransom, Gaensler & Slane (2002) has already placed an upper limit on
a pulse fraction of 4.5%. This Letter presents spectroscopic and independent timing analyses
of these data.
2. Observations and Data Reduction
The observations analysed in this paper Pipeline-processed (CXC software version 6.3.1)
photon event lists were reduced and analysed using the CIAO software package version 2.2,
and independently using custom IDL1 software. Processing included extra pulse-height fil-
tering to reduce background (Wargelin et al., in preparation) and barycentric correction of
event times. Dispersed photon events were extracted using the now standard “bow tie” win-
dow, and a small circular region (31-pixel radius2 ) was used to extract the 0th order events.
We note that an extraneous bright feature (likely a ghost image of a bright off-axis source)
appeared on the +1 order outer HRC-S plate which introduces spurious features into the
extracted spectrum and background near 110 A. We applied corrections to extracted event
times to ameliorate an HRC electronics problem that assigns time tags for each event to the
event that immediately follows. If every event were telemetered to the ground, correct times
could be easily reassigned, but because of the high HRC background rate and a telemetry
limit of 184 events/sec, only valid events, which make up approximately 45% of the total
1 Interactive Data Language, Research Systems Inc.
2Our 31 pixel radius extraction region for the 0th order source corresponds to an encircled energy fraction
of about (92 ± 3)%, which combines with an average deadtime of 0.59% to yield a correction factor for a 2π
steradian aperture of 1.09.
– 4 –
events in these observations, are recorded in data received on the ground. We have reas-
signed the time of every valid event to the preceding valid event during ground processing;
the event times will then be correct about Nvalid /Ntotal of the time. One can place an upper
limit of δ t on the time errors associated with this process, however, simply by excluding
events with a time-shift of more than δ t; the average timing error of such events will, of
course, be less than δ t. However, if δ t is too small, too many events get excluded to perform
sensitive timing studies. We have adopted a δ t of 2 ms, retaining 20% of the original counts
in the corrected data set.
3. Count Rates and Timing Analysis
Count rates were derived for each observation segment from event lists filtered to exclude
times of high background and telemetry saturation, when the telemetered valid event rate
exceeded 184 count s−1 , and to exclude high pulse-height events that arise entirely from
background. These rates corresponding to 0th order only are listed in Table 1.
The count rates for the different observation segments are statistically consistent, though
the 2000 March observation (113) rate of 0.2195 ± 0.0020 count s−1 lies 1.9% and 1.9σ above
the rate for the combined 2001 October series rate of 0.2155 ± 0.0008. Such deviations will be
obtained by chance about 3% of the time when the count rates do not vary; it is thus more
likely attributable to quantum efficiency (QE) variations in the detector on small scales.
Indeed, Pons et al. (2001) note two ROSAT HRI observations obtained 3 yr apart that are
consistent to 1%. The fluxes corresponding to HRI and PSPC count rates are higher than
that obtained from our Chandra data by 20% and 30%, respectively. As remarked by Burwitz
et al. (2001), these differences are likely attributable to absolute calibration uncertainties.
We have also examined the ASCA SIS observation described Pons et al. (2001) and find
fluxes 30-40 % lower than obtained by Chandra for the 20-30 A range, but in agreement
with Chandra shortward of 20 A. The EUVE DS count rate of Pons et al. (2001) is also
consistent with the Chandra observation within allowed uncertainties.
Our search for pulsations used three different techniques, none of which found any
evidence for significant variability. In contrast to Ransom et al. (2002), we did not include
a deceleration term; this will be discussed below.
We applied the Bayesian method of Gregory & Loredo (1992) to both the time-corrected
and uncorrected 0th order event lists. The method tests for variability by comparing the fits
of periodic stepwise models to the data with the fit of a constant model. The odds favoring
variability (based on eqn. 5.28 of Gregory & Loredo 1992, with a maximum number of steps
– 5 –
mmax = 12 and limiting angular frequencies ωlo and ωhi equal to 10−4 and 103 , respectively)
were found to be 1.45 × 10−4 for the whole data set and 3.75 × 10−3 for the data filtered on
δ t = 2 ms, both to be compared with an odds value of 102 needed for a confident pulsation
or variability detection.
Our second method employed an FFT analysis applied to the combined 0th and 1st
order events, followed by a likelihood ratio test (LRT) to determine limits on the pulse
fraction. The FFT power distribution was consistent with shot noise. For the LRT of a
given period, P , the data were binned into N phase bins, giving ni counts in each. The
source model was y = A + f cos(φi + φ0 ) where φi is the phase of bin i and φ0 is the phase
of the pulse and f /A is the pulse fraction. The likelihood equations were then solved for A
and the process applied to > 500 frequencies where the FFT power exceeded a critical level
in the frequency range 0.001-50 Hz. By including also the dispersed events we improved the
signal-to-noise of the result by a factor of 1.47 compared to using 0th order alone and could
obtain a pulse fraction limit lower than the value of 4.5% obtained by Ransom et al. (2002)
in their unaccelerated search. A pulse fraction upper limit (99% confidence) of 2.7% was
derived applying our likelihood ratio method using all data, including the dispersed events
with 1 < λ < 70 A. Taking only the events limited by δ t < 2 ms, the pulse fraction limit is
10%.
Thirdly, we computed the Lomb-Scargle periodogram for both the time-corrected and
uncorrected photon arrival time differences in the frequency range 0.01-1 Hz for the events
in ObsID’s 3380 3381, 3382 and 3399. Again, no significant peaks were present.
The assumption of a negligible deceleration term in our period search restricts the range
of periods and dipole magnetic field strengths for which our search is valid. The coherence
limit for phase slippage by 10 % over the duration of the 2001 October observations implies
that our result is valid for a magnetic field upper limit B < 2.3 × 1013P 3/2 G for period
P s (e.g. Shapiro & Teukolsky 1983). As noted by Ransom et al. (2002), this range would
exclude very young and energetic neutron stars, such as the Crab and Vela pulsars, though
most of these younger ob jects are also conspicuously strong radio pulsars. All anomalous
X-ray pulsars would lie within our sensitivity limit range. RX J1856.5−3754 is also most
unlikely to be an extremely young ob ject based on its modest temperature and luminosity,
which are consistent with an ob ject of age ∼ 105 yr on canonical cooling curves (e.g. Tsuruta
1997).
– 6 –
4. Spectral Analysis and Model Parameter Estimation
Spectral analysis in the form of model parameter estimation was undertaken using the
CIAO/Sherpa fitting engine and independently using specially-written IDL software. Cur-
sory inspection of the spectrum leads immediately to the conclusion that there are no obvious
features indicative of absorption lines or edges. We found that blackbody models represent
the high resolution spectra well, in agreement with earlier studies. We modelled +1 and −1
orders both separately and simultaneously, and added together; results from these differ-
ent approaches were statistically indistinguishable. Representative results of model fits and
residuals are illustrated in Figure 1. Two sets of best-fit parameters were obtained from in-
dependent analyses that invoked (1) the existing first and higher order CXC calibration3 and
(2) the same first order effective area with higher orders modified slightly to improve model
fits to sources with power law spectra (Marshall et al., in preparation). Both sets are con-
sistent with the results of Burwitz et al. (2001) based on the 2000 March observation alone.
Parameters and 1σ statistical uncertainties for best-fit models were: (1) T = 61.2 ± 0.3 eV;
NH = (1.10 ± 0.02) × 1020 cm−2 ; X-ray luminosity (2.96 ± 0.03) × 1031D2
100 erg s−1 , where
D100 is the distance in units of 100 pc; (2) T = 61.1 ± 0.3 eV; NH = (0.81 ± 0.02) × 1020 cm−2 ;
X-ray luminosity (3.16 ± 0.03) × 1031D2
100 erg s−1 . Parameters producing minima in the χ2
test statistic were not sensitive to the exact binning adopted, though of course the reduced χ2
values were: values ranged from 0.94 for data binned to a signal-to-noise ratio of S/N = 10,
to 1.7 for S/N = 30. The latter value is dominated by residual effective area calibration
uncertainties. To investigate the effects of these uncertainties, which are estimated to be
about 15% absolutely, a first order polynomial term was included in the source model to
mimic an effective area lower by 15% at 20A and higher by 15% at 100 A, and vice-versa.
Such a slope skews the blackbody curve and leads systematically to higher or lower temper-
ature solutions by about 1 eV—clearly the true temperature uncertainty is driven by this
uncertainty in the effective area. Allowing for this uncertainty, we adopt a final tempera-
ture of 61.2 ± 1.0—in agreement with, but much more tightly constrained than, the ROSAT
temperatures derived by Burwitz et al. (2001; 63 ± 3 eV) and Pons et al. (2002; 55.3 ± 5.5).
We note that a temperature of 61 eV results in an optical flux only 10% higher than that
from a 55 eV blackbody, so that the discrepency between the observed optical flux and that
predicted by the hot blackbody noted by Walter & Matthews (1997) and Pons et al. (2002)
remains essentially unchanged.
Blackbody models were found by both Pons et al. (2001) and Burwitz et al. (2001) to
represent observed spectra better than sophisticated model atmospheres, though a uniform
3Version dated 2000 October 31; http://asc.harvard.edu/cal/Links/Letg/User
– 7 –
temperature blackbody model was formally excluded in X-ray-EUV-UV-optical modelling
in the former work. However, additional cooler components were found to contribute at
most only a few percent to the observed ROSAT PSPC X-ray flux. We have ruled out
the significant presence of additional thermal and non-thermal emission components by trial
of models combining two blackbodies, and a blackbody component with arbitrary power
laws: in both cases the additional components were completely unconstrained and resulted
in no improvement in the goodness of fit. The 3σ upper limit to power law flux is 5.2 ×
1028D2
100 erg s−1keV−1 at 1 keV. Our blackbody models from methods (1) and (2) correspond
formally to a radius over distance ratio (angular size) of R∞/D100 = 4.12 ± 0.68 km/100pc,
where R∞ = R/p1 − 2GM/Rc2 is the “radiation radius” corresponding to the true radius
R for a star of mass M , and the quoted uncertainty represents the combined temperature
determination uncertainty (±1 eV) and the (dominant) absolute effective area uncertainty
of the LETG+HRC-S combination (±15% ).
In all model comparisons for binning at S/N > 10 the residual differences between the
observed counts and the best-fit model show systematic departures from normal statistical
deviations (Figure 1). Broad deviations are characterised by a general overprediction of
observed counts for λ < 30 A and in the region 75-100 A which is dominated by higher order
flux, underprediction by an average of ∼ 10 % for the range 25-38 A, and deviations around
the instrumental C K edge region 40-44 A. The latter results from a residual calibration error
in the HRC-S UV/Ion shield. Other deviations could arise either as a result of impropriety of
a blackbody model for RX J1856.5−3754, or through calibration errors which are currently
estimated to be ≤ 15% over broad spectral ranges and less over narrower ranges (Drake
et al., in preparation). An apparent edge at 60 A and flux excess in the range 60-70 A
arises because of one of the HRC-S plate gap boundaries coincides with small residual QE
differences between positive and relative negative order outer plates.
In the case of narrow line or edge features, in different combinations of spectral order,
data reduction method and binning size, we identify possible structure in residuals at 26.5,
27.6, 34.4, 32.4, and 35.9 A (emission like features), and at 28.2, 39.1, and 86.5 A (absorption
like features), that might be tempting to attribute to the source. However, we cannot exclude
the possibility that any are chance fluctuations at the 1% level. The issue is complicated
by possible calibration uncertainties on smaller scales, believed to be at a level of about
5% or less, that should be largely smoothed out by dither. We have also used a 60ks
observation (ObsID 331) of PKS 2155−304, currently thought to be a featureless continuum
in the spectral range we are concerned with here (Marshall et al., in preparation), as a
flat field source to aid in feature identification. This spectrum comprises about eight times
the number of first order counts of the combined RX J1856.5−3754 data. Moreover, we
have imposed a constraint that features must appear in both positive and negative orders
– 8 –
in the RX J1856.5−3754 spectrum. We examined deviations from the model on scales up
to 3 A and find only the expected normal distribution of residuals after allowing for smooth
departures resulting from calibration. The Kolmogorov-Smirnov test applied to deviations
on different scales also revealed no evidence for significant features these scales. In summary,
all significant deviations in the residuals that have been found can reasonably be explained
by instrumental effects. Equivalent width upper limits were derived by applying counting
statistics to a convolution of the spectrum with a triangular kernel (Figure 2).
5.
Interstellar Medium Absorption
The distance of 62 pc derived by Walter (2001) seems at odds with the neutral H column
density, NH , of 1020 cm−2 derived from the Chandra spectra: measured NH values for ob jects
at this distance based on different techniques are typically in the range 1018 -1019 cm−2 (e.g.
Fruscione et al. 1994). Walter (2001) indeed remarked on this, citing reddening values
EB−V of up to 0.1 derived by Knude & Høg (1998) in support of the distance. However,
these reddening values show considerable scatter at low reddening and are based only on a
relatively coarse attribution of spectral type to the stars considered.
We have estimated NH and the mean local neutral hydrogen number density, nH , in the
line-of-sight toward RX J1856.5−3754 as a function of distance by spatial interpolation in
the measurements compiled by Fruscione et al. (1994) and Diplas & Savage (1994) using a
technique developed by P. Jelinsky (unpublished). We illustrate this in Figure 3, together
with the allowed distance ranges from Walter (2001) and Kaplan et al. (2002). The value
of NH estimated in this way at a distance of 60pc is an order of magnitude lower than the
X-ray measurement, but is in good agreement with the distance of 140pc derived by the
latter authors. This distance would place RX J1856.5−3754 on the outskirts of the R CrA
cloud using the cloud distance of Knude & Høg (1998) of ∼ 170pc and within the cloud using
the canonical cloud distance of ∼ 130 pc. In either case, RX J1856.5−3754 would likely lie
in a region of relatively high ISM density (∼ 1-10 cm−3 ). While these estimates of NH are
crude and will smooth out any small-scale ISM inhomogeneities, the larger distance is easier
to reconcile with the measured column. The cloud distance of Knude & Høg (1998) then
represents an upper limit to the distance of RX J1856.5−3754.
– 9 –
6. Discussion
Our results, combined with the recent analysis of Ransom et al. (2002), demonstrate a
lack of pulsed features above a level of 2.7 % (unaccelerated search; 4.5% from the accelerated
search of Ransom et al.) and no unequivocal detection of spectral features. This dearth of
indices with which to restrict parameter space precludes an obvious answer to the problem
of the nature and origin of RX J1856.5−3754.
The apparent lack of electron or proton resonance cyclotron absorption suggests that
magnetic field strengths in the ranges (1-7) × 1010 and (0.2-1.3) × 1014 G are less likely,
as discussed by Paerels et al. (2001) for RXJ 0720.4−3125 and by Burwitz et al. (2001),
but, as emphasised by the latter, should not be excluded owing to possible difficulties in
detecting the absorption features.
Indeed, neutron stars with different levels of magnetic
field up to 1015 G have now been observed with high resolution X-ray spectrometers and
none have so far shown absorption features that are intrinsic to the stellar photosphere (e.g.
RXJ 0720.4−3125—Paerels et al. 2001; PSR 0656+14—Marshall & Schulz 2002; Vela—
Pavlov et al. 2001; and 4U 0142+61—Juett et al. 2002).
Our derived angular size based on modelling of the LETGS spectra, R∞/D100 = 4.12 ±
0.68 km/100pc, is consistent with that of Burwitz et al. (2001), though the revised allowed
distance range of 111-200 pc (Kaplan et al. 2002), together with the distance upper limit
constraint based on the R CrA cloud, D100 ≤ 1.70, now implies a radiation radius in the
range R∞ = 3.8-8.2 km. The high end of this range, corresponding to the largest allowed
distance, is still inconsistent with current “normal” NS equations of state (R∞ & 12 km;
e.g. Lattimer & Prakash 2000), as well as with those with extreme softening, such as kaon
condensate models.
Pons et al. (2001) find that heavy element-dominated atmosphere models provide a
plausible match to the low resolution ROSAT PSPC spectra and UV and optical fluxes of
RX J1856.5−3754, while Kaplan et al. (2002) alleviate conflicts with standard EOS through
their revised distance. The heavy element models yield larger radii by virtue of having cooler
effective temperatures than blackbody spectra with similar energy distributions. However,
Burwitz et al. (2001) argued against such uniform temperature heavy element-dominated
atmosphere solutions based on a lack of the expected spectral line features. The appar-
ently featureless but much higher quality LETGS spectra presented here strengthen these
conclusions.
An alternative favoured by Pons et al. (2001), Burwitz et al. (2001) and Ransom et al.
(2002) is a non-uniform temperature model—that we are only seeing a localised hot region
on the surface of a cooler star. The latter authors argue that the gravitational smearing
– 10 –
effects described by Psaltis, Ozel, & DeDeo (2000) account for a lack of observed pulsations.
However, the pulse fraction expectations of Psaltis et al. (2000) indicate X-ray pulse fraction
levels below our 2.7 % limit would be seen only ∼ 10-15 % of the time.
Pulsation would also be expected for a young (∼ 106yr), cooling INS with a strong
magnetic field. The alternative—that RX J1856.5−3754 is an older ob ject re-heated by ISM
accretion—has been dismissed by Kaplan et al. (2002) and van Kerkwijk & Kulkarni (2001a),
largely based on a Bondi-Hoyle accretion rate for the space velocity of Walter (2001) and
Kaplan et al. (2002) that would be much too low to explain the observed luminosity, and
on a possibly strong, accretion-inhibiting magnetic field (Pons et al. 2001). Nevertheless,
RX J1856.5−3754 appears to be in the outskirts of the R CrA cloud and, based on its HST-
derived velocity vector, has likely passed through more dense regions than it now resides
in. Finding one of only a handful of INS candidates in such a region by chance is very
unlikely, yet it is just such dense ISM regions that are expected to power accretion-heated
INSs. Based on the large velocity, however, accretion could only be significant if the Bondi-
Hoyle formalism were to be inapplicable for RX J1856.5−3754. While we tentatively ascribe
the ∼ 2% change in observed 0th order count rate in the 19 months separating the two
observation sets to detector effects, such variability could be accommodated by an accretion
model as the star passed through ISM density fluctuations on ∼AU scales.
The lack of spectral features is also consistent with a pure hydrogen atmosphere model
that is expected to result from modest accretion, whereby heavier elements undergo rapid
gravitational settling. Pons et al. (2001) argue that standard pure H models overpredict the
optical flux of RX J1856.5−3754 by a factor of 30 and that the magnetic accreting models of
Zane, Turolla & Treves (2000), while capable of reproducing the observed X-ray to optical
flux ratio, would need to be two orders of magnitude brighter and an order of magnitude
hotter than observed. However, different accretion scenarios and atmospheric models might
bear examination in light of the otherwise coincidental location of RX J1856.5−3754.
The slightly unfavourable odds of a non-uniform temperature model failing to show signs
of pulsation leads us to consider a third possibility. Taken at face value, the distance, NH ,
and lack of spectral and temporal features and pulsations favour a more compact ob ject than
current NS models permit, but one that is allowed for in strange quark matter solutions (e.g.
Lattimer & Prakash 2000). There now exists a body of evidence from heavy-ion collision
experiments supporting the viability of a quark-gluon plasma (Heinz 2001), and the possible
existence of “strange stars” (e.g. Alcock et al. 1986) is perhaps not as speculative as it once
was. As noted by Pons et al. (2001), such an ob ject would be expected to have a thermal
spectrum as we observe. Such a suggestion is not unprecedented and there now exists a small
handful of ob jects whose apparent compactness could be explained if they are composed of
– 11 –
quark matter (e.g. Cheng et al. 1998; Li et al. 1995; Xu et al. 1999). Of the existing quark
star candidates, RX J1856.5−3754 arguably presents the strongest and most direct case. If
this case survives future scrutiny, then the likelihood of such an ob ject being the brightest
and closest of the current few INS candidates would add some support to speculation that
such a state of matter is a common product of supernovae explosions, or a common phase or
endpoint in the evolution of a neutron star (e.g. Alcock et al. 1986; Kapoor & Shukre 2001;
Xu, Zhang & Qiao 2001).
We extend warm thanks to Pete Ratzlaff for invaluable assistance and “wonder scripts”
developed at short notice. We also thank members of the CfA HEAD for useful comments and
corrections that enabled us to improve the manuscript. The SAO authors were supported by
NASA contract NAS8-39073 to the Chandra X-ray Center during the course of this research.
HLM was supported by NASA contract SAO SV1-61010.
Alcock, C., Farhi, E., & Olinto, A. 1986, ApJ, 310, 261
REFERENCES
Burwitz, V., Zavlin, V. E., Neuhauser, R., Predehl, P., Trumper, J., & Brinkman, A. C.
2001, A&A, 379, L35.
Campana, S., Mereghetti, S., & Sidoli, L. 1997, A&A, 320, 783
Caraveo, P. A., Bignami, G. F., & Trumper, J. E. 1996, A&A Rev., 7, 209
Cheng, K. S., Dai, Z. G., Wei, D. M., & Lu, T. 1998, Science, 280, 407
Diplas, A. & Savage, B. D. 1994, ApJS, 93, 211
Fruscione, A., Hawkins, I., Jelinsky, P., & Wiercigroch, A. 1994, ApJS, 94, 127
Gregory, P. C. & Loredo, T. J. 1992, ApJ, 398, 146
Heinz, U. 2001, Nuclear Physics A, 685, 414
Li, X.-D., Bombaci, I., Dey, M., Dey, J., & van den Heuvel, E. P. J. 1999, Physical Review
Letters, 83, 3776
Juett, A. M., Marshall, H. L., Chakrabarty, D., & Schulz, N. S. 2002, ApJ, 568, L31.
Kaplan, D. L., van Kerkwijk, M. H., & Anderson, J. 2002, ApJ, in press
– 12 –
Kapoor, R. C. & Shukre, C. S. 2001, A&A, 375, 405
Knude, J. & Høg, E. 1998, A&A, 338, 897
Lattimer, J. M. & Prakash, M. 2001, ApJ, 550, 426
Lattimer, J. M. & Prakash, M. 2000, Phys. Rep., 333, 121
Marshall, H.L., & Schulz, N.S. 2002, ApJ, in press
Paerels, F. et al. 2001, A&A, 365, L298
Pavlov, G. G., Zavlin, V. E., Sanwal, D., Burwitz, V., & Garmire, G. P. 2001, ApJ, 552,
L129
Pons, J. A., Walter, F. M., Lattimer, J. M., Prakash, M., Neuhauser, R., & An, P. 2002,
ApJ, 564, 981.
Psaltis, D., Ozel, F., & DeDeo, S. 2000, ApJ, 544, 390
Ransom, S. M., Gaensler, B. M., & Slane, P. O. 2002, ApJ, in press
Shapiro, S. L. & Teukolsky, S. A. 1983, Black Holes, White Dwarfs, and Neutron Stars. New
York: John Wiley and Sons 1983
Treves, A., Turolla, R., Zane, S., & Colpi, M. 2000, PASP, 112, 297
Tsuruta, S. 1998, Phys. Rep., 292, 1
van Kerkwijk, M. H. & Kulkarni, S. R. 2001a, A&A, 380, 221
van Kerkwijk, M. H. & Kulkarni, S. R. 2001b, A&A, 378, 986
Walter, F. M. 2001, ApJ, 549, 433
Walter, F. M. & Matthews, L. D. 1997, Nature, 389, 358
Walter, F. M., Wolk, S. J., & Neuhauser, R. 1996, Nature, 379, 233
Xu, R. X., Qiao, G. J., & Zhang, B. 1999, ApJ, 522, L109
Xu, R. X., Zhang, B., & Qiao, G. J. 2001, Astroparticle Physics, 15, 101
Zane, S., Turolla, R., & Treves, A. 2000, ApJ, 537, 387
This preprint was prepared with the AAS LATEX macros v5.0.
– 13 –
Fig. 1.— The combined positive and negative order spectra of RX J1856.5−3754 binned at
0.5 A intervals shown with the best fit blackbody model with parameters corresponding to
method (2) in §4 and residuals (observations−model). The deviations from this model are
consistent with Poisson statistics after allowing for calibration uncertainties at the C K-edge
and over broader wavelength intervals. The apparent edge at 60 A results primarily from
one of the HRC-S plate gap boundaries and small residual QE differences between positive
and relative negative order outer plates.
– 14 –
Fig. 2.— The 3σ equivalent width upper limit to line features as a function of wavelength
based on a convolution of the spectrum with a triangular kernel.
– 15 –
Fig. 3.— The estimated NH (solid curves) and nH (dotted curves) as a function of distance in
the line-of-sight toward RX J1856.5−3754. The pairs of curves represent the likely ranges of
these quantities and correspond to deviations of factors of 3 from a smooth locus through the
results of the 3D interpolations. Vertical shaded regions indicate the distance ranges found
by Walter (2001; W) and Kaplan et al. (2002; KvKA). The horizontal stripe illustrates the
allowed range (±1σ) of NH derived in this study.
– 16 –
Table 1. Summary of Chandra LETG+HRC-S Observations
Obs ID
UT Start
UT End
Exposure
[s]
Net Eventsa
0th + 1st
0th
0th Rateb
[Hz]
113
3382
3380
3381
3399
2000-03-10 07:55:12
2001-10-08 08:18:49
2001-10-10 05:06:28
2001-10-12 19:19:26
2001-10-15 11:47:06
2000-03-10 23:37:24
2001-10-09 03:01:50
2001-10-12 04:00:48
2001-10-14 09:14:28
2001-10-15 14:42:59
55121
101172
166325
169956
9282
12202
20949
35097
36011
1962
— 0.2195 ± 0.0020
0.2166 ± 0.0016
86516
0.2154 ± 0.0013
135230
141349
0.2157 ± 0.0013
0.2126 ± 0.0051
7136
aThe net event numbers includes background events; while including the 1st order events increases the
net source events, the number of background events also increases.
bRates were derived from 0th order data when valid event rates did not exceed 184 count s−1 ; this
selection criterion yielded 417786 seconds of the 501856s total exposure time.
|
astro-ph/0111323 | 1 | 0111 | 2001-11-16T10:18:23 | CN and HNC Line Emission in IR Luminous Galaxies | [
"astro-ph"
] | We have observed HNC 1-0, CN 1-0 and 2-1 line emission in a sample of 13 IR luminous (LIRGs, L_IR > 10E11 Lo) starburst and Seyfert galaxies. HNC 1-0 is detected in 9, CN 1-0 is detected in 10 and CN 2-1 in 7 of the galaxies. We also report the first detection of HC3N (10-9) emission in Arp220. The excitation of HNC and CN emission requires densities n > 10E4 cm-3. We compare their intensities to that of the usual high density tracer HCN. The I(HCN)/I(HNC}) and I(HCN)/I(CN) 1-0 line intensity ratios vary significantly, from 0.5 to >6, among the galaxies. This implies that the actual properties of the dense gas is varying among galaxies who otherwise have similar I(CO)/I(HCN) line intensity ratios. We suggest that the HNC emission is not a reliable tracer of cold (10 K) gas at the center of LIRGs, as it often is in the disk of the Milky Way. Instead, the HNC abundance may remain substantial, despite high gas temperatures, because the emission is emerging from regions where the HCN and HNC formation and destruction processes are dominated by ion-neutral reactions which are not strongly dependent on kinetic temperature. We find five galaxies (four AGNs and one starburst) where the I(HCN)/I(HNC) intensity ratio is close to unity. In other AGNs, however, I(HCN)/I(HNC}) is >4. The CN emission is on average a factor of two fainter than HCN, but the variation is large and there seems to be a trend of reduced relative CN luminosity with increasing IR luminosity. One galaxy, NGC3690, has a CN luminosity twice that of HCN and its ISM is thus strongly affected by UV radiation. We discuss the I(HCN)/I(HNC) and I(HCN)/I(CN) line ratios as indicators of starburst evolution. | astro-ph | astro-ph |
Astronomy & Astrophysics manuscript no.
(will be inserted by hand later)
CN and HNC Line Emission in IR Luminous Galaxies
S. Aalto1, A. G. Polatidis1, S. Huttemeister1
,
2, S. J. Curran3
1 Onsala Rymdobservatorium, Chalmers Tekniska Hogskola, S - 439 92 Onsala, Sweden
2 Astronomisches Institut der Universitat Bochum, Universitatsstrasse 150, D - 44780 Bochum, Germany
3 School of Physics, University of New South Wales, Sydney NSW 2052, Australia
Received / Accepted
I(HNC) 1-0 and I(HCN)
Abstract. We have observed HNC 1-0, CN 1-0 & 2-1 line emission in a sample of 13 IR luminous (LIRGs,
LIR > 1011 L⊙) starburst and Seyfert galaxies. HNC 1-0 is detected in 9, CN 1-0 is detected in 10 and CN 2-1 in 7
of the galaxies and all are new detections. We also report the first detection of HC3N (10-9) emission in Arp 220.
The excitation of HNC and CN emission requires densities n > 104 cm−3. We compare their intensities to that of
the usual high density tracer HCN. The I(HCN)
I(CN) 1-0 line intensity ratios vary significantly, from
∼ 6, among the galaxies. This implies that the actual properties of the dense gas is varying among galaxies
0.5 to >
who otherwise have similar I(CO)
I(HCN) line intensity ratios. We suggest that the HNC emission is not a reliable tracer
of cold (10 K) gas at the center of LIR galaxies, as it often is in the disk of the Milky Way. Instead, the HNC
abundance may remain substantial, despite high gas temperatures, because the emission is emerging from regions
where the HCN and HNC formation and destruction processes are dominated by ion-neutral reactions which are
not strongly dependent on kinetic temperature. We find five galaxies (Mrk 231, NGC 7469, NGC 7130, IC 694
and NGC 2623) where the I(HCN)
I(HNC) intensity ratio is close to unity. Four are classified as active galaxies and one
as a starburst. In other active galaxies, however, the I(HCN)
I(HNC) is > 4. The CN emission is on average a factor of
two fainter than the HCN for the luminous IR galaxies, but the variation is large and there seems to be a trend
of reduced relative CN luminosity with increasing IR luminosity. This trend is discussed in terms of other PDR
tracers such as the [C II] 158 µm line emission. One object, NGC 3690, has a CN luminosity twice that of HCN
and its ISM is thus strongly affected by UV radiation. We discuss the I(HCN)
line ratios as indicators
of starburst evolution. However, faint HNC emission is expected both in a shock dominated ISM as well as for a
cloud ensemble dominated by dense warm gas in the very early stages of a starburst. Additional information will
help resolve the dichotomy.
I(HNC) and I(HCN)
I(CN)
Key words. galaxies: evolution -- galaxies: ISM -- galaxies: starburst -- radio lines: galaxies -- radio lines: ISM
1. Introduction
The polar molecule HCN (dipole moment 2.98 debye) is
commonly used as a tracer of dense molecular gas, i.e. gas
at n(H2) ≥ 104 cm−3. In particular in distant luminous
(LIR > 1011 L⊙, LIRGs) and ultraluminous (LIR > 1012
L⊙, ULIRGs) systems the HCN 1 -- 0 line is the prototypi-
cal tracer of dense gas content (e.g., Solomon et al. 1992;
Helfer & Blitz 1993; Curran et al. 2000 (CAB)). Solomon
et al. (1992) find a tighter correlation between FIR and
HCN luminosity than the one found between FIR and CO.
They suggest that, in general, the IR luminosities origi-
nate from star formation rather than AGN activity in FIR
luminous galaxies. The HCN to CO intensity ratio, how-
ever, varies substantially ( 1
40 ) among luminous galax-
ies, and it is unclear whether this difference can simply
be interpreted as variations in the dense gas content or
3 - 1
Send offprint requests to: S. Aalto
is also due to abundance and/or excitation effects. Apart
from being collisionally excited, HCN may become excited
via electron collisions (at X(e) ≈ 10−5) or be pumped
by 14 µm continuum radiation through vibrational tran-
sitions in its degenerate bending mode. It is also difficult
to know if the gas is really engaged in star formation, or
if it is simply dense in response to being near the central
potential of the galaxy (e.g., Helfer & Blitz 1993; Aalto et
al. 1995) where other mechanisms (AGN, turbulence etc)
may heat the gas and dust.
In order to understand the activities in the centers of
luminous galaxies it is essential to also understand the
prevailing conditions of the dense gas. Apart from ob-
serving higher transitions of HCN it is important also
to study the emission from other high density tracers.
One such tracer is the HNC molecule, the isomer of (and
chemically linked to) HCN. For example, at high temper-
atures HNC can be transferred into HCN via the reaction
2
S. Aalto, A. G. Polatidis, S. Huttemeister, S. J. Curran: CN,HNC in Galaxies
HNC + H → HCN + H. It is predicted, e.g. in (maybe
oversimplified) chemical steady state models, but also by
shock models, that the HCN
HNC ratio increases with increas-
ing temperature and gas density (e.g., Schilke et al. 1992
(S92)). This is supported by the fact that the measured
HCN
HNC abundance ratio is especially high in the vicinity of
the hot core of Orion KL. Most of the temperature de-
pendence is between 10 and 50 K, after which there is a
considerable flattening (S92).
Compared to these results, the HCN
HNC intensity ratios
found (so far) in nearby starburst galaxies are rather low
(ranging from 1-5) closer to dark clouds than to hot cores
(Huttemeister et al. 1995 (H95)). This result is in apparent
∼ 50 K)
contradiction with the idea that the gas is warm (T >
in the centers of starburst galaxies (e.g., Wild et al. 1992;
Wall et al.1993). However, Aalto et al. (1995) suggest that
the dense cores of the molecular clouds of the starburst
NGC 1808 are cold (10 K) and thus these cores could
be responsible for the HNC emission in NGC 1808, but
possibly also in other galaxies.
The radical CN is another tracer of dense gas, with
a somewhat lower (by a factor of 5) critical density
than HCN. Observations of the CN emission towards the
Orion A molecular complex (Rodriguez-Franco et al. 1998)
show that the morphology of the CN emission is dom-
inated by the ionization fronts of the HII regions. The
authors conclude that this molecule is an excellent tracer
of regions affected by UV radiation. Thus, the emission
from the CN molecule should serve as a measure of the
relative importance of gas in Photon Dominated Regions
(PDRs).
>
We have searched for HNC and CN emission in
a sample of LIRG and ULIRG galaxies with warm
( f (60 µm)
∼ 0.75) FIR colours. We were interested to see
f (100 µm)
whether the HNC emission would be relatively fainter
compared to the cooler, nearby objects studied by H95. Is
HNC a reliable cold gas tracer, or would we find evidence
for the contrary? We furthermore wanted to assess the rel-
ative importance of dense PDRs in these objects through
comparing the CN line brightness with that of HCN. If in-
deed the HNC emission is a tracer of the amount of cold,
dense gas, then perhaps an anti-correlation between the
CN and HNC emission is to be expected. Many of the
galaxies in the survey are powered by prodigious rates of
star formation and thus a bright CN line relative to that
HCN is to be expected. Some of the galaxies are domi-
nated by an AGN where the expected CN brightness may
also be high (e.g., Krolik & Kallman 1983).
In Sect. 2, we present the observations and in Sect. 3
the results in terms of line intensities and line ratios. In
Sect. 4.1 we discuss the interpretation of the HNC results
and in Sect. 4.2 we discuss CN. In 4.3 possible connec-
tions to starburst evolution and scenarios of the possibly
dominating gas components are discussed.
Table 1. Beamsizes and efficiencies
Transition
ν [GHz]
HPBW [′′]
ηmb
OSO SEST OSO SEST
HCN 1-0
HNC 1-0
CN 1-0
CO 1-0
88.6
90.6
113.5
115.2
44
42
34
33
57
55
46
45
0.59
0.59
0.50
0.50
0.75
0.75
0.70
0.70
2. Observations
I(CO)
I(HCN) 1-0 intensity ratios <
We have used the SEST and OSO 20m to measure the
HNC 1-0 (90.663 GHz) and the CN 1-0 113.491 GHz (1-
0, J=3/2-1/2, F =5/2-3/2) line intensity in a selection
of 13 LIRGs and ULIRGs. We also include observations
of NGC 1808 which is of lower luminosity. The selected
∼ 15
galaxies all have global
(apart from NGC 3256 and NGC 1808). For the south-
ern galaxies observed with SEST, we were also able to
measure the CN 2-1 line (226.874 GHz (2-1, J=5/2-3/2,
F =7/2-5/2), 226.659 GHz (2-1 J=3/2-1/2 F=5/2-3/2)).
The CN 1-0 113.191 GHz line (1-0 J=1/2-1/2 F=3/2-3/2)
is shifted +806 km s−1 from the main line and we have ob-
tained limits to its intensity in several cases. For Arp 220
the two CN 1-0 spingroups are blended because the line
is so broad. Thus, even in the 1 GHz correlator backend
(see below) it was necessary to observe CN 1-0 at two dif-
ferent LO settings and then join the spectra together to
get enough baseline. For four galaxies the bandwidth was
wide enough to also include the 90.983 GHz 10-9 line of
HC3N.
Observations were made in 1999 October (HNC,
OSO), December (HNC, SEST) and 2000 June (CN,
OSO), August (CN, SEST). For OSO, the system tem-
peratures were typically 300 K for HNC and 500-600
K for CN. For SEST, typical system temperatures were
230 K for the HNC measurements and 400 K for both the
113 GHz and the 226 GHz CN observations. Pointing was
checked regularly on SiO masers and the rms was found
to be 2′′ for OSO, and 3′′ for SEST. Arp 220 was observed
both with OSO (CN) and SEST (HNC). We have also
measured the 115 GHz CO 1-0 and the HCN 1-0 lines for
some galaxies where we did not have values from the lit-
erature. Beamsizes and efficiencies are shown in Table 1.
For the OSO observations a 500 MHz filterbank was used
for backends for all observations, and for some a 1 GHz
autocorrelator was also used. For the SEST observations
we alternated between a 500 MHz and 1 GHz backend de-
pending on whether simultaneous observations with the 1
and 3 mm receiver were taking place. We used the software
package xs (written by P. Bergman) to subtract baselines
and add spectra.
3. Results
S. Aalto, A. G. Polatidis, S. Huttemeister, S. J. Curran: CN,HNC in Galaxies
3
Table 2. Integrated Line Intensitiesa
Galaxy
Telescope
vc
I(HNC) 1-0b
I(CN) 1-0c
I(CN)2-1d
5/2 - 3/2
3/2 - 1/2
I(CO) 1-0
km s−1
K km s−1
K km s−1
K km s−1
K km s−1
K km s−1
K km s−1
Arp 220
"
IC 694f
NGC 3690f
Mrk 231
Mrk 273
NGC 34
NGC 1614
NGC 2146
NGC 2623
NGC 6240
NGC 3256
NGC 7130
NGC 7469
OSO
SEST
OSO
OSO
OSO
OSO
SEST
SEST
OSO
OSO
SEST
SEST
SEST
OSO
5400
5400
3100
3050
12650
11320
5930
4500
900
5500
7335
2800
4840
4960
. . .
0.95 ± 0.2
0.75 ± 0.2
. . .
0.5 ± 0.1
2.0 ± 0.3
. . .
<
∼ 0.85
1.5 ± 0.3
0.5 ± 0.2:
<
<
∼ 0.25
∼ 0.3
. . .
. . .
0.6 ± 0.15
0.4 ± 0.2:
0.6 ± 0.05
0.4 ± 0.05
0.7 ± 0.1
<
∼ 0.25
∼ 0.3
<
0.6 ± 0.3:
1.2 ± 0.3
<
∼ 0.6
0.7 ± 0.1
1.2 ± 0.05
0.5 ± 0.1
0.8 ± 0.1
. . .
0.65 ± 0.2h
. . .
. . .
. . .
. . .
0.25 ± 0.05
0.5 ± 0.1
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
1.3 ± 0.1
0.55 ± 0.05
0.45 ± 0.05
. . .
0.9 ± 0.1e
0.4 ± 0.05
0.3 ± 0.05
. . .
0.4 ± 0.1e
0.15 ± 0.05
0.15 ± 0.05
. . .
15.0 ± 2.0g
13.0 ± 2.0
14.0 ± 1.0
11.0 ± 1.0
g
g
5.5 ± 0.5g
9.0 ± 0.5
43.0 ± 1.0
13.0 ± 0.5
15.5 ± 0.5
45 ± 1.5
9.5 ± 0.5g
g
NGC 1808
SEST
960
1.2 ± 0.1
3.8 ± 0.1
2.0 ± 0.1
1.4 ± 0.1
0.7 ± 0.1
66 ± 0.5
A and in K km s−1, and upper limits are all 3σ. SEST, Tmb: 27 Jy K−1 (115 GHz), 41 Jy
a): Integrated line intensities are in T ∗
K−1 (230 GHz) for point sources; OSO, Tmb: 27 Jy K−1 (115 GHz) for point sources.
b): In Arp 220 we also detect the 90.98 GHz (10-9) line of HC3N with the intensity 0.4 ± 0.15. For NGC 3256 the (3σ) limit to
the line is 0.18, for NGC 7130 it is 0.15, and for NGC 1808 we have a very tentative detection with the intensity 0.2 ± 0.1. For
Arp 220 we also find an integrated HCN intensity in the SEST beam of 2.5 ± 0.4.
c): This is only the integrated intensity of the 1-0 J=3/2-1/2 line. To get the total 1-0 intensity multiply with 1.3 (in the
optically thin case). For Arp 220 both lines are blended and included in the integrated line intensity listed here.
d): The total integrated intensity from the two 5/2 - 3/2 and 3/2 - 1/2 spingroups together. The following two columns contain
the integrated intensity in each spin group whenever it was possible to measure.
e): Line blending is severe therefore the intensity ratio was estimated from fitting of two gaussians with fixed linewidths of
∆V = 450 km s−1 at v1 = 7200 km s−1and v2 = 7500 km s−1.
f): IC 694 and NGC 3690 are also known as the merger Arp 299.
g): See CAB.
h): very broad lines, may be affected by unknown baseline errors.
3.1. HNC line intensities and ratios
The line intensities are presented in Table 2 and the ra-
tios in Table 3. CO spectra are presented in Figure 1 and
HNC spectra (plus a SEST HCN spectrum for Arp 220)
in Figure 2. For some of the galaxies the center veloc-
ity vc is put to 0. This is because the observer in that
case chose to set the velcocity to zero and work with the
redshifted frequency instead. The velocity used to calcu-
late the redshifted frequency (thus the velocity the band
was centered on) can be found in Table 2. All galaxies,
∼ 1011 L⊙
except NGC 1808, have FIR luminosities LFIR
(see Table 3). Five of the investigated sources (Mrk 231,
NGC 7469, NGC 2623, IC 694 and NGC 7130) have global
HCN
HNC luminosity ratios close to unity. The rest have ra-
tios ranging from 2 to >
∼ 6. The HNC luminous objects
are all AGNs (three Seyferts, one LINER (NGC 2623))
except for IC 694 which we suspect is a starburst (e.g.,
Polatidis & Aalto 2000), but there are also Seyfert galax-
ies with faint HNC emission (such as Mrk 273, NGC 6240
and NGC 34). We do not find that an increase in FIR lu-
minosity is followed by an increase in HCN
HNC line intensity
>
ratio. Instead, there might be a weak trend to the opposite
and two (Mrk 231 and Arp 220) of the three ULIRGs in
our sample have relatively bright HNC emission (see Sect.
3.1.1).
3.1.1. Comparisons with other HNC surveys
In the H95 HNC survey of nearby starburst galaxies, the
majority of the sources show HCN
HNC line ratios greater than
or equal to two. Three objects have I(HCN)≈ I(HNC):
the nearby starburst NGC 253, the nearby post-starburst
NGC 7331 and the Seyfert NGC 3079. For most of the
galaxies in that sample, the line ratios are not global, but
reflect the conditions in the inner 0.2 - 1 kpc of the galaxy.
Since both the HCN and CO emission for most of our
sample galaxies comes from the inner kpc (e.g., Bryant
1996; Scoville et al. 1997; Downes & Solomon 1998; Bryant
& Scoville 1999), it is meaningful to compare the HCN
HNC
ratios of the two samples. In H95 the average HCN
HNC line
intensity ratio is 2 for 14 galaxies (excluding limits and
their value for Arp 220 (see below)). In our sample, the
ratio is somewhat lower, 1.6, when only detections are
4
S. Aalto, A. G. Polatidis, S. Huttemeister, S. J. Curran: CN,HNC in Galaxies
Table 3. Line Ratios and FIR luminosities
Galaxy
typea
CO
CN 1-0
b
CO
HCN
HCN
CN
HCN
HNC
HNC
CN
CN 5/2−1/2
3/2−1/2 CN 2−1
1−0
c
log L(FIR)d
Arp 220
IC 694
NGC 3690
Mrk 231
Mrk 273
NGC 34
NGC 1614
NGC 2146
NGC 2623
NGC 6240
NGC 3256
NGC 7130
NGC 7469
HII
HII
HII
Sey 1
Sey 2
Sey 2/HII
HII
HII
LINER
Seyfert 2
HII
Sey 2/HII
Sey 1/HII
>
>
>
7.5
∼ 17
6
8
∼ 10
∼ 17
12
28
∼ 22
18
29
14
11
>
NGC 1808f HII
13
8
11
13
5
2g
4
14
16
12
9
16
13
8h
18
1
>
>
>
∼ 1.2
0.5
1.6
∼ 5
∼ 4
1
1.8
∼ 2
2
2
1.1
1.4
>
>
>
1.6
1.0e
. . .
1
∼ 4
∼ 4
. . .
. . .
1.4
2.7
3
1.2
1.2
>
0.5
∼ 1
. . .
1.6
. . .
. . .
. . .
. . .
∼ 1.3
0.7
0.7
1
1
>
0.7
2
0.4
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
2.1
2.6
2.4
. . .
2.2
0.15
. . .
. . .
. . .
. . .
>
∼ 0.35
0.35
. . .
. . .
0.6
0.14
0.25
. . .
0.25
12.13
11.77
. . .
12.37
12.04
11.16
11.43
11.00
11.49
11.69
11.52
11.26
11.41
10.48
a): Type of central activity that is suggested to dominate the IR luminosity.
b): The HCN data come from CAB apart from for: IC 694 & NGC 3690 (Arp 299) (Solomon et al. 1992); NGC 1614 (Aalto et
al. 1995); NGC 2623, NGC 6240 (Bryant 1996); NGC 3256 (Aalto et al. 1995). The Bryant (1996) ratios are measured with the
OVRO interferometer, but since both the CO and HCN emission is compact for both NGC 2623 and NGC 6240 the line ratios
are global and can be compared with single dish ratios. For Arp 299 (IC 694 & NGC 3690) the situation is more complicated
since the high density tracer emission is emerging from compact structures, while a fraction of the CO emission comes from
more extended gas (e.g., Aalto et al. 1997). In (Solomon et al. 1992) HCN was measured in a 28′′ beam (IRAM).
c): We have not mapped the CN line emission which is of course necessary for an accurate determination of the CN 2−1
1−0 line
ratio. For most galaxies we expect the CN distribution to be effectively a point source in the beam -- except for NGC 1808
where we adopt a source size of 18′′ (Aalto et al. 1995).
d): The FIR luminosities come from Bryant (1996) (apart from NGC 1808 (Aalto et al. 1994); NGC 34 (CAB); NGC 3256,
NGC 7130, NGC 2146 (Aalto et al. 1991))
e): We have assumed that the HNC emission is a point source (see footnote b) and recalculated the line intensity to that of the
28′′ IRAM HCN beam.
f): Line ratios are not global, but apply only for the inner 2-3 KPC.
g): We use the temperature line ratio of 2 instead of the integrated line ratio of 1 given in CAB (because the HCN line appear
too broad).
h): There is an error in table 4 of CAB. The correct ratio (8) can be found in table 3 in CAB.
included (but when limits are included the ratio increases
to 2). Joining the H95 galaxies with ours in one sample
we still can find no strong trend in the line ratio with
increasing FIR luminosity. We note, however, that we have
a larger number of galaxies with ratios close to unity (6)
compared to H95 (3) despite our smaller sample.
We observed HNC in Arp 220 to see if we could repro-
duce the remarkable result in H95 that HNC was brighter
than HCN. We observed HNC on two consecutive days,
with observations of HCN in between, but could not con-
firm the bright HNC emission found earlier. Instead we
obtain an HCN
HNC intensity ratio of 1.4, which is more con-
sistent with results found in other galaxies.
3.1.2. HC3N
For some galaxies the bandwidth was large enough to in-
clude the HC3N 10-9 line, shifted in velocity by -1000
km s−1 from the HNC line. The line is detected in Arp 220
at 40% of the HNC 1-0 intensity. In NGC 1808 the line is
tentatively detected at 16% of the HNC 1-0 intensity and
in NGC 7130 and NGC 3256 we have upper limits to the
line (see footnote to Table 2).
3.2. CN line ratios and intensities
CN 1-0: The line intensities are presented in Table 2 and
the ratios in Table 3. The CN 1-0 spectra are displayed in
Figure 3. For most of the galaxies we only detect one of
the spingroups in the 1-0 transition. Because of the broad
line in Arp 220 the two groups are however blended --
even though they are separated by 800 km s−1. Also for
NGC 6240 the second spingroup is somewhat blended with
the first one.
The HCN
CN 1-0 integrated line ratio varies substan-
tially: from 0.5 to greater than four. I(CN) >
∼ I(HCN) in
three objects (Arp 220, NGC 3690 and NGC 1808), while
CN remains undetected in IC 694, Mrk 273, NGC 34
and NGC 2623. We also find three galaxies (Mrk 231,
NGC 2623, and IC 694) where I(CN) <
∼ I(HNC). In four
galaxies (Arp 220, NGC 1808, NGC 3256, NGC 6240)
I(HNC) < I(CN) -- three are starburst galaxies and
S. Aalto, A. G. Polatidis, S. Huttemeister, S. J. Curran: CN,HNC in Galaxies
5
NGC 6240 is classified as a Seyfert. In H95, CN was
brighter than HNC in all four sample galaxies where CN
was measured.
The HCN
CN line intensity ratio seems to increase slightly
with FIR luminosity. Dividing our galaxies into two lumi-
nosity bins we find that the average ratio of 1.5 for the
lower luminosity galaxies and 2.3 for the higher luminos-
ity objects. This, somewhat surprising, result is discussed
in Sect. 4.2. We included NGC 3690 in the lower lumi-
nosity batch since most of the FIR emission is believed to
originate in IC 694.
CN 2-1: The 2-1 spectra are presented in Figure 4.
Both 2-1 spingroups are detected in four of the cases (in
NGC 6240 the blending is severe). In NGC 34 and in
NGC 1614 we detect only the brighter J=5/2-3/2 tran-
sition. The CN 2 -- 1 spectrum of Arp 220 shows a much
weaker line than the 1 -- 0 spectrum, and emission is only
detected in the lower velocity part (vc ≈ 5200 km s−1)
of the spectrum. Also the CN 1-0 line peaks around 5200
km s−1. The CO emission peaks at vc ≈ 5400 km s−1. This
implies that most of the CN emission emerges from the
western nucleus (see Scoville et al. 1997). The HC3N line
on the other hand appears to peak at vc = 5550 km s−1,
and thus most of the emission likely emerges from the
eastern nucleus. This tentative velocity difference should
be investigated at higher resolution.
3.2.1. The CN 2−1
1−0 ratio
The CN 2−1
1−0 line intensity ratio (see footnote to Table 3
on how it was calculated) suggests the CN emission be-
ing subthermally excited apart from in NGC 6240 (and
NGC 34 with an upper limit to the CN 1-0 emission). In
general the ratio is lower than 0.4 indicating gas densities
∼ 5 × 104 cm−3 (Fuente et al. 1995) which are about one
order of magnitude below the critical density.
<
3.2.2. The CN spingroup ratio
We can use the 1mm spingroup ratio to estimate the av-
erage optical depth of the two lines. The ratio between
the lines is quite large, >2 for all cases where it can be
measured. The spingroup ratios are very accurate since
they were obtained at the same time (apart from possi-
ble baseline errors). In NGC 6240 the line blending is so
severe that the ratio is difficult to determine. The LTE ra-
tio should be close to 1.8 (Bachiller et al. 1997) and ratios
around 2 thus indicate that the lines are of low optical
depth. For most galaxies, the other spingroup of the 1-0
line (around 113.191 GHz (J=1/2-1/2)) was outside the
observed bandwidth. In the optically thin case, this line
is about 1/3 of the J=3/2-1/2 at 113.490. For Arp 220,
the two lines are blended (see above) and we fitted two
gaussians each centered on one of the spingroups. The fits
shows that a line intensity ratio between the two lines of
1/3 is possible.
4. Discussion
We initially expected that the relative HNC luminosity
would be lower in our sample of warm, luminous galaxies
compared to the H95 sample. Instead (see Sect. 3.1.1) the
very luminous galaxies even seemed to be somewhat more
luminous, on average, in HNC even though this change is
not statistically significant. A possibility could be that the
telescope beam is picking up more extended, cooler gas in
the distant, more luminous, galaxies. However, most of
them are known to have very compact molecular cloud
distributions and it is unlikely that significant HNC emis-
sion emerges from the outskirts of the galaxies. Our results
seem to challenge the notion of HNC as a reliable tracer
of cold gas.
Furthermore, the variation in HCN
HNC line ratio is large
among the galaxies that otherwise have similar properties.
For example, despite having similar CO
HCN 1-0 intensity ra-
tios, Mrk 273 and Mrk 231 have quite different HCN
HNC and
CN line intensity ratios. The HCN
HCN
HNC ratio of Mrk 231 is
close to unity, while HNC is not detected in Mrk 273 re-
sulting in a HCN
∼ 6. It is unlikely that the dense gas
is cold in Mrk 231 but warm in Mrk 273 (both are hot,
ultraluminous AGN/Starburst mergers) and therefore the
interpretation of HNC needs to be reevaluated.
HNC ratio >
4.1. Why is HNC often bright in centers of galaxies?
There are a number of possible explanations. Let us ex-
amine them one by one and see how they can be tested:
1. The dense gas is cold.
Is it possible that a significant fraction of the dense gas
is cold in centers of starburst galaxies? Possibly. Aalto et
al. (1994) discuss the presence of cold dense gas in the
"mild" starburst galaxy NGC 1808. The high dust tem-
peratures observed towards NGC 1808 could be explained
by clouds with hot surfaces and cold interiors. In this sce-
nario, the HNC 1-0 emission would emerge from the cold
cloud cores, while HCN 1-0 emission also would come from
the outer, warmer parts. Gas at densities > 105 cm−3
should become thermalized with the dust. Thus, if the
HNC emission is coming from dense, cold (10-20 K) gas
there should be submm dust continuum emission associ-
ated with it. Therefore, a study of the submm and mm
excess in conjunction with the strength of the HNC emis-
sion should be quite interesting. However, the most HNC
luminous object in our sample, Mrk 213, shows very weak
mm thermal dust emission and Braine & Dumke (1998)
find a dust temperature of 70 K -- the kinetic tempera-
ture of the associated dense gas should be at least as high.
This is not consistent with the idea that the HNC emis-
sion arises from a cold component. Based on ISOPHOT
data Klaas et al. (1997) find that the bulk of the dust
mass in Arp 220 is at a temperature of about 50 K. Thus,
the bright HNC 1-0 emission is unlikely originate in cold
cloud cores.
We conclude that for objects like Mrk 231 and Arp 220,
where there is no mm excess emission, and where the over-
6
S. Aalto, A. G. Polatidis, S. Huttemeister, S. J. Curran: CN,HNC in Galaxies
Arp 220
NGC 34
IC 694
NGC 1614
NGC 1808
NGC 2146
NGC 2623
NGC 3256
NGC 3690
NGC 6240
NGC 7130
Fig. 1. CO spectra of all galaxies, apart from MRK 231, MRK 273 and NGC 7469 which can be found in CAB. The
A. The velocity resolution ranges from 10 TO 50 km s−1. It was selected to be 10% of the FWHM of the
scale is in T ∗
line itself.
all dust temperature is high, HNC 1-0 does not trace cold
dense gas. Instead one or several of the scenarios below
must apply. However, for less extreme objects such as
NGC 1808 there could be enough mass in cold cores that
at least a fraction of the HNC emission could emerge from
them.
2. Chemistry.
Steady state chemistry models by S92 show both a
temperature and a density dependence in the [HCN]
[HNC] abun-
dance ratio. For example, there is a very significant differ-
ence between n(H2)=104 cm−3 and 107 cm−3. For Tkin=
50 K, [HCN]
[HNC] =67 for 107. If the bulk
[HNC] =0.8 for 104 and [HCN]
of the HCN and HNC emission is emerging from gas of
densities 104 −105 then the relative HNC abundance there
may be substantial, despite the high temperature. The
reason for this is that at lower densities reactions with
HCNH+ (HCN and HNC reacts with H+
3 to form HCNH+)
become more important. The ion abundance is higher and
once HCN and HNC become protonised, HCNH+ will re-
combine to produce either HCN or HNC with 50% prob-
ability. At higher densities, the ion abundance is likely
lower and reactions like HNC + O → CO + NH become
more important at high temperatures. This scenario is in-
teresting since the electron and ion abundance is likely
HC N3
cV
HC N3
S. Aalto, A. G. Polatidis, S. Huttemeister, S. J. Curran: CN,HNC in Galaxies
7
Arp 220
HNC
IC 694
Mrk 231
NGC 1808
HNC
NGC 2623
NGC 3256
NGC 6240
NGC 7130
NGC 7469
Fig. 2. HNC 1-0 spectra for the galaxies for which we claim a detection. The scale is in T ∗
A. The 10-9 HC3N line
is indicated in the spectra for Arp 220 and NGC 1808. The velocity resolution ranges from 10 to 50 km s−1. It was
selected to be 10% of the FWHM of the line itself, but for a few cases the S/N of the spectrum required further
smoothing.
higher in PDRs. Therefore, in a PDR chemistry, the con-
nection between HNC abundance and kinetic temperature
may also be weak since we there expect the HCNH+ reac-
tions to be important. Since the CN 2−1
1−0 ratios we measure
indicate subthermal excitation it is reasonable to assume
that most of the HCN,HNC and CN emission is indeed
emerging from gas where the density is below 105 cm−3.
3. Optical depth effects
In Orion, S92 find peak-to-peak intensity ratios be-
tween HCN
HNC of 3 to 4 towards the hot core and ridge.
However, the abundance ratio is much higher, ≈80. Thus,
it is possible that the fairly bright HNC emission in some
galaxy centers is caused by optical depth effects. In objects
where the HCN emission is subthermally excited (like in
Mrk 231) the optical depth of the 1-0 line could be quite
high and perhaps explain part of the apparently too-bright
HNC.
4. IR pumping
Both HCN and HNC may be pumped by an intense
mid-IR radiation field boosting the emission from low den-
sity regions. There has been no direct evidence IR pump-
ing is dominating the HCN excitation in external galax-
ies. However, Barvainis et al. (1997) suggest IR pump-
ing as a possible mechanism behind the HCN emission
of the Cloverleaf quasar. Ultraluminous galaxies, such as
Mrk 231 and Arp 220, have central mid-IR sources with
optically thick radiation temperatures well in excess of
those necessary to pump the HCN molecule (Soifer et al.
1999). For HNC the coupling to the field is even stronger
than for HCN, thus increasing the probability for IR
pumping in extreme galaxies, such as Mrk 231. Even if
the HNC abundance is lower compared to HCN the HNC
emission may have a higher filling factor due to the IR
pumping (if it allows emission from gas clouds otherwise
at too low density to excite the HNC molecule). A com-
parative excitation study of HCN and HNC would help
8
S. Aalto, A. G. Polatidis, S. Huttemeister, S. J. Curran: CN,HNC in Galaxies
Arp 220
Mrk 231
NGC 1614
NGC 1808
NGC 2146
NGC 3256
NGC 3690
NGC 6240
NGC 7130
NGC 7469
Fig. 3. CN 1-0 spectra for the galaxies with detections. The scale is in T ∗
A. The spectrum for Arp 220 shows the
Gaussian fit for both spingroups, since the line is broad enough for them to be blended. The velocity resolution ranges
from 10 to 50 km s−1. It was selected to be 10% of the FWHM of the line itself, but for a few cases the S/N of the
spectrum required further smoothing.
cast light on this issue. Of course, the HNC abundance
must be high enough so that the pumping can be effective
and it must thus happen in regions where the chemistry
is dominated by ion-neutral reactions.
4.2. Interpreting the CN emission
4.2.1. CN chemistry
Studies of Galactic molecular clouds have shown that the
[CN]
[HCN] abundance ratio is increasing in the outer regions
of UV irradiated clouds (e.g., Greaves & Church 1996;
S. Aalto, A. G. Polatidis, S. Huttemeister, S. J. Curran: CN,HNC in Galaxies
9
Arp 220
J=3/2-1/2
NGC 34
J=3/2-1/2
NGC 1614
J=3/2-1/2
J=5/2-3/2
J=5/2-3/2
J=5/2-3/2
NGC 1808
J=3/2-1/2
NGC 3256
J=3/2-1/2
NGC 6240
J=3/2-1/2
J=5/2-3/2
J=5/2-3/2
J=5/2-3/2
NGC 7130
J=3/2-1/2
J=5/2-3/2
Fig. 4. CN 2-1 spectra for the galaxies with detections. The scale is in T ∗
A. The band is centered in between the two
spingroups, apart from Arp 220, where the spectrometer was centered on the J=5/2-3/2 line. The two spingroups are
marked with arrows and are blended for all galaxies, apart from NGC 7130. FOR Arp 220 emission is not detected at
the line center (marked by the arrow) instead, there is a tentative detection at V =5170 km s−1 (blueshifted from Vc
by approximately 250 km s−1). The velocity resolution ranges from 10 to 50 km s−1. It was selected to be 10% of the
FWHM of the line itself, but for a few cases the S/N of the spectrum required further smoothing.
Rodriguez-Franco et al. 1998). The abundance of the CN
radical becomes enhanced at the inner edge of a PDR (at
an AV of about 2 magnitudes) via the reaction N + C2 →
CN+C or via N+CH → CN+H. At larger depths into the
cloud the CN abundance radically declines and the [HCN]
[CN]
abundance ratio increases (Jansen et al. 1995). Most of the
CN is present in a part of the cloud where the abundance
of free electrons is rather large, X(e) ≈ 3×10−5. CN is also
a photodissociation product of HCN. Thus the CN abun-
dance should be favoured in a molecular cloud ensemble
dominated by PDRs. Furthermore, chemical models (e.g.,
Krolik & Kallman 1983; Lepp & Dalgarno 1996) show that
the CN abundance should also be enhanced when the X-
ray ionization rates are high -- as might be the case near
an AGN.
It is of course difficult to translate a measured HCN
CN 1-0
line intensity ratio to an abundance ratio between the two
species. The spingroup line ratios (see Sect. 3.2.1.) show
that the CN 2-1 emission is optically thin for most galax-
ies we have measured. The emission of the CN molecule is
distributed in a greater number of transitions than HCN,
thus the optical depth per transition is often lower for CN,
reducing the intensity per line. The CN luminosity then
becomes a measure of the total number of CN molecules
(at least in a comparative sense, given a constant excita-
tion situation from galaxy to galaxy). For the CN 1-0 line
we have only information for Arp 220 where the relative
faintness of the second spingroup suggests that the optical
depth of the 1-0 line is also low. So, the measured (total)
HCN
CN line intensity ratio will give a reasonable idea of the
abundance ratio if also the HCN line is close to being op-
10
S. Aalto, A. G. Polatidis, S. Huttemeister, S. J. Curran: CN,HNC in Galaxies
tically thin (and a lower limit to the abundance ratio if
it is not) and if the same excitation temperature can be
assumed. The critical density of the CN line is lower by a
factor of a few, so its Tex is likely somewhat higher.
4.2.2. AGNs or starbursts?
∼ I(HCN)
There are only a few galaxies where I(CN) >
(Arp 220, NGC 3690 and NGC 1808) -- all three are
starburst galaxies. For these galaxies N (CN) is greater
than N (HCN) and the dense gas is to a large degree af-
fected by photodissociation. The most extreme example
is NGC 3690 where the CN line is on average more than
a factor of two brighter than HCN. In Arp 220 the CN
appears to be bright only towards the western of the two
nuclei (Sect. 3.2). This is interesting, since the situation is
similar for the IC 694/NGC 3690 system: bright CN emis-
sion towards one of the galaxy nuclei only. It is interesting
to speculate on whether the burst towards NGC 3690 is
older than the one in IC 694 because of the development
of the PDRs. However, the IC 694 burst is more compact
and it is possible that the actual properties of the ISM are
intrinsically different.
We were surprised to find that CN was difficult to
detect in several of the brightest galaxies like Mrk 273,
NGC 2623, NGC 6240, IC 694 and NGC 34. Four of these
galaxies are AGNs and it is tempting to speculate that
the CN deficiency is related to the nuclear activity. This
seems contrary to models (see above) which predict an in-
crease in the CN abundance in an X-ray chemistry. High
resolution studies of nearby systems which contain both
starburst and AGN activity (like NGC 1068) will reveal
whether CN emission is associated with one or both of the
activities.
4.2.3. The CN and the [C II] 158 µm line
In ULIRGs such as Arp 220 and Mrk 231 the [C II] 158
µm fine structure line is found to be abnormally faint com-
pared to other, less FIR luminous, starburst galaxies like
NGC 3690 (e.g., Luhman et al. 1998). This is interesting,
since one would expect the emission from a standard PDR
tracer, like the [C II] line, to be bright in a galaxy that
is believed to be powered largely by mighty starbursts.
Malhotra reports a decreasing trend in F[CII]
with increas-
FFIR
ing f (60)
f (100) µm flux ratio.
Several possible explanations for the [C II] faintness are
brought forward (e.g., Malhotra et al. 1997; Luhman et al.
1998; van der Werf 2001). The PDRs may be quenched in
the high pressure, high density environment in the deep
potentials of the ULIRGs and the HII regions exist in
forms of small-volume, ultracompact HII regions that are
dust-bounded. The [C II] line may become saturated ei-
ther in low density (n ∝ 102 cm−3) regions of very high
UV fields (G0 ∝ 103) or in dense (n ∝ 105 cm−3) re-
gions of more moderate UV fields (G0 = 5 − 10). A soft
UV field from an aging starburst is another possibility. A
higher dust-to-gas ratio would also decrease the expected
F[CII]
FFIR
ratio.
The molecular ISM of ULIRGs seems to be character-
ized by subthermally excited CO and very bright emis-
sion from HCN (e.g., Downes & Solomon 1998; Aalto et
al. 1995; Solomon et al. 1992). Crudely, this can be mod-
elled as dense clouds (n = 104 − 105 cm−3) embedded in
a low density (n = 100 − 500 cm−3) continous medium.
This simple scenario may fit well with the two scenarios
resulting in saturated [C II] emission.
We have three ULIRGs in our sample: Arp 220,
Mrk 231 and Mrk 273 their HCN
CN line intensity ratio
changes from 1 to >
∼ 6. The deficiency of CN in Mrk 273
is consistent with the lack of [C II] emission and can be
an indication that the PDRs are not forming in the dense
gas. In Arp 220 the CN emission from the western nucleus
is strong and an indicator that a fair fraction of the dense
gas is in fact in a PDR state. The UV radiation is strongly
affecting the dense molecular clouds here. Clearly we need
more information on the properties of the dense gas there
to find an explanation for the lack of [C II] emission.
4.3. Line ratios and starburst evolution
We speculate whether line ratios of dense gas tracers can
be used to explore the evolutionary stage of a starburst.
The observed galaxies can be sorted into rough categories
based on their line ratios.
4.3.1. Warm dense gas -- a young starburst or shocks?
Objects that show bright HCN emission, but little or no
HNC or CN, may be dominated by warm dense gas early
in their starburst development. The Orion KL region is
an example of a Galactic warm (T >
∼ 50 K), dense core
where I(HCN) is significantly greater than both I(CN)
and I(HNC) (e.g., Ungerechts et al. 1997). Also the emis-
sion from the 10-9 transition of HC3N is brighter than
the CN and HNC emission which is a typical signature
of warm, dense gas. However, shocked gas may be an im-
portant part of the molecular ISM in the center of a star-
burst galaxy -- in particular if there is a bar in the center
where clouds on intersecting orbits collide. The interac-
tion between supernova remnants at the surrounding ISM
may also lead to the presence of shocked gas. The effect
of the shock is to compress and heat the gas, which in
some respects will make it look like an ISM dominated by
warm dense cores. The major difference is that the shock,
partially or fully, destroys the dust grains and thereby re-
leasing molecules, such as SiO, into the ISM. Therefore,
SiO emission is often used as a tracer of shocked gas (e.g.,
Martin-Pintado et al. 1992). If the shocked gas is allowed
to cool after the shock (which may happen quickly since
it has been compressed) the HNC abundance will increase
rapidly (S92). An example of a galaxy that could have
a shock dominated ISM is NGC 6240. Very strong IR
emission is emerging from shock-excited H2 (van der Werf
S. Aalto, A. G. Polatidis, S. Huttemeister, S. J. Curran: CN,HNC in Galaxies
11
2001) in between the two merger nuclei and both HNC
and CN line emission is faint relative to that of HCN. In
a high pressure environment the gas may be dense and
warm -- but perhaps heated by dissipation of turbulence
rather than very young embedded stars. This may not oc-
cur during the very early stages of star formation, but
rather be a form of aftermath.
>
(e.g., Churchwell et al. 1984). We know, however, from
other studies of IC 694 that the dense gas is warm
(Tkin
∼ 80K) (Aalto et al. 1998) and the compact CO nu-
cleus of NGC 2623 likely harbours an ISM similar to that
of IC 694. The starburst may have evolved beyond a strong
radiative impact from the stars while the chemisty is still
dominated by ion-neutral reactions at moderate gas den-
sity and relatively high electron abundance.
4.3.2. PDRs
For those galaxies where I(CN) is >
∼ I(HCN) a significant
part of the ISM should be in PDRs. In the Galaxy they
are often found in interface regions between HII regions
and molecular clouds (e.g., Jansen et al. 1995) and near
planetary nebulae. CN luminous galaxies are very clearly
in the phase where the ISM is being strongly affected by
an intense UV field. The filling factor of UV illuminated
gas must be high, and the clouds probably not too large
since the [CN]
[HCN] abundance ratio drops dramatically with
increasing AV. The mechanical impact of the starburst
(superwinds - supernovae) may help in fragmenting the
molecular clouds. Emission from complex molecules, such
as HC3N, is faint because of photodestruction. Since the
chemistry now, to a large degree, involves ion-neutral re-
actions (see Sect. 4.1) the HNC abundance becomes less
dependent on temperature and I(HNC) may be signifi-
cant even from a warm PDR. For example, the temper-
ature around a planetary nebula may become very high
(≈ 100 K) but the HNC abundance is substantial enough
to result in HCN
HNC line intensity ratios close to unity (e.g.,
Herpin & Cernicharo 2000).
For Arp 220, it appears that HC3N is mainly emerging
from the eastern nucleus (see Sect 3.2.1) which would sup-
port the notion of an evolutionary difference between the
two nuclei. Rodriguez-Franco et al. (1998) show that the
emission from HC3N is bright toward hot, dense cores,
while the HC3N/CN abundance ratio is only 10−3 in
PDRs. Thus the eastern nucleus seems to be in an earlier
evolutionary phase where star formation has just begun.
In the mid-IR the western nucleus is more prominent than
the eastern one (Soifer et al. 1999).
As discussed in Sect. 4.2.1. the absence of PDR tracers,
such as CN emission and the 158 µm [C II] fine structure
line, does not necessarily mean that the burst is young
(or shock dominated). In high-pressure, dusty ULIRGs it
is possible that the formation of PDRs is suppressed --
or that the PDRs are associated with the diffuse lower
density molecular material where the classic PDR lines
will not be excited.
4.3.3. An evolved burst
Two galaxies (NGC 2623 and IC 694) show fairly bright
HNC emission but with undetected CN emission. Very
cold (10 K) and dense gas would result in I(HCN)=
0.3 − 0.5 × I(HNC) and I(CN) < I(HCN). In the Galaxy
such conditions dominate clouds like TMC-1 and TMC-2
5. Summary
We have undertaken a SEST and OSO survey of CN and
HNC line emission in a sample of 13 luminous IR galax-
ies, plus one more "normal" starburst (NGC 1808). This
survey is the first in its kind for IR luminous galaxies. The
main conclusions we draw from this survey are as follows:
1. We have detected HNC 1-0 in 9 of the galaxies,
while CN 1-0 is detected in 10 galaxies. The CN 2-1
lines (J=5/2-3/2, F =7/2-5/2; J=3/2-1/2 F=5/2-3/2)
(which could only be measured at SEST) are detected
in 7 galaxies -- in one (NGC 34) the CN 1-0 line was
not detected, while the 2-1 line was.
2. The line intensitites vary significantly from galaxy to
galaxy and the line ratios relative to the "standard"
high density tracer, HCN, are far from constant. The
HCN
HNC ratios vary from 1 to >
CN ratios range
from 0.5 to >
∼ 6. From this we can learn that the ac-
tual properties of the dense gas vary significantly from
galaxy to galaxy, even if their HCN luminosities are
similar.
∼ 6 and the HCN
3. We report the detection of HC3N 10-9 emission in
Arp 220 at 40% of the HNC 1-0 intensity. In NGC 1808
the line is tentatively detected.
4. We find that the IR luminous galaxies are at least as
HNC luminous as more nearby (IR-fainter) galaxies.
We conclude that the HNC emission is not a reliable
tracer of cold (10 K) gas in the center of luminous
IR galaxies, the way it may be in clouds in the disk
of the Milky Way. Instead, we list several alternative
(testable) explanations. Provided that the situation in
the centers of starburst galaxies can be approximated
with a steady state chemistry, the observed HCN
HNC line
ratios can be explained by the emission originating in
∼ 105 cm−3 where the chemistry is
gas of densities n <
dominated by ion-neutral, instead of neutral-neutral,
reactions. The temperature dependence of the HNC
production in the ion-neutral reactions is strongly sup-
pressed.
5. We were surprised to find that the CN emission of the
more luminous galaxies is often faint. This can be con-
sistent with the relative faintness of other PDR tracers
such as the 158 µm [C II] line. Both these phenom-
ena can be understood in terms of a two-component
molecular ISM consisting of low density molecular gas
surrounding dense clouds. However, the relative bright-
ness of CN in Arp 220 may be difficult to reconcile with
its relative [C II] faintness.
12
S. Aalto, A. G. Polatidis, S. Huttemeister, S. J. Curran: CN,HNC in Galaxies
Lepp, S., &, Dalgarno, A., 1996, A&A, 306, L21
Luhman M. L., Satyapal S., Fischer, J., et al., 1998, ApJ, 504,
L11
Malhotra S., Helou G., Stacey G., et al., 1997, ApJL 491, 27
Martin-Pintado, J., Bachiller, R., & Fuente, A., 1992, A&A
254, 315
Polatidis, A., & Aalto, S., 2000, in Proceedings of the 5th EVN
Symposium, Eds. J. Conway, A. Polatidis, R.Booth. Y.
Pihlstrm, Onsala Space Observatory, Chalmers Technical
University, Gothenburg, Sweden (2000), 127
Rodriguez-Franco, A., Martin-Pintado, J., & Fuente, A., 1998,
A&A, 329, 1097
Schilke P., Walmsley C.M., Pineau de Forets G., et al., 1992,
A&A 256, 595 (S92)
Scoville, N.Z., Yun, M.S, & Bryant, P. M., 1997, ApJ, 484, 702
Soifer B. T., Neugebauer G., Matthews K., et al., 1999, ApJ,
513, 207
Solomon P.M., Downes D., & Radford S.J.E., 1992, ApJ 387,
L55
Ungerechts H., Bergin E.A., Goldsmith P.F., et al., 1997, ApJ
482, 245
van der Werf, P.P., 2001, in Starburst galaxies - near and far,
proceedings of the Ringberg workshop, eds. D. Lutz & L.J.
Tacconi, in press
Wall, W.F., Jaffe, D.T, Bash, F.N., et al., 1993, ApJ, 414, 98
Wild, W., Harris, A.I., Eckart, A., et al., 1992, A&A, 265, 447
6. Within Arp 299 (the merging pair IC 694 and
NGC3690) there is a line ratio difference between the
two nuclei. NGC 3690 has a CN luminosity twice that
of HCN and its ISM is thus strongly affected by UV
radiation while CN is not detected towards IC 694.
This reflects an evolutionary gradient in the burst --
or that the star formation activity takes place in very
different environments in the two galaxies. For Arp 220
there may also be an evolutionary gradient between the
two nuclei because of differences in emission velocities
for CN and HC3N.
7. The measured line ratios can in general be used to
identify stages in the starburst evolution, but we con-
clude that the dichotomy between scenarios is a com-
plication. For example, faint HNC emission is expected
both in a shock dominated ISM as well as for a cloud
ensemble dominated by dense warm gas in the very
early stages of a starburst.
Acknowledgements. Many thanks to the OSO and SEST staff
for their help. We are grateful to M. Walmsley and P. Schilke
for discussions on the HNC chemistry. We thank the referee,
F. Herpin, for many useful comments and suggestions.
References
Aalto S., Black J.H., Booth R.S., Johansson L.E.B., 1991,
A&A, 247, 291
Aalto S., Booth R.S., Black J.H., Johansson, L.E.B., 1994,
A&A, 286, 365
Aalto, S., Booth, R.S., Black, J.H., Johansson, L.E.B. 1995,
A&A, 300, 369
Aalto S., Radford, S.J.E., Scoville, N.Z., Sargent, A.I., 1997,
ApJ, 475, L107
Aalto S., Radford, S.J.E., Scoville, NZ., Sargent, A.I., 1998,
IAU Symposium 186 'Galaxy Interactions at Low and High
Redshift', Ed. D. Sanders
Bachiller, R., Fuente, A., Bujarrabal, V., Colomer, F., Loup,
C., Omont, A., de Jong, T., 1997, A&A, 319, 235
Barvainis, R., Maloney, P., Antonucci, R., Alloin, D., 1997,
ApJ, 484, 695
Braine, J., & Dumke, M., 1998, A&A, 333, 38
Bryant, P.M., 1996, PhD thesis, Caltech
Bryant, P.M., & Scoville, N.Z., 1999, AJ, 117, 2632
Churchwell, E., Nash A. G., & Walmsley C. M., 1984, ApJ 287,
681
Curran, S.J., Aalto, S., & Booth, R.S. 2000, A&AS, 141, 193
(CAB)
Downes, D., & Solomon, P.M., 1998, ApJ, 507, 615
Fuente, A., Martin-Pintado, J., & Gaume, R, 1995, ApJ, 442,
L33
Greaves J. S., & Church S. E., 1996, MNRAS 283, 1179
Helfer, T.T., & Blitz L., 1993, ApJ, 419, 86
Herpin, F., & Cernicharo, J., 2000, ApJ, 530, L129
Huttemeister S, Henkel C., Mauersberger R., et al., 1995, A&A
295, 571
Jansen D. J., van Dishoeck,E. F., Black J. H., Spaans M., Sosin
C., 1995, A&A, 302, 223
Klaas, U., Haas, M., Heinrichsen, I., Schulz, B., 1997 A&A,
325, L21
Krolik, J.H., &, Kallman, T.R., 1983, ApJ, 267, 610
|
astro-ph/0101271 | 1 | 0101 | 2001-01-16T16:54:48 | The X-ray Emission in Post-Merger Ellipticals | [
"astro-ph"
] | The evolution in X-ray properties of early-type galaxies is largely unconstrained. In particular, little is known about how, and if, remnants of mergers generate hot gas halos. Here we examine the relationship between X-ray luminosity and galaxy age for a sample of early-type galaxies. Comparing normalized X-ray luminosity to three different age indicators we find that L_X/L_B increases with age, suggesting an increase in X-ray halo mass with time after a galaxy's last major star-formation episode. The long-term nature of this trend, which appears to continue across the full age range of our sample, poses a challenge for many models of hot halo formation. We conclude that models involving a declining rate of type Ia supernovae, and a transition from outflow to inflow of the gas originally lost by galactic stars, offers the most promising explanation for the observed evolution in X-ray luminosity. | astro-ph | astro-ph |
Mon. Not. R. Astron. Soc. 000, 1 -- 8 (0000)
Printed 27 October 2018
(MN LaTEX style file v1.4)
The X -- ray Emission in Post -- Merger Ellipticals
Ewan O'Sullivan1, Duncan A. Forbes1,2 and Trevor J. Ponman1
1School of Physics and Astronomy, University of Birmingham, Edgbaston, Birmingham B15 2TT
(E-mail: [email protected])
2Astrophysics & Supercomputing, Swinburne University, Hawthorn VIC 3122, Australia
27 October 2018
ABSTRACT
The evolution in X -- ray properties of early -- type galaxies is largely unconstrained. In
particular, little is known about how, and if, remnants of mergers generate hot gas
halos. Here we examine the relationship between X -- ray luminosity and galaxy age
for a sample of early -- type galaxies. Comparing normalized X -- ray luminosity to three
different age indicators we find that LX /LB increases with age, suggesting an increase
in X -- ray halo mass with time after a galaxy's last major star-formation episode. The
long-term nature of this trend, which appears to continue across the full age range of
our sample, poses a challenge for many models of hot halo formation. We conclude
that models involving a declining rate of type Ia supernovae, and a transition from
outflow to inflow of the gas originally lost by galactic stars, offers the most promising
explanation for the observed evolution in X-ray luminosity.
Key words:
evolution -- X-rays: galaxies
galaxies: interactions -- galaxies: elliptical and lenticular -- galaxies:
1
INTRODUCTION
A potential problem with forming ellipticals from merging
spirals is how to account for the different gas properties in
the two types of galaxies. In particular, spirals contain rel-
atively high masses of cold (T ∼ 100 K) gas, whereas el-
lipticals have very little. The opposite situation is true for
the hot (T ∼ 106 K) gas masses -- normal spirals contain
rather little hot gas, while ellipticals may possess exten-
sive hot halos. Since the total gas masses per unit stellar
mass are broadly comparable in early and late-type galax-
ies, this raises the possibility that inefficient merger -- induced
star formation might heat the cool interstellar gas in spirals
to form the hot halo in post -- merger ellipticals. Recently,
Georgakakis et al. (2000) have shown that the cold gas mass
indeed decreases in an 'evolutionary sequence' from merging
spirals to post -- merger ellipticals.
However, it is not clear that gas heated in an intense
starburst can be retained within the galactic potential. Read
& Ponman (1998) examined the X -- ray properties of eight
on -- going mergers placed in a chronological sequence. Their
study revealed that material is ejected from merging galax-
ies soon after the first encounter. Massive extensions of hot
gas are seen (involving up to 1010 M⊙) at the ultralumi-
nous peak of the interaction, as the two nuclei coalesce.
There is evidence for this in in the Antennae, Arp 220, and
NGC 2623. However, after this phase of peak activity, the
c(cid:13) 0000 RAS
X -- ray luminosity actually declines. For example, NGC 7252
(a prime example for the remnant of a merger that occurred
0.7 Gyr ago) shows some evidence of a hot halo, but one that
is much smaller than the halos seen in typical ellipticals.
Two post -- merger ellipticals were studied in the X -- ray
by Fabbiano & Schweizer (1995). They found that neither
galaxy had an extensive hot halo. Given the correlation be-
tween X -- ray luminosity and isophotal boxiness (thought to
be a signature of a past merger) demonstrated by Bender
et al. (1989), this was a somewhat surprising result. If post --
merger ellipticals are to resemble 'normal' ellipticals they
need to acquire a hot gas halo. Possible mechanisms include:
1) The late infall (i.e. after a few Gyr) of HI gas associated
with the tidal tails (Hibbard et al. 1994). This cold gas may
be shocked to X -- ray emitting temperatures as it falls back
into the merger remnant.
2) A reservoir of hot gas, possibly expelled at the nuclear
merger stage, might infall from large radii, where its low
density makes it undetectable to current X -- ray satellites.
3) After the initial violent starburst, the continued mass
loss from stars and heating from stellar winds recreates a
hot ISM.
The first study to attempt an 'evolutionary sequence'
for post -- merger ellipticals was that of Mackie & Fabbiano
(1997). For 32 galaxies they showed a weak trend for LX /LB
to increase with a decreasing Σ parameter. The Σ parame-
2
Ewan O'Sullivan et al.
ter is defined by Schweizer & Seitzer (1992), and is a mea-
sure of a galaxy's fine structure, i.e. optical disturbance.
They showed that it correlates with blue colours and Balmer
line strength, and is thus a rough indicator of dynamical
youth. The Mackie & Fabbiano trend therefore suggested
that post -- merger ellipticals became more X -- ray luminous,
for a given optical luminosity, as the galaxy aged. They also
plotted LX/LB against Hβ absorption line EW, which re-
vealed a similar trend. A more recent study by Sansom et al.
(2000) confirms the trend in X -- ray overluminosity with Σ
for 38 galaxies, and suggests that it might be explained by
the build up of hot gas in post -- merger galaxies. However,
both Σ and Hβ EW have their drawbacks -- Σ is only semi --
quantitative at best and the strength of Hβ is affected by
both stellar age and metallicity.
Here we reexamine the trend seen by Mackie & Fab-
biano (1997) for a larger sample, and investigate two new
measures of galaxy age: residual from the Fundamental
Plane and spectroscopic age. Using these three measures we
explore the X -- ray luminosity evolution of post -- merger el-
lipticals.
Throughout the paper we assume H0 = 75 km s−1
Mpc−1 and normalise LB using the solar luminosity in the
B band, LB⊙ = 5.2×1032 erg s−1.
2 RESULTS
We first reexamine the trend of normalized X -- ray lumi-
nosity with Fine Structure parameter Σ, presented initially
by Mackie & Fabbiano (1997) for 32 galaxies. Here we
use a sample of 47 early -- type galaxies with Σ taken from
Schweizer & Seitzer (1992), and normalized LX values from
our recent catalogue (O'Sullivan et al. 2000) of X -- ray lu-
minosities (mostly based on ROSAT data). The LX val-
ues are approximately bolometric. The LB values are based
wherever possible on BT magnitudes taken from Prugniel
& Simien (1996). If these are unavailable, values from NED
are used. Fig. 1 shows the sample of 47 early -- type galaxies
plotted in order of decreasing Σ or increasing age.
As noted by Mackie & Fabbiano, the scatter at low val-
ues of Σ is large, covering around 2 orders of magnitude in
LX /LB, while at higher Σ the scatter appears to be smaller
and LX /LB limited to low values. A similar trend was seen
by Sansom et al. (2000) for 38 galaxies. This suggests that
dynamically young galaxies have low X -- ray luminosities and
that aging produces a range of luminosities, presumably de-
pendent on other factors.
In order to test the strength of this relation we apply
Kendall's K test to the X -- ray detections from our sample.
The test does not assume any distribution in the data, and
the K statistic is unit normal distributed when at least 10
data points are present. For our data we find K = -2.31054
(using 27 data points), indicating an anti correlation be-
tween Log LX /LB and Σ of ∼2-σ significance. For com-
parison, we also apply the test to the sample of Sansom et
al. , and find a correlation of very similar significance, K =
-2.28749 for 30 detections.
2.1 LX /LB versus Spectroscopic Age
Fine structure only persists in dynamically young galaxies,
so Fig. 1 provides information only on the early galaxy evo-
lution of these objects. A more general measure of age is
needed to show how X -- ray properties evolve over a longer
timescale. Galaxy spectroscopic ages are now available for
a large number of early -- type galaxies from the catalogue of
Terlevich & Forbes (2000). These ages are generally based
on Hβ absorption line measurements and the stellar popu-
lation models of Worthey et al. (1994). The line index mea-
surements come from the galaxies' central regions, and are
luminosity weighted. Thus they are dominated by the last
major burst of star formation, which is presumably triggered
by a major merger event. Although not a reliable absolute
measure of the age of each galaxy, these spectroscopic ages
do provide us with a much more useful estimate of their ages
relative to one another. In Fig. 2 we show LX/LB plotted
against spectroscopic age (from Terlevich & Forbes 2000).
The plot shows a large degree of scatter, most notably
in the age range between 4 and 10 Gyrs. This is likely to
be in part caused by the uncertainties in calculating ages,
which we estimate lead a typical error of ∼20%. We also
expect a mean 1-σ error in X -- ray luminosity of ∼15%. Typ-
ical error bounds for points at 1, 7 and 15 Gys are shown
in Fig. 2 by the diamonds at the top of the plot. The graph
also contains a relatively large number of upper limits, so
we have used the survival analysis packages available under
IRAF to assess the strength of any correlation and fit regres-
sion lines. The Cox proportional hazard, Spearman's rho and
generalized Kendall's tau tests are used to determine corre-
lation strength. Linear regression fitting is carried out using
the expectation and maximization (EM) algorithm and the
Buckley -- James (BJ) algorithm. In all cases we found that
the two fitting algorithms agreed closely.
Using the full sample of 77 early-type galaxies, we find
a correlation of at least 99.6% significance between LX /LB
and age. Line fits to the sample produce slopes of 0.066 ±
0.022 (EM) and 0.063 ± 0.025 (BJ). The former is shown as
the dashed line in Fig. 2.
Recent work on the X -- ray properties of galaxies in
groups (Helsdon et al. 2000) lead us to suspect that some
of the scatter seen in Fig. 2 might be caused by inclusion of
group or cluster dominant galaxies (marked as crossed cir-
cles) in our sample. It is also important to note that there is
evidence to suggest that the largest ellipticals may have non --
solar abundance ratios (e.g. Carollo et al. 1993). This could
lead to their ages being underestimated by a small amount,
again adding to the scatter. We therefore removed from our
sample a number of galaxies listed by Garcia (1993) as group
dominant or known cDs and retested the sample. This im-
proves the correlation slightly (to >99.75%) and makes the
line fit slightly shallower; 0.063 ± 0.019 (EM) and 0.061 ±
0.022 (BJ). The EM fit is plotted as a solid line in Fig. 2.
To confirm that the increase in LX /LB occurs across the
range of ages, rather than just the first few Gyrs, we also
binned the sample into subsets by age. The age range of each
subset was chosen so as to have roughly equal numbers of
detections in each bin. We then calculated a mean LX /LB
for each bin, using the Kaplan -- Meier estimator. The results
are shown in Table 1, and as large crosses on Fig. 2. These
show the same trend for increasing LX /LB with age, and
c(cid:13) 0000 RAS, MNRAS 000, 1 -- 8
X -- rays from Post -- Mergers
3
32
31
30
29
10
8
6
4
2
0
Figure 1. Normalized X -- ray luminosity versus fine structure parameter Σ for early -- type galaxies. Filled circles are detections, arrows
denote upper limits. Dynamically young galaxies (high Σ) have low LX /LB .
Age
(Gyrs)
A < 4
4 ≤ A < 6
6 ≤ A < 8
8 ≤ A < 10
A ≥ 10
No. of detections No. of upper limits
Log LX /LB
(erg s−1 L−1
B⊙)
9
6
9
9
8
12
11
4
6
4
29.56
29.83
29.93
30.15
30.32
error
±0.11
±0.15
±0.22
±0.14
±0.09
Table 1. Mean LX /LB values for 5 age bins, calculated using the Kaplan -- Meier estimator.
indicate that the trend is continuous across the age range
covered by the data.
The Kaplan -- Meier estimator (Feigelson & Nelson 1985)
includes upper limits by constructing a probability distribu-
tion function for the data in which the probability asso-
ciated with each upper limit is redistributed equally over
detected values which lie below the upper limit. So long as
there is no systematic difference between the detected and
undetected systems, this should be a reasonable procedure.
A problem arises when the lowest point in a given bin is an
upper limit (since there are no lower points over which to re-
distribute the corresponding probability). When this occurs,
the Kaplan -- Meier estimator treats the limit as a detection
and hence may be biased.
Of the five bins chosen, two (the youngest and second
oldest) have upper limits as their lowest values. For the sec-
ond oldest bin, the problem data value is only slightly lower
than a detected point, so the bias will be small. However,
the youngest bin has two upper limits as its lowest points,
one of which is considerably below the nearest detection. To
check for bias, we recalculated a mean log LX /LB of 29.66
for this bin excluding these points, compared to 29.56 pre-
c(cid:13) 0000 RAS, MNRAS 000, 1 -- 8
4
Ewan O'Sullivan et al.
Figure 2. Log (LX /LB ) plotted against spectroscopic age. Crossed circles represent group or cluster dominant galaxies, filled circles all
other detections. Arrows denote upper limits and the filled triangle represents NGC 7252, a young post -- merger. The three diamonds at
the top of the graph show typical age and LX /LB errors for galaxies of age 1, 7 and 15 Gys. The two lines are fits to the data, including
(dashed line) or excluding (solid line) the group and cluster dominant galaxies. The large crosses represent mean LX /LB for the five age
bins described in the text. The dotted line shows the expected increase in LX /LB caused by the decrease of LB with age. Normalisation
of this line is arbitrary, and the position shown can be considered to represent a "worst case" scenario.
viously. This indicates that although the youngest bin may
be slightly biased, it still supports the trend observed.
As a further check of the robustness of our conclusions,
we examined the LX /LB:Age relation using detected points
only, and find a correlation significant at ∼2.1-σ. The upper
limits also show a correlation with age (at ∼2.3-σ). These
trends cannot be selection effects in the data, since neither
source distance, nor X-ray exposure time are correlated with
age. They therefore point to a genuine trend in the data.
Galaxies with larger spectroscopic ages typically have higher
LX /LB than younger ones.
As the merger -- induced starburst fades, we would expect
the blue luminosity to decline with time. This will cause an
increase in LX /LB, even for constant LX , which will com-
bine with any trends in LX which are present. With this
is mind, we show in Fig. 2 the expected change due to a
fading starburst from stellar population models (Worthey
1994). Here we have crudely assumed that the progenitor
galaxies consisted of a solar metallicity, 15 Gyr old popula-
tion, and that the new stars created in the merger also have
solar metallicity. The relative mass ratios are 90% progeni-
tor stars and 10% new stars. The total B band luminosity
fading of this composite stellar population is shown in Fig.
2 (we assumed that the fading of the progenitor stars after
reaching 18 Gyrs old is insignificant). It can be seen that the
starburst fades rapidly in the first few Gyrs, roughly match-
ing the trend seen in LX /LB, but at later times there is
very little change, whereas the observed mean LX /LB value
continues to rise. So although the LX /LB evolution at early
times could be simply driven by the fading starburst, this
cannot account for the evolution at late times.
A further possibility to be checked is that our sample
might contain a trend in mean optical luminosity with age.
Since LX/LB has been widely reported (see however Hels-
don et al. (2000)) to rise with LB, this could lead to a con-
sequent correlation between age and LX /LB. To test this,
we show in Fig. 3 a plot of log LB with measured age. An
anti -- correlation between LB with age could produce the ob-
c(cid:13) 0000 RAS, MNRAS 000, 1 -- 8
served trend in LX /LB. Fig. 3 does not appear to show any
such trend, and in order to test for a correlation, we apply
Kendall's K test to the data. We find a K statistic (which is
unit normal distributed) of ∼0.44 (for 77 galaxies), and ex-
cluding group dominant galaxies reduces this to ∼0.33 (for
65 galaxies). This result indicates that there is no trend in
the data, even at the 1σ level. We therefore conclude that
our sample does not show an anti -- correlation of LB with
age, and that the LX /LB trend with age reflects an increas-
ing X -- ray luminosity.
2.2 LX /LB versus Fundamental Plane Residual
As further evidence of X -- ray evolution with age we have
compared LX/LB to residual from the Fundamental Plane
(FP). Work by Forbes et al. (1998) has shown that galaxy
age appears to be a "fourth parameter" affecting the position
of a galaxy relative to the FP plane. Galaxies below the
plane (negative residual) are generally young, while older
objects lie on or above the plane (positive residual). Fig. 4
shows 212 galaxies with Fundamental Plane residuals taken
from Prugniel & Simien (1996).
As with Figs. 1 and 2, this shows a trend of increasing
LX /LB with age. For the complete data set, the correla-
tion strength is >99.98%. However, only a small number of
points lie outside the range -- 0.5<FP residual<0.5. These are
unlikely to be representative of the general population, but
will have a strong influence on the statistical tests. Exclud-
ing them lowers the correlation strength to ∼99.95%, with
a slope of 1.39 ± 0.40 (EM). This line is shown in Fig. 4.
Again the trend shown in Fig. 4 is one of increasing X -- ray
luminosity, relative to the optical, as a galaxy evolves.
3 DISCUSSION
In the previous section we have used three age estimators to
show a strong correlation between normalized X -- ray lumi-
nosity and galaxy age. The X -- ray emission from early -- type
galaxies is known to be produced by sources of two main
types; discrete sources such as X -- ray binaries, and hot gas.
The contribution of discrete sources has been shown to be
important in low luminosity early-type galaxies (Fabbiano
et al. 1994; Irwin & Sarazin 1998; Sarazin et al. 2000) and
in late-type galaxies, where they are generally the dominant
source of X -- ray emission. On the other hand, most early --
type galaxies have a strong hot gas component, and massive
ellipticals are certainly dominated by hot gas emission (Mat-
sushita et al. 2000; Matsushita 2000). As the contribution
of hot gas is known to vary a great deal, while the contri-
bution from discrete sources is generally believed to scale
with LB (Matsushita et al. 2000; Fabbiano et al. 1992) , it
seems likely that the trend we observe in LX /LB is caused
by changes in the hot gas content of these objects with age.
We now go on to discuss possible hot halo formation mecha-
nisms which may be responsible for this trend of increasing
X -- ray gas content with age.
3.1 Gas infall
It has been suggested (Hibbard & van Gorkom 1996) that
the hot gas in elliptical galaxies may be produced during
c(cid:13) 0000 RAS, MNRAS 000, 1 -- 8
X -- rays from Post -- Mergers
5
a merger by the shock heating or photoionization of cool
gas from the progenitor galaxies. In some ongoing mergers
the tidal tails are thought to contain up to half the Hi gas
originally present in the progenitor galaxies (Hibbard et al.
1994). When this gas falls back into the body of the galaxy,
shock heating should be capable of heating it to X -- ray tem-
peratures. An alternative is that the temperature of the gas
is caused by heating in the starburst phase of the merger. In
this case the hot gas would be blown out to large radii, but
might be contained by the dark matter halo of the galaxy
(Mathews & Brighenti 1998). At these large radii it would
be too diffuse to be detected, but would eventually fall back
into the galaxy, forming the observed X -- ray halo.
In terms of available masses of gas, both these mod-
els appear to be viable formation processes. Bregman et al.
(1992) derived X -- ray and cold gas masses for a large sample
of early -- type galaxies, finding MX of between ∼ 106 and
1011 M⊙. The mean X -- ray gas mass was a few 109 M⊙.
On the other hand, very few detections of Hi were made
in early -- type galaxies, although Sa galaxies were found to
contain between 107 and ∼ 5×1010 M⊙ of Hi . Studies of Hi
masses in later -- type galaxies (Huchtmeier & Richter 1989;
Huchtmeier & Richter 1988) give an average of a few 109
M⊙, depending on the range of morphologies chosen. Given
that large elliptical galaxies are likely to have been formed by
the merger of many smaller spiral galaxies, these results sug-
gest that fairly modest conversion efficiencies could produce
the expected X -- ray halo gas masses either from infalling Hi
or via starburst heating and later infall.
However, the infalling Hi model fails to explain the
timescale over which we observe the LX /LB trend. In the
well known merging galaxy NGC 7252, 50% of the cold gas
currently in the tidal tails is expected to fall back into the
main body of the galaxy within the next 2 -- 3 Gyr (Hibbard
& van Gorkom 1996). This suggests that we should expect
to see a rapid build up of X -- ray emission in the first few
gigayears after merger. This increase should be made even
more noticeable by the fading of the stellar population, as
this is the period during which LB is dropping most quickly.
Rapid generation of the X -- ray halo is inconsistent with our
results.
A model involving infall of hot gas is likely to have sim-
ilar problems. It is difficult to see how gas falling back into
the body of the galaxy can do so at the steady rate needed to
produce a smooth increase in LX /LB. Infall may be delayed,
as the hot gas reaches larger radii than the cold, but it will
still occur on a galaxy dynamical timescale. Models involv-
ing cooling and infall of gas onto central dominant galaxies
in groups or clusters (Brighenti & Mathews 1999) may be
able to produce an inflow over a longer period, but these are
unlikely to be generally applicable. Group dominant galax-
ies have exceptionally large dark halos, making them more
able to retain (or accrete) hot gas. They also lie at the bot-
tom of a group potential well, and may be at the centre of
a group X -- ray halo, providing them with an extra reservoir
of hot gas. These conditions do not apply to the majority of
early-type galaxies. Therefore we suggest that infall of (cold
or hot) gas is unlikely to be the dominant process in the
generation of X -- ray halos in typical ellipticals.
6
Ewan O'Sullivan et al.
11
10.5
10
9.5
9
0
5
10
15
Figure 3. Optical luminosity versus spectroscopic age. Crossed circles represent cluster or group dominant galaxies and filled circles
all other detections. The triangle is the recent merger remnant NGC 7252. The cE galaxy, M32, has been excluded from the plot and
associated statistical tests.
3.2 Ongoing stellar mass -- loss and galaxy winds
Since infall of either hot or cold gas seems unable to provide
the sort of long term trend in LX/LB seen in Fig. 2, we now
look to models involving gas generated by mass loss from the
galactic stars. This source of gas is fairly well understood,
and can provide the sort of gas masses required to account
for the halos of many early-type galaxies (though proba-
bly not the very brightest -- Brighenti & Mathews 1998),
provided that the gas is retained within the galactic poten-
tial. Two main timescales are involved. One is the mass loss
rate from stars, primarily giant stars. The second is the rate
of type Ia supernovae (SNIa) which provide the main heat
source (after the brief SNII phase). These two factors, and
the interplay between them, can potentially lead to changes
in the hot gas content of elliptical galaxies on timescales
much longer than a dynamical time, and so have the poten-
tial to explain the slow trend we observe. The stellar mass
loss rate is fairly well understood (e.g. Ciotti et al. 1991), and
for a single-aged stellar population it declines approximately
as t−1.3. However, the SNIa heating rate, which dominates
the effective specific energy of the injected gas, is highly un-
certain. Recent estimates of the SNIa rate in old stellar pop-
ulations (e.g. Cappellaro et al. 1999) have revised the classic
Tammann (1982) value of the rate in old stellar populations
down by a factor 3-4, and the evolution of this rate with
population age is very model-dependent. Given our contin-
uing ignorance about the precise nature of SNIa, it must
therefore be regarded as largely unknown.
The specific energy of the injected gas determines
whether gas is retained in a hydrostatic hot halo, or escapes
the galaxy as a wind. The way in which the specific energy
evolves, depends upon the evolution of the SNIa rate rela-
tive to the stellar wind loss rate. There are therefore two
fundamentally different classes of models: those in which
the SNIa rate is constant, or declines more slowly than the
stellar mass loss rate (e.g. Loewenstein & Mathews 1987;
David et al. 1991) and those in which the supernova rate
drops more quickly than the gas injection rate (Ciotti et al.
1991; Pelegrini & Ciotti 1998). Broadly speaking, when the
specific energy of the injected gas is smaller than its grav-
itational binding energy it will be retained in a hot halo.
It may subsequently cool in a luminous cooling flow. How-
ever, if the specific energy of the gas substantially exceeds
its binding energy it will escape from the galaxy in a fast
c(cid:13) 0000 RAS, MNRAS 000, 1 -- 8
X -- rays from Post -- Mergers
7
32
31
30
29
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
Figure 4. X -- ray luminosity versus Fundamental Plane residual for early -- type galaxies. Filled circles represent detections and arrows
upper limits. The solid line is a fit to all points with -- 0.5<FP residual<0.5
wind, which results in a much lower X -- ray luminosity. The
first class of models, in which the specific energy rises with
time, therefore produces an evolution from an early inflow
phase to a later wind phase, predicting a declining X-ray
luminosity with time. Such models are clearly inconsistent
with our results.
Models in which the SNIa rate drops more quickly than
t−1.3 evolve, in broad terms, from a low luminosity wind
phase, towards a more luminous halo/cooling flow phase.
The timescale for this evolution depends upon a variety of
factors, such as the depth and shape of the potential, but it
is typically ∼ 10 Gyr (Ciotti et al. 1991). In fact, none of
the models studied to date shows the rather simple mono-
tonic rise in LX with age which we observe. The models of
Ciotti et al. (1991) predict an initial luminous phase, when
the gas loss rate from stars is very high. The X -- ray luminos-
ity then declines as the wind loss rate drops, and then rises
again during the transition to a bound hydrostatic halo, at
which point the X-ray luminosity is essentially equal to the
SNIa luminosity. This whole development describes a fairly
symmetrical dip and rise, lasting ∼ 5 − 15 Gyr, which is
again not what we see in the data. However, this model de-
scribes a galaxy in which all stars are formed at t = 0. In
c(cid:13) 0000 RAS, MNRAS 000, 1 -- 8
contrast, the spectroscopic age of our galaxies probably de-
notes the time since a merger-induced starburst involving
only a few percent of the stellar mass. In this case, the early
stellar mass loss rate will be much lower than in the Ciotti
et al models, and the X-ray luminosity correspondingly less
(scaling approximately as the square of the mass loss rate).
The main change in LX will be a rise associated with the
slowing wind and transformation into a hydrostatic halo. In
addition, as shown in Fig. 2, the decrease in optical lumi-
nosity as the starburst population fades will produce a rise
in LX /LB over the first 1-2 Gyr.
4 CONCLUDING REMARKS
We have examined the evolution of the X -- ray properties
of early -- type galaxies. For three galaxy age estimators (fine
structure parameter, Fundamental Plane residuals and spec-
troscopic ages) we find that the normalized X -- ray luminos-
ity evolves with time. In particular, the X -- ray luminosity,
which reflects the mass of hot halo gas, appears to increase
at a steady rate over ∼ 10 Gyrs.
Comparing the long term trend in LX/LB which we
8
Ewan O'Sullivan et al.
observe with expectations from possible mechanisms for hot
halo formation, we conclude that the only viable mechanism
appears to be the slow evolution from an outflowing wind
to hydrostatic halo phase driven by a declining SNIa rate.
Infalling gas seems unlikely to be the main cause of such
a long term trend. Our results suggest that some of the
scatter seen in the global LX versus LB relation is due to
the evolutionary state, and past merger history, of early --
type galaxies.
Hibbard J. E., Guhathakurta P., van Gorkom J. H.,
Schweizer F., 1994, AJ, 107, 67
Huchtmeier W. K., Richter O.-G., 1988, A&A, 203, 237
Huchtmeier W. K., Richter O.-G., 1989, A&A, 210, 1
Irwin J., Sarazin C., 1998, ApJ, 494, L33
Loewenstein M., Mathews W. G., 1987, ApJ, 319, 614
Acknowledgements
We would like to thank R. Brown for early work on this
project and S. Helsdon, W. Mathews and A. Renzini for
useful discussions. We also thank an anonymous referee for
several suggestions which have improved the paper. This
research has made use of the NASA/IPAC Extragalactic
Database (NED) which is operated by the Jet Propulsion
Laboratory, California Institute of Technology, under con-
tract with the National Aeronautics and Space Adminis-
tration. E. O'S. acknowledges the receipt of a PPARC stu-
dentship.
REFERENCES
Mackie G., Fabbiano G.,
in Arnaboldi M.,
Da Costa G. S., Saha P., eds, ASP Conf. Ser. 116:
The Nature of Elliptical Galaxies; 2nd Stromlo Sym-
posium. p. 401
1997,
Mathews W. G., Brighenti F., 1998, in ASP Conf. Ser. 136:
Galactic Halos. p. 277
Matsushita K., Ohashi T., Makishima K., 2000, PASJ, 52,
685
Matsushita K., 2000, preprint, astro-ph/0009276
O'Sullivan E., Forbes D. A., Ponman T. J., 2000, in prepa-
ration
Pelegrini S., Ciotti L., 1998, A&A, 333, 433
Bender R., Surma P., Dobereiner S., Mollenhoff C., Made-
Prugniel P., Simien F., 1996, A&A, 309, 749
jsky R., 1989, A&A, 217, 35
Bregman J. N., Hogg D. E., Roberts M. S., 1992, ApJ, 387,
484
Brighenti F., Mathews W. G., 1998, ApJ, 495, 239
Brighenti F., Mathews W. G., 1999, ApJ, 512, 65
Read A. M., Ponman T. J., 1998, MNRAS, 297, 143
Sansom A. E., Hibbard J., Schweizer F., 2000, AJ, 120, 1946
Sarazin C. L., Irwin J. A., Bregman J. N., 2000, preprint,
astro-ph/0009448
Schweizer F., Seitzer P., 1992, AJ, 104, 1039
Cappellaro E., Evans R., Turatto M., 1999, A&A, 351, 459
Carollo C. M., Danziger I. J., Buson L., 1993, MNRAS, 265,
553
Tammann G. A. in Rees M., Stoneham R., eds, Supernovae:
A Survey of Current Research, Dordrecht: Reidel,
1982
Ciotti L., D'Ercole A., Pelegrini S., Renzini A., 1991, ApJ,
Terlevich A. I., Forbes D. A., 2000, MNRAS, submitted
376, 380
Worthey G., Faber S. M., Gonzalez J. J., Burstein D., 1994,
David L. P., Forman W., Jones C., 1991, ApJ, 369, 121
ApJS, 94, 687
Fabbiano G., Schweizer F., 1995, ApJ, 447, 572
Fabbiano G., Kim D. W., Trinchieri G., 1992, ApJS, 80, 531
Fabbiano G., Kim D.-W., Trinchieri G., 1994, ApJ, 429, 94
Feigelson E. D., Nelson P. I., 1985, ApJ, 293, 192
Forbes D., Ponman T., Brown R., 1998, ApJ, 508, L43
Garcia A. M., 1993, A&AS, 100, 47
Georgakakis A., Forbes D. A., Norris R. P., 2000, MNRAS,
318, 124
Helsdon S. F., Ponman T. J., O'Sullivan E., Forbes D. A.,
2000, MNRAS, submitted
Hibbard J. E., van Gorkom J. H., 1996, AJ, 111, 655
c(cid:13) 0000 RAS, MNRAS 000, 1 -- 8
|
astro-ph/0209100 | 1 | 0209 | 2002-09-05T14:42:55 | Bose-Einstein condensation in dense nuclear matter and strong magnetic fields | [
"astro-ph",
"nucl-th"
] | Bose-Einstein condensation of antikaons in cold and dense beta-equilibrated matter under the influence of strong magnetic fields is studied within a relativistic mean field model. For magnetic fields $> 5 \times 10^{18}$G, the phase spaces of charged particles are modified resulting in compositional changes in the system. The threshold density of $K^-$ condensation is shifted to higher density compared with the field free case. In the presence of strong fields, the equation of state becomes stiffer than that of the zero field case. | astro-ph | astro-ph | Bose-Einstein condensation in dense nuclear matter and strong
magnetic fields
Prantick Dey(a), Abhijit Bhattacharyya(b) and Debades Bandyopadhyay(c)
(a)Physics Department, Raja Peary Mohan College, Uttarpara, Hooghly 712258, India
(b)Physics Department, Scottish Church College, 1 & 3 Urquhart Square, Calcutta 700006, India
(c)Saha Institute of Nuclear Physics, 1/AF Bidhannagar, Calcutta 700 064, India
Abstract
Bose-Einstein condensation of antikaons in cold and dense beta-equilibrated
matter under the influence of strong magnetic fields is studied within a rela-
tivistic mean field model. For magnetic fields > 5 × 1018G, the phase spaces
of charged particles are modified resulting in compositional changes in the
system. The threshold density of K − condensation is shifted to higher den-
sity compared with the field free case. In the presence of strong fields, the
equation of state becomes stiffer than that of the zero field case.
PACS: 03.75.Fi, 26.60.+c, 21.65.+f
2
0
0
2
p
e
S
5
1
v
0
0
1
9
0
2
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
1
Recently, it has been inferred that some soft gamma ray repeaters (SGRs) and perhaps
certain anomalous X-ray pulsars (AXPs) could be neutron stars having large magnetic fields
∼ 1014 − 1016 G [1]. Those objects are called "magnetars" [2]. Earlier large magnetic fields
∼ 1013G were estimated to be associated with the surfaces of some radio pulsars [3]. The
origin of such ultra strong magnetic fields is still an unsolved problem. An attractive idea
about the origin is that the small magnetic field of a progenitor star is amplified due to
the magnetic flux conservation during the gravitational collapse of the star [3]. Recently,
Thomson and Duncan argued that a convective dynamo mechanism might result in large
fields ∼ 1015G [4]. On the other hand, it is presumed from the scalar virial theorem [5]
based on Newtonian gravity that the limiting interior field in neutron stars could be as large
as ∼ 1018G [6]. From the general relativistic calculation of axis-symmetric neutron stars in
magnetic fields, it follows that neutron stars could sustain magnetic fields ∼ 1018G [7,8].
Because of highly conducting core, such large interior fields may be frozen and could not
be directly accessible to observation. Its effects may be manifested in various observables
such as the mass-radius relationship, neutrino emissivity etc. Motivated by the existence of
large fields in the core of neutron stars, its influence on the gross properties of neutron stars
was studied by various groups [6,9,10,11]. The calculations in the relativistic mean field
(RMF) approach showed that the equation of state (EoS) was modified due to the Landau
quantization and also by the interaction of magnetic moments of baryons with the field [11].
The intense magnetic field was found to change the composition of beta equilibrated matter
relevant to neutron stars drastically [9,10,11]. The neutrino emissivity in neutron stars was
reported to be enhanced in strong magnetic fields [12].
Besides strong interior fields, many exotic forms of matter may exist in the dense core
of neutron stars. One such possibility is the appearance of the Bose-Einstein condensate of
strange particles. Nelson and Kaplan first pointed out that antikaons may undergo the Bose-
Einstein condensation (BEC) in dense matter at zero temperature because of the attractive
s-wave antikaon-nucleon interaction [13]. Later, this idea was applied to neutron stars by
various authors [14,15,16]. Bose-Einstein condensation in a magnetic field is an old and
interesting problem in other branches of physics also namely condensed matter physics and
statistical physics. It was shown by Schafroth [17] that a non-relativistic Bose gas could not
condense in an external magnetic field. There are some calculations on the condensation
of relativistic charged Bose gas in magnetic fields in the literature [18,19]. Elmfors and
collaborators [18] noted that the relativistic Bose gas might condense for spatial dimension
d ≥ 5. They showed that the number density of bosons in the ground state diverges for
d < 5 in the presence of a magnetic field. On the other hand, it was argued [19] that the
condensation of bosons in a magnetic field could occur in three dimension if the chemical
potential of bosons was taken as a function of density, temperature and magnetic field [19].
In this case, the BEC would be a diffuse one because there is no definite critical temperature.
It was also shown in the latter calculation [19] that the number density of bosons in the
ground state was finite. Recently, Suh and Mathews [10] have studied pion condensation in
a beta equilibrated non-interacting n-p-e-µ system in magnetic fields.
In this paper, we investigate the influence of strong magnetic fields on the Bose-Einstein
condensation of antikaons in cold and dense matter relevant to neutron stars. This may have
profound implications on the gross properties of neutron stars. This method of studying the
BEC in strong magnetic fields and dense matter is rather general; therefore it should be of
2
correspondingly broad interest.
We consider strong magnetic field effects on antikaon condensation in the beta equili-
brated neutron star matter composed of neutrons, protons, electrons, muons and K − mesons
within the framework of a relativistic field theoretical model [20]. As the constituents in
neutron stars are highly degenerate, the chemical potentials of baryons are larger than the
temperature of the system. Therefore, the gross properties of neutron stars are calculated
at zero temperature. The total Lagrangian density may be written as the sum of baryonic,
kaonic and leptonic parts i.e. L = LB + LK + Ll. In a uniform magnetic field, the baryonic
Lagrangian density [21] is given by
¯ψB (iγµDµ − mB + gσBσ − gωBγµωµ − gρBγµtB · ρµ − κBσµνF µν) ψB
LB = XB=n,p
1
+
2 (cid:16)∂µσ∂µσ − m2
− Xk=ω,ρ(cid:20) 1
4 (cid:16)∂µV k
σσ2(cid:17) − U(σ)
µ(cid:17)2
ν − ∂νV k
−
1
2
m2
k(V k
µ )2(cid:21) +
1
4
g4 (ωµωµ)2 −
1
4
F µνFµν .
(1)
Here ψB denotes the Dirac spinor for baryon B with vacuum mass mB and isospin operator
tB. The scalar self-interaction term [22] is, U(σ) = g2σ3/3 + g3σ4/4. Following Ref. [11],
the interaction of anomalous magnetic moments of baryons with magnetic fields is given
by the last term under the summation in Eq.(1). Here, F µν is the electromagnetic field
tensor, σµν = [γµ, γν]/2 and κB is the experimentally measured value of magnetic moment
for baryon B. The covariant derivative for a charged particle is Dµ = ∂µ + iqAµ with the
choice of gauge corresponding to the constant magnetic field (Bm) along z-axis is A0 = 0,
A ≡ (0, xBm, 0). The form of 4-component spinor solutions for baryons is given by Ref. [11].
The (anti)kaon-nucleon interaction is treated in the same footing as that of the nucleon-
nucleon interaction [15]. Therefore, the kaonic Lagrangian density in a magnetic field is
given as,
K K ∗K ,
LK = D∗
µK ∗DµK − m∗2
(2)
where the covariant derivative Dµ = ∂µ + iqAµ + igωKωµ + igρK tK · ρµ. There is
no interaction term involving magnetic moments in the kaonic Lagrangian density be-
cause (anti)kaons having zero spin angular momentum do not possess magnetic mo-
ments. The effective mass of (anti)kaons in this minimal coupling scheme is given by
m∗
K = mK − gσKσ. The solution for negatively charged kaons in a magnetic field is
K ∝ (qBm/π)1/4 (1/√2nn!)e−iωK− t+ipyy+ipzze−qBmη2/2Hn(√qBmη), where η = x + py/qBm,
"H" denotes the Hermite polynomial with n the Landau principal quantum number. The
Lagrangian density for neutrons is obtained by putting q = 0 in the covariant derivatives of
Eq. (1). In the mean field approximation [20], the meson field equations in the presence of
antikaon condensate and magnetic field are
m2
ωω0 + g4ω3
m2
σσ = −
0 = XB
ρρ03 = XB
m2
m∗
K
K + qBm
∂U
∂σ
gσBnS
B + gσK
+XB
gωBnB − gωKnK − ,
gρBI3BnB + gρKI3K −nK − .
qm∗2
nK − ,
(3)
(4)
(5)
3
where nB and ns
B are baryon and scalar density for baryon B respectively; I3B = +1/2 for
protons, −1/2 for neutrons and I3K − = −1/2 for K − mesons. The expressions of the scalar
and baryon density corresponding to protons are given by [11]
p = qpBm
ns
2π2 Xν Xs
m∗
p
mp
mp − sκpBm
ln (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)
and
Ep
f + kp
mp
f,ν,s
! ,
(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)
np = qpBm
2π2 Xν Xs
kp
f,ν,s ,
where the energy spectrum for protons is given by
Ep,ν,s = rk2
z +(cid:16)qm∗ 2
p + 2νqpBm + sκpBm(cid:17)2
+ gωpω0 +
1
2
gρpρ03 ,
and
mp = qm∗ 2
p + 2νqpBm + sκpBm ,
Similarly for neutrons, those expressions are given by [11]
kp
f,ν,s = rEp 2
ns
n =
m∗
n
4π2 Xs
nn =
1
2π2 Xs
1
3
k3
f,s +
1
2
.
f −(cid:16)qm∗ 2
kf,sEn
p + 2νqpBm + sκpBm(cid:17)2
f − m2 ln (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)
m (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)
f arcsin
sκnBm"mkf,s + En 2
f + kf,s
! ,
En
m
f −
En
π
2!# ,
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
and
where
and
En,s = rk2
z +(cid:16)qm∗ 2
n + k2
x + k2
y + sκnBm(cid:17)2
+ gωnω0 −
1
2
gρnρ03 ,
m = m∗
n + sκnBm ,
kf,s = qEn 2
f − m2 .
magnetic field is obtained as ωK − = qp2
Solving the equation of motion for antikaons, the in-medium energy of K − meson in a
K + qBm(2n + 1) − gωKω0 − gρKρ03/2. The
condition for the condensation of K − meson in a magnetic field is pz = 0 and n = 0.
The number density of K − meson in a magnetic field and in the ground state is obtained
z + m∗2
4
from the relation J K
2(ωK − + gωKω0 + gρKρ03/2)K ∗K. The total energy density is given by
µ = i(K ∗∂L/∂µK ∗ − ∂L/∂µK K) and it is given by, nK − = −J K −
0 =
ε =
1
2
m2
σσ2 +
+ XB=n,p
εB +Xl
1
3
g2σ3 +
1
4
g3σ4 +
1
2
m2
ωω2
0 +
3
4
g4ω4
0 +
1
2
m2
ρρ2
03
εl + ε ¯K ,
(16)
where εB and εl correspond to the kinetic energy densities of baryons and leptons respec-
tively. The kinetic energy densities of protons and neutrons in a magnetic field are given by
[11]
εp = qpBm
4π2 Xν Xs
and
Ep
f kp
f,ν,s + mp
εn =
1
1
2
4π2 Xs
+ (cid:18)1
sκnBm −
3
En 3
f kf,s +
sκnBmEn 3
2
3
1
4
m(cid:19)"mkf,sEn
Ep
2 ln (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)
f arcsin
f + m3 ln (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)
En
f + kp
mp
f,ν,s
! ,
(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)
π
m
f −
En
f + kf,s
2!
m (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)
!# .
(17)
(18)
Similarly, the expression for the kinetic energy density of electrons has the same form as that
of protons but electrons are noninteracting and anomalous magnetic moment of electrons
is not considered here [11]. The energy density for antikaons in the condensate state is
ε ¯K = qm∗2
K + qBm nK −. The other terms in Eq. (16) represent interaction energy densities.
The pressure of the system follows from the relation P = µnnb − ε, where µn and nb are the
neutron chemical potential and total baryon density, respectively. In the core of neutron
stars, strangeness changing processes such as n ⇀↽ p + K − and e− ⇀↽ K − + νe occur. The
chemical equilibrium yields µn − µp = µK − = µe, where µp and µK − are respectively the
chemical potentials of protons and K − mesons. Employing Eq. (3) in conjunction with the
chemical equilibrium conditions and charge neutrality np − nK − − ne − nµ = 0, we obtain
the effective masses self-consistently.
In the effective field theoretical approach adopted here, two different sets of coupling
constants for nucleons and kaons with σ, ω and ρ meson are required. The nucleon-meson
coupling constants are obtained by fitting experimental data for binding energies and charge
radii for heavy nuclei [21]. This set of parameters is known as TM1 set. The values of
coupling constants are gσN = 10.0289, gωN = 12.6139, gρN = 4.6322, g2 = −7.2325f m−1,
g3 = 0.6183 and g4 = 71.3075. The incompressibility of matter at normal nuclear matter
density (n0 = 0.145f m−3) is 281 MeV for the TM1 model. According to the simple quark
model and isospin counting rule, the kaon-vector meson coupling constants are gωK = 1
3gωN
and gρK = gρN . On the other hand, the scalar coupling constant is obtained from the
real part of the antikaon optical potential at normal nuclear matter density i.e. U ¯K(n0) =
−gσKσ − gωKω0 [16]. In this calculation, we have taken U ¯K(n0) = −160 MeV and the scalar
coupling is gσK = 2.0098.
The TM1 model was adopted earlier for the description of heavy nuclei and the equation
of state for neutron stars [21]. Besides the non-linear σ meson terms, the model also includes
5
c are denoted by the dashed lines. The critical field for electrons (Be
non-linear ω meson term. It was shown [21] that the TM1 model reproduced scalar and
vector potentials close to those of the relativistic Brueckner Hartree Fock calculation using
the realistic nucleon-nucleon interaction [23]. Recently, the TM1 model was used for the
investigation of antikaon condensation in neutron star matter [16] for zero field. For TM1
parameter set, it was found that the phase transition was of second order [15,16]. In the TM1
model, the maximum masses and central densities of neutron stars without and with antikaon
condensation where UK − = −160 MeV are respectively, 2.179(1.857)M⊙ and 5.97(6.37)n0.
In Figure 1, number densities of various particles are plotted with baryon density. The
particle densities for Bm = 0 are shown by the solid lines, whereas those corresponding to
Bm = 1.5 × 105Be
c ) is
that value where cyclotron quantum is equal to or above the rest energy of an electron and
its value is Be
c = 4.414 × 1013G. Here we note that the formation of K − condensation is
delayed to higher density than the field free case. The threshold densities of K − condensation
corresponding to Bm = 0 and Bm = 1.5 × 105Be
c are 2.67n0 and 3.85n0 respectively. The
delayed appearance of K − condensation may be attributed to the stiffer EoS because of the
effects of magnetic moments. In the presence of the field, the enhancement of electron and
muon fraction are pronounced whereas the proton fraction is smaller than the zero field value
beyond 2.7n0. With the appearance of K − condensate, it would try to diminish electron and
muon density. On the other hand, the phase spaces of electrons and muons are so strongly
modified in a quantizing field that their fractions are significantly increased. The net result
is the reduction in the density of K − condensate than that of the field free case. The proton
density increases after the onset of K − condensation. The neutron fraction also increases
because of the interaction of anomalous magnetic moment of neutrons with the field. This
may have important effects on the equation of state.
In the presence of magnetic fields > 5 × 1018G, the nucleon effective mass is enhanced
in the high density regime than that of the field free case. This may be attributed to the
effects of magnetic moments as it was also noted in Ref. [11]. The (anti)kaon effective mass
in magnetic fields does not change appreciably from the zero field case.
The onset of K − condensation is given by the equality of K − chemical potential (µK −)
with electron chemical potential (µe). In the presence of magnetic field, we find the hadronic
phase smoothly connects to the antikaon condensate phase resulting in a second order phase
transition as it is evident from equation of state (pressure versus energy density curve) in
Figure 2. For TM1 parameter set, we note that the phase transition is of second order with
and without magnetic field.
c (curve II) and Bm = 1.5× 105Be
In Figure 2, matter pressure (P ) versus matter energy density (ǫ) is displayed for Bm = 0
c (curve III). For Bm = 4× 104Be
(curve I), Bm = 4× 104Be
c ,
we note that the curve becomes slightly stiffer with the onset of K − condensation. This
stiffening may be attributed to the large enhancement in electron and muon fraction in the
field. However, this effect is reduced in the high density regime where electron and muon
fraction become small. As the field is further increased to Bm = 1.5 × 105Be
c , not only
electrons and muons are strongly Landau quantized, but also protons are populated in the
zeroth Landau level. It was shown [9,11] that Landau quantization of charged particles was
responsible for the softening in the equation of state. On the other hand, the effects of baryon
magnetic moments for Bm = 1.5 × 105Be
c overwhelm the effects of Landau quantization.
Consequently, the curve corresponding to Bm = 1.5× 105Be
c stiffens further. It is found here
6
m/(8π) = 4.814 × 10−8(Bm/Be
that the effects of magnetic moments are important for Bm > 105Be
c . Besides the effects
of Landau quantization and magnetic moments, the contribution of electromagnetic field to
the matter energy density and pressure is to be taken into account. The magnetic energy
density and pressure, εf = Pf = B2
c )2MeV f m−3, become
significant in the core of the star for Bm ≥ 105Be
c .
In this calculation, we have considered interior magnetic field > 5 × 1018G. However, it
was found in a recent calculation [24] that the maximum value of the magnetic field within
a star may not exceed 3 × 1018G for a particular choice of a constant current function but
independent of an EoS. In this case, the ratio of the maximum field to the average field is not
large because of small spatial gradient. The authors [24] argued that the value of maximum
field at any point may well exceed the average value as mentioned above for a different field
geometry. In that event the effects of strong magnetic field > 5 × 1018G on the threshold of
antikaon condensation, particle composition and EoS might be important.
To summarise, in this paper, we have focused on the formation of the antikaon conden-
sation in dense nuclear matter in the presence of magnetic fields. We have considered the
interaction of magnetic moments of baryons with the field and the magnetic energy density
and pressure in this work. In the presence of strong magnetic fields > 5 × 1018G, we find a
considerable change in the phase spaces of charged particles. The threshold density of K −
condensation is delayed to higher density in the presence of such a strong field and the EoS
becomes stiffer. For Bm > 1018G, the effects of magnetic moments are important and it adds
to further stiffening of the equation of state. Also, the electromagnetic field contribution
to the energy density and pressure becomes important in the core for Bm > 1018G. The
stiffening of the EoS in the presence of magnetic fields might have significant impact on
the gross properties of neutron stars such as the mass-radius relationship, cooling etc. It
is worth mentioning here that (anti)kaons do not interact with magnetic fields in the same
way as fermions do because their spin angular momentum is zero.
In this calculation, we do not include the role of hyperons, pion condensation and nucleon-
nucleon correlation on the antikaon condensation. Negatively charged hyperons, in particular
Σ− hyperon, could delay the onset of K − condensation [14]. However, it was estimated that
Σ−-nucleon interaction is highly repulsive in normal nuclear matter [25]. Recently, it has
been also shown that threshold densities of most hyperons including Σ− are substantially
increased in strong magnetic field Bm > 5 × 1018G both due to Landau quantisation and
magnetic moment interactions [24]. In this situation, Σ− hyperons might have no impact on
K − condensation. Pion condensation could occur in neutron stars because of the attractive
p-wave pion-nucleon interaction [26]. The condensation of π− may modify the electron
chemical potential which, in turn, would delay K − condensation. In this paper, we have
employed the RMF model which does not include nucleon-nucleon correlations.
It was
shown in non relativistic models [27] that nucleon-nucleon correlations shifted the threshold
density of K − condensation to higher density. We believe that the qualitative features of
strong magnetic fields presented here would survive even in other models which include
hyperons, pion condensation and nucleon-nucleon correlations. It would be interesting to
look into the neutrino emissivity from an antikaon condensed matter and the structure of
compact stars having antikaon condensate in the presence of a strong magnetic field.
7
REFERENCES
[1] Kouveliotou C, Dieter S, Strohmayer T, Paradijs J van, Fishman G J, Meegan C A and
Hurley K 1998 Nature 393 235; Kouveliotou C, Strohmayer T, Hurley K, Paradijs J van,
Finger M H, Dieter S, Woods P, Thompson C and Duncan R C 1999 Astrophys. J. 510
L115; Heyl J S and Kulkarni S R, 1998 Astrophys. J. 506 L61; Chatterjee P, Hernquist
L and Narayan R, 2000 Astrophys. J. 534 373.
[2] Duncan R C and Thompson C 1992 Astrophys. J. 392 L9; Thompson C and Duncan R
C 1995 MNRAS 275 255; Thompson C and Duncan R C 1996 Astrophys. J. 473 322;
Usov V V 1992 Nature 357 472; Paczynski B 1992 Acta. Astron. 42 145; Vasisht G and
Gotthelf E V 1997 Astrophys. J. 486 L129.
[3] Chanmugam G 1992 Annu. Rev. Astron. Astrophys. 30 143.
[4] Thompson C and Duncan R C 1993 Astrophys. J. 498 194.
[5] Chandrasekhar S and Fermi E 1953 Astrophys. J. 118 116.
[6] Lai D and Shapiro S L 1991 Astrophys. J. 383 745.
[7] Bocquet M, Bonazzola S, Gourgoulhon E and Novak J 1995 Astron. Astrophys.301 301.
[8] Cardall C Y, Prakash M and Lattimer J M 2001 Astrophys. J. 554 322.
[9] Chakrabarty S, Bandyopadhyay D and Pal S 1997 Phys. Rev. Lett. 78 2898; Bandyopad-
hyay D, Chakrabarty S and Pal S 1997 Phys. Rev. Lett. 79 2176.
[10] Suh I -S and Mathews G J 2001 Astrophys. J. 546 1126.
[11] Broderick A, Prakash M and Lattimer J M 2000 Astrophys. J. 537 351.
[12] Leinson L B and Perez A 1998 JHEP 09 020; Bandyopadhyay D, Chakrabarty S, Dey
P and Pal S 1998 Phys. Rev. D58 121301; Baiko D A and Yakovlev D G 1999 Astron.
Astrophys. 342 192; Dalen E N E van, Dieperink A E L, Sedrakian A and Timmermans
R G E 2000 Astron. Astrophys. 360 549.
[13] Kaplan D B and Nelson A E 1986 Phys. Lett. B175 57; Nelson A E and Kaplan D B
1987 Phys. Lett. B192 193.
[14] Prakash M, Bombaci I, Prakash M, Ellis P J, Lattimer J M and Knorren R 1997 Phys.
Rep. 280 1.
[15] Glendenning N K and Schaffner-Bielich J 1998 Phys. Rev. Lett. 81 4564; Glendenning N
K and Schaffner-Bielich J 1999 Phys. Rev. C60 025803.
[16] Pal S, Bandyopadhyay D and Greiner W 2000 Nucl. Phys. A674 553; Banik S and Bandy-
opadhyay D 2001 Phys. Rev. C63 035802.
[17] Schafroth M R 1955 Phys. Rev. 100 463.
[18] Elmfors P et al. 1995 Phys. Lett. B348 462 and references therein.
[19] Perez R H 1996 Phys. Lett. B379 148 and references therein.
[20] Serot B D and Walecka J D 1986 Adv. Nucl. Phys. 16 1.
[21] Sugahara Y and Toki H 1994 Nucl. Phys. A579 557.
[22] Boguta J and Bodmer A R 1977 Nucl. Phys. A292 413.
[23] Brockmann R and Machleidt R 1990 Phys. Rev. C42 1965.
[24] Broderick A, Prakash M and Lattimer J M 2001 astro-ph/0111516
[25] Friedman E, Gal A, Mares J and Cieply A 1999 Phys. Rev. C60 024314.
[26] Akmal A, Pandharipande V R and Ravenhall D G 1998 Phys. Rev. C58 1804.
[27] Pandharipande V R, Pethick C J, Thorsson V 1995 Phys. Rev. Lett. 75 4567; Carlson J,
Heiselberg H and Pandharipande V R 2001 Phys. Rev. C63 017603.
8
1.000
0.100
0.010
)
3
-
m
f
(
i
n
Bm = 1.5 x 105 Bc
e
n
p
e-
-
K-
n
p
e-
-
K-
0.001
1
2
3
n/n0
4
5
6
Fig.1 : The particle abundances are plotted with normalised baryon density for Bm = 0 and
Bm = 1.5×105Be
c . Solid lines indicate particle abundances for field free case whereas dashed
c ) is 4.414 × 1013G.
lines denote those with the magnetic field. The critical electron field (Be
9
m
m
III
II
I
100
)
3
-
m
f
V
e
M
(
P
10
200
400
600
e (MeV fm -3)
800
1000
Fig.2 : The matter pressure (P ) is shown as a function of matter energy density (ε) for
different values of Bm. The field free case is shown by curve I (solid line) and curve II
(dash-dotted line) and curve III (dashed line) represent calculations for Bm = 4× 104Be
c and
Bm = 1.5 × 105Be
c , respectively. The critical electron field (Be
c ) is 4.414 × 1013G.
10
|
astro-ph/0107505 | 1 | 0107 | 2001-07-26T14:54:47 | High Speed photometry of faint Cataclysmic Variables: I. V359 Cen, XZ Eri, HY Lup, V351 Pup, V630 Sgr, YY Tel, CQ Vel, CE-315 | [
"astro-ph"
] | The first results of a photometric survey of faint Cataclysmic Variables are presented. V359 Cen is an SU UMa star with a period of 112 min. Even though observed at quiescence, the mass transfer rate in this old nova may be sufficiently high that in such a short period system (with its implied small mass ratio) the disc may be excited into an elliptical shape with the result that the observed brightness modulation gives a superhump period rather than an orbital period. XZ Eri is an eclipsing dwarf nova with an orbital period (P(orb)) of 88.1 min. HY Lup has only slight variability. V351 Pup, the remnant of Nova Puppis 1991, has P(orb) = 2.837 h and a light curve that strongly resembles that of the magnetic Nova Cyg 1975. V630 Sgr is the first nova remnant that has both positive superhumps (P(sh) = 2.980 h) and eclipses (P(orb) = 2.831 h). The YY Tel identification is somewhat uncertain. The correct identification for CQ Vel is provided from discovery of its flickering activity. The light curve of CE-315, a recently discovered AM CVn star, shows similarities to that of GP Com, with no apparent orbital modulation. | astro-ph | astro-ph |
Mon. Not. R. Astron. Soc. 000, 000 -- 000 (0000)
Printed 26 October 2018
(MN LATEX style file v1.4)
High Speed photometry of faint Cataclysmic Variables: I.
V359 Cen, XZ Eri, HY Lup, V351 Pup, V630 Sgr, YY Tel,
CQ Vel, CE-315
Patrick A. Woudt⋆ and Brian Warner†
Department of Astronomy, University of Cape Town, Private Bag, Rondebosch 7700, South Africa
ABSTRACT
The first results of a photometric survey of faint Cataclysmic Variables are presented.
V359 Cen is an SU UMa star with a period of 112 min. Even though observed at
quiescence, the mass transfer rate in this old nova may be sufficiently high that in
such a short period system (with its implied small mass ratio) the disc may be excited
into an elliptical shape with the result that the observed brightness modulation gives
a superhump period rather than an orbital period. XZ Eri is an eclipsing dwarf nova
with an orbital period (Porb) of 88.1 min. HY Lup has only slight variability. V351
Pup, the remnant of Nova Puppis 1991, has Porb = 2.837 h and a light curve that
strongly resembles that of the magnetic Nova Cyg 1975. V630 Sgr is the first nova
remnant that has both positive superhumps (Psh = 2.980 h) and eclipses (Porb =
2.831 h). The YY Tel identification is somewhat uncertain. The correct identification
for CQ Vel is provided from discovery of its flickering activity. The light curve of CE-
315, a recently discovered AM CVn star, shows similarities to that of GP Com, with
no apparent orbital modulation.
Key words: techniques: photometric -- binaries: eclipsing -- close -- novae, cataclysmic
variables
1
INTRODUCTION
Cataclysmic Variable stars (CVs) are a class of interact-
ing close binaries which attract great attention. They en-
able a quantitative study of the many processes involved in
mass transfer through accretions discs or through magneti-
cally channeled accretion. A list of these physical processes,
which occur also in many other astrophysical environments,
is given in Warner (2000); a review of CVs in general is given
in Warner (1995a).
Fewer than 40 CVs have been studied in any great de-
tail (e.g., with doppler tomography, eclipse deconvolution).
These, and the phenomena seen in other CVs, show that
no two CVs are exactly alike. The different combinations of
masses of primaries and secondaries, temperatures, magnetic
field strengths, orbital inclinations and mass transfer rates
( M ) suffice to produce great variety of behaviour. Much of
this variety is already understood in terms of the compo-
nent physics accompanying the gross differences. The effects
of more subtle differences -- e.g., original chemical composi-
tion of the secondary, effects on the secondary of past nova
⋆ E-mail: [email protected]
† E-mail: [email protected]
c(cid:13) 0000 RAS
explosions, compositional and environmental (for instance,
irradiation) dependence of disc viscosity -- have still to be
teased out of the rich observational phenomenology.
Much of the progress in the past two or three decades
has come from the discovery of ever more variety of be-
haviour. There are often conspicuous effects in the longer
term light curves; examples are the ER UMa stars (dwarf
novae with outbursts every few days and superoutbursts ev-
ery few weeks), 'echo' outbursts after a superoutburst, con-
tinuous small outbursts in the discs of nova-like variables.
New phenomena continue to be found also in the 'close-up'
details; examples are short period quasi-periodic oscillations
(from ∼2 s in the X-ray flux of SS Cyg in outburst to ∼1000
s modulations in many high M discs), spiral waves in discs,
tilted discs, desynchronisation of polars by nova explosions.
Although some of the new phenomena have accompa-
nied the discovery of new CVs (e.g., the ∼ 108 G magnetic
fields in AR UMa and V884 Her), most have been discovered
in long known CVs (e.g., TeV gamma rays from AE Aqr, ZZ
Cet pulsations in GW Lib). It is clear, therefore, that while
the great majority of CVs remain under-observed new phe-
nomena may remain undiscovered. At a less exotic level, the
determination of a significant number of new orbital periods,
and in particular the discovery of eclipsing systems, provide
objects for more detailed structural studies. With the ad-
2
Patrick A. Woudt and Brian Warner
Table 1. Observing log.
Object
Type
Run No.
Date of obs. HJD of first obs.
(+2451000.0)
(start of night)
Length
(h)
tin
(s)
Tel.
<V>
(mag)
V359 Cen
SU UMa
XZ Eri
DN
HY Lup
NR
V351 Pup NR
V630 Sgr
NR
YY Tel
SU UMa
CQ Vel
NR
CE-315
AM CVn
S6165
S6167
S6170
S6186
S6189
S6197
S6216
S6156
S6158
S6163
S6191
S6157
S6160
S6162
S6185
S6192
S6124
S6126
S6128
S6130
S6100
S6218
S6223
S6231
S6071
S6209
S6212
29 Dec 2000
30 Dec 2000
1 Jan 2001
20 Feb 2001
21 Feb 2001
25 Feb 2001
18 May 2001
26 Dec 2000
27 Dec 2000
29 Dec 2000
22 Feb 2001
26 Dec 2000
27 Dec 2000
28 Dec 2000
20 Feb 2000
23 Feb 2000
24 Aug 2000
25 Aug 2000
26 Aug 2000
27 Aug 2000
5 Jun 2000
18 May 2001
21 May 2001
26 May 2001
908.50289
909.51368
911.51125
961.37008
962.41615
966.35024
1048.22183
905.29165
906.29347
908.30081
963.47488
905.47151
906.45415
907.43106
961.27432
964.26594
781.22202
782.21315
783.21314
784.20912
701.36536
1048.61265
1051.61806
1056.64150
10 Mar 2000
614.26263
15 May 2001
16 May 2001
1045.35878
1046.25888
2.15
1.95
2.02
2.05
4.03
2.22
4.02
4.02
0.36
0.65
3.76
3.06
3.62
3.96
1.93
1.20
6.72
7.15
7.16
6.15
1.39
1.72
1.57
1.07
4.07
4.26
6.63
30
30
30
30, 60
60
60
60
20, 60
30
20, 30
45
15
20
20
60
60
10
5
5, 10
5, 10
10
30, 60
60
30, 60
74-in
74-in
74-in
40-in
40-in
40-in
40-in
74-in
74-in
74-in
18.8
18.7
18.7
18.6
18.7
18.6
18.6
19.1
19.2
19.1
40-in
19.6
74-in
74-in
74-in
40-in
40-in
74-in
74-in
74-in
74-in
74-in
40-in
40-in
74-in
18.9
19.1
18.9
18.8
19.0
17.6
17.5
17.8
17.4
18.6
18.6
18.5
18.6
60
10
20
74-in
21.1
40-in
40-in
17.6:
17.6:
Notes: NR = Nova Remnant, DN = Dwarf Nova, tin is the integration time.
vent of 8-m class telescopes such studies can be made even
to 20th magnitude. The ongoing survey, of which this paper
presents the first general results, was initiated with this in
mind.
The need for such a survey, particularly in the south-
ern hemisphere and capable of succeeding in the crowded
fields in the general direction of the Galactic centre where
the majority of nova remnants reside, was recognised in the
early 1990s and as a result the CV group at the University
of Cape Town (UCT) used several years' equipment funds
(in advance) for the purchase of a state-of-the-art CCD de-
tector. The photometer and its software were designed and
constructed by Dr. D. O'Donoghue, who was a member
of the Astronomy Department at UCT at that time. The
UCT CCD photometer, completed in 1993, has been exten-
sively used on telescopes at the Sutherland site of the South
African Astronomical Observatory. An outline of its func-
tional properties can be found in O'Donoghue (1995) and
Koen & O'Donoghue (1995). In the CV group at UCT it
was first largely used for observations of well known CVs at
higher time resolution or improved signal-to-noise. Its use
for a general survey of faint CVs started in 1997, with the
almost immediate discovery of the first ZZ Cet (i.e., non-
radial) pulsations in the white dwarf primary of the SU
UMa type dwarf nova GW Lib (Warner & van Zyl 1998)
and independent discovery of the intermediate polar nature
of the nova remnant HZ Pup (see Abbott & Shafter 1997 for
detailed description). Follow-up observations of these led to
neglect of the survey for some time, but we are now actively
observing faint CVs again. We present here initial results.
The observations are given in Section 2 and some brief
conclusions in Section 3.
2 OBSERVATIONS
All observations have been made with the UCT CCD pho-
tometer attached either to the 40-in or 74-in reflectors at
Sutherland. In order to capture as many photons as possible
'white light' was used, with the result that extinction cor-
rections and magnitude calibrations are only approximate.
The observing log for the stars included in this paper is
given in Table 1.
2.1 V359 Centauri
V359 Cen was discovered bright in April 1929‡ and entered
the lists as a probable nova, unconfirmed because no spec-
tra were obtained. There are still no spectra available at
‡ Note that Duerbeck (1987) states 1939, and this has been prop-
agated in later literature.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
High Speed photometry of faint Cataclysmic Variables: I.
3
Figure 1. Light curves of V359 Cen at quiescence phased ac-
cording to the 112 min period. The light curves are plotted top
to bottom in the same order as in Table 1. All light curves, except
for the top one, are displaced along the magnitude axis for display
purposes only. The mean magnitude of V359 Cen for each run is
given in Table 1.
minimum light, but in recent years further outbursts to a
maximum of V ∼ 13.8 have been observed at intervals ∼ 1
y, showing the star to be of SU UMa type. The quiescent
magnitude given in the Downes, Webbink & Shara (1997)
catalogue is 21.0 from J plates, and Munari & Zwitter (1998)
gave V ≥ 20.5 in 1996, but we find V ∼ 18.7, which is a great
deal easier to observe at moderate time resolution. Our May
2001 observation was obtained three weeks after a superout-
burst, but the star had returned to its quiescent brightness.
Our light curves are displayed in Fig. 1 and show a re-
current hump with an amplitude ∼ 0.4−0.5 mag. Our obser-
vations are not distributed adequately to remove aliases; the
best determined period from the February 2001 run is 112
min (0.0779 days). The variations of hump profile and their
occasional almost triangular shape resemble more what is
seen in superhumps than orbital modulations, but it is rare
to observe superhumps at quiescence -- the only well docu-
mented examples are EG Cnc which sustained 'late' super-
humps for at least 90 d after the 1996 superoutburst (at a
time when 'echo' normal outbursts were occuring) (Patter-
son et al. 1998) and AL Com with a persistent superhump as
well as an orbital hump at quiescence (Patterson et al. 1996).
If what we are seeing is an orbital hump then the inclination
of V359 Cen must be ∼ 65◦.
2.2 XZ Eridanus
XZ Eri has been known as a probable dwarf nova for nearly
70 years (Shapley & Hughes 1934). In recent years only three
outbursts have been detected: Mar/Apr 1995, Jan 1998,
Feb 1999. The first of these reached V ∼ 14.1, the others
reached V ∼ 14.6. CCD photometry at quiescence gives V
= 18.7 (Howell et al. 1991). The rarity and range of the
outbursts suggests that they are superoutbursts. Any nor-
mal outbursts may have gone unnoticed. There has been no
high speed photometry during an outburst.
A quiescence spectrum, obtained by Szkody & How-
ell (1992) shows the Balmer emission and continuum char-
acteristic of a dwarf nova. There is a hint of doubling in
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Figure 2. The mean light curve of XZ Eri.
the strongest Balmer lines, indicative of a high inclination
system. Nearly 4 h of CCD photometry through a V filter
at quiescence, with 220 s integration on a 1.8-m telescope,
showed a scatter of ∼ 1.0 mag and, to quote the authors, no
indications of periodic modulation (Howell et al. 1991).
Our observations (Fig. 2) show XZ Eri to be, in brief,
a Z Cha lookalike, with the succession of immersions and
emersions of the white dwarf and bright spot all quite clear.
At V ∼ 19 we cannot achieve sufficient time resolution to
separate these components as clearly as in Z Cha (e.g., Wood
et al. 1986). Using our best estimate for the time of mid-
eclipse of the white dwarf we derive the following ephemeris:
HJDmin = 2451905.4419 + 0.d0612 E.
(1)
The short Porb (1.47 h = 88.1 min) implies that XZ Eri
will have superhumps when observed in outburst.
2.3 HY Lupi
HY Lup was Nova Lupi 1993, which reached V ∼ 8 and was
moderately fast (t2 ∼ 15 d). It is currently at V ∼ 18.9 and
has a resolved ejecta shell (Downes & Duerbeck 2000). No
pre-outburst candidate brighter than V ∼ 17 is detectable
(McNaught & Garradd 1993) and it is probable that the
remnant has not yet reached its quiescent brightness. No
high speed photometry has been published for this nova rem-
nant.
Our photometry (Fig. 3) shows little evidence for rapid
flickering and only slow variations of small amplitude. There
is no suggestion of any orbital modulation which would en-
courage further photometry. The Porb for HY Lup will need
to be obtained spectroscopically.
2.4 V351 Puppis
Nova Puppis 1991 (V351 Puppis) was a moderately fast nova
(t2 = 16 d), probably a neon nova (Pachoulakis & Saizar
1995). The pre-eruption magnitude was variable over the
range 21− > 22 (McNaught 1992). The eruption light curve
through 1996 is illustrated in Saizar et al. (1996); in 1998
the remnant was at V ∼ 19.6 (Downes & Duerbeck 2000).
There are one or two additional aspects of this nova
that are of interest. Sixteen months after optical maximum,
4
Patrick A. Woudt and Brian Warner
Figure 3. The light curve of HY Lup obtained on 22 February
2001. The data are binned along the horizontal axis by a factor
of five.
hard X-rays were detected by ROSAT which, from the ab-
sence of a soft X-ray component, were deduced to be caused
either by shocks in the ejected shell or by magnetically con-
trolled accretion onto the white dwarf (Orio et al. 1996).
Variability (time scale < days) observed in the X-ray flux
seemed to favour the latter interpretation. However, a later
analysis explains the absence of soft X-rays as due to in-
ternal and external absorption, allowing the hard X-rays to
be the product of surface nuclear burning, as in most novae
(Vanlandingham et al. 2001). Multiwavelength observations
of the nebular phase of the eruption made in 1992-1993 re-
vealed a strong red excess attributed to the secondary star,
implying that this component is either a giant or an irradi-
ated dwarf (Saizar et al. 1996). In the former case an orbital
period Porb ∼ days would be expected, in the latter Porb <∼ 4
h would be required in order for the secondary to be close
enough to the hot primary to be strongly irradiated. The
relatively small mass of the ejected shell of N Puppis 1991
implies a high mass for the white dwarf -- near the Chan-
drasekhar limit (Vanlandingham et al. 2001).
We observed V351 Pup in December 2000 and found it
to be modulated in brightness with a range of up to ∼1.2
mag and a period of 2.837 h. The light curves are given in
Fig. 4 and show variable amplitude, low amplitude flickering,
rounded maxima and a V-shaped minimum which may be
filled in to form a flattened region. The ephemeris from these
observations is
HJDmax = 2451905.56746 + 0.d1182 E.
(2)
The additional observations we made in February 2001
are too far removed in time to enable us to improve on the
precision of the period.
Our light curves for V351 Pup have a striking similar-
ity to those of V1500 Cyg in the late 1980s (see especially
Fig. 1 of Kaluzny & Chlebowski 1988). V1500 Cyg was Nova
Cygni 1975 and in the late 1980s it was about 15.3 mag be-
low its maximum brightness. Nova Pup 1991 was not discov-
ered until after maximum, but there are reasons to believe
it did not exceed V ∼ 5.5 (Saizar et al. 1996) and it there-
fore at present is ∼14 mag below maximum. V1500 Cyg is
a desynchronised polar (Stockman, Schmidt & Lamb 1988;
Kaluzny & Chlebowski 1988) with an orbital period of 3.35
h, which suggests the possibility that V351 Pup may also
be magnetic. With this in mind we asked Gary Schmidt to
check for circular polarization and he reports "A sequence
of spectropolarimetric observations was obtained by G. and
P. Schmidt on 2 January 2001 using the 74-in reflector at
Mt. Stromlo and covering the range 4220 -- 7300 A at ∼10
Figure 4. The light curves of V351 Pup obtained in December
2000, phased according to Eqn. 2. A vertical shift of 0.5 mag has
been introduced between light curves for display purposes.
A resolution. The summed spectrum, which spans a total
of 3.7 h, reveals a weak continuum (V ∼ 19.4) that is ap-
proximately flat in fλ, plus broad (±1700 km s−1) nebular
emission lines of H, He II and O III. The polarimetric results
fail to show significant circular polarization anywhere in the
observational sequence, and the coadded sum is unpolarized
to a 3σ upper limit of 1%".
Before dismissing the possibility that V351 Pup is a
magnetic nova we note that (a) the level of circular polariza-
tion in V1500 Cyg is ∼ 1.5% (Stockman et al. 1988), which,
compared with the 10 -- 30% in polars, shows the presence of
much unpolarized light, probably that from irradiation of
the secondary by the still hot primary; (b) V351 Pup is 10 y
from its eruption and less far in its recovery than was V1500
Cyg in the late 1980s -- and should therefore have even more
unpolarised light; (c) GQ Mus (Nova Mus 1983) also has a
polar-like orbital light curve and shows strong spectral sig-
natures of magnetic accretion, but no circular polarization
(Diaz & Steiner 1994); (d) similar remarks apply to V2214
Oph (Nova Oph 1988: Baptista et al. 1993).
2.5 V630 Saggitarii
V630 Sgr was the last of four novae detected in Saggitarius
in 1936. It was discovered in October of that year, at a time
when Harvard objective prism plates were being exposed,
whereby it attained the status and alias HD 321353. The
observed maximum was V ∼ 4.5, but it probably reached V
∼ 1.6 (Downes et al. 1997), and at minimum it is V ∼ 19,
giving it the unusually large eruption amplitude of ∼ 17.5
mag, even for a t2 = 4 d nova, suggesting a high inclination
system (Warner 1986: this conclusion arises from the appar-
ent intrinsic faintness of the nearly edge-on accretion disc
which such an object has in quiescence).
There have been no modern photometric or spectro-
scopic observations of V630 Sgr, which is a faint object in a
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
High Speed photometry of faint Cataclysmic Variables: I.
5
Figure 5. The light curves of V630 Sgr obtained in August 2000, phased according to Eqn. 3. These have been binned to 30 s to give
greater clarity of reproduction. The light curves are plotted top to bottom in the same order as in Table 1. All light curves, except for
the top one, are displaced along the magnitude axis for display purposes only. The mean magnitude of V630 Sgr for each run is given in
Table 1.
V630 Sgr
Figure 6. The Fourier transform of the light curves shown in
Fig. 5.
very crowded field. It is listed by Diaz & Steiner (1991), on
the basis of its similarity to V1500 Cyg in being a very fast
nova of large amplitude, as a candidate magnetic system.
Our light curves are shown in Fig. 5 and illustrate that
V630 Sgr is indeed a moderately high inclination system,
with eclipses 0.4 -- 0.6 mag deep. The ephemeris derived from
our light curves, measured from the fundamental Fourier
component. The time of the first minimum of our first light
curve of this set is given in the ephemeris, which is
HJDmin = 2451781.2480 + 0.d1180 E.
(3)
The orbital period of 2.831 h is just at the top of the
orbital period gap.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
It is evident from Fig. 5 that there is a hump of com-
plex and variable profile that has a period comparable to
Porb. The Fourier amplitude spectrum of the complete set
of light curves is shown in Fig. 6. The main groups of power
are associated with the fundamental and strong harmon-
ics of the narrow eclipses. The window pattern of the data
set can be judged from the first and second harmonics. Su-
perimposed, but slightly offset from the Porb fundamental
window pattern, is a second pattern. Prewhitening at the
orbital frequency gives a clear pattern with an amplitude of
0.07 mag centred on Psh = 2.980 h, which is the period of
the drifting humps. Therefore, V630 Sgr is a nova remnant
that is also an eclipsing permanent superhumper. This is the
first CV found to combine all of these features. The super-
hump excess of 5.3% and beat period of 2.4 d are typical
for superhumps in superoutbursting dwarf novae and nova-
like variables (Warner 1995a). The beat period is evident in
the night-to-night variation of mean brightness, as seen in
Fig. 7. The Fourier transform of this delivers an approximate
period of 2.1 d and an amplitude of 0.22 mag.
The eclipse depth of ∼ 0.6 mag is relatively shallow,
but at the Porb of V630 Sgr it implies a quite high orbital
inclination. The width at half depth of the narrowest eclipses
is ∆ψ ∼ 0.085Porb. The secondary mass M (2) at Porb = 2.8
h is ∼ 0.24 M⊙ (Warner 1995a) and as V630 Sgr was a
very fast nova we expect M (1) ∼ 1.2 M⊙, giving a mass
ratio q ∼ 0.20. These figures are not quite compatible with
the theoretical ∆ψ, q, i relationship (Horne 1985), but they
serve to show that i ≥ 85◦. The presence of superhumps
shows V630 Sgr to be a high M system which will have a
disc radius at the 3:1 resonant radius, which is rd = 0.45a.
The Roche lobe radius for the secondary is R(2) = 0.25a for
q = 0.2. Therefore, even at large inclination, the eclipses are
partial, as is seen in the eclipse profiles in Fig. 5.
The observed eruption range of 17.5 mag is 3.5 mag
greater than expected for a t2 = 4 d nova (see Fig. 5.4 of
6
Patrick A. Woudt and Brian Warner
Figure 7. The long term optical behaviour of V630 Sgr.
Figure 8. The light curve of YY Tel obtained on 6 June 2000. The
lower light curve shows a comparison star of the same brightness
(displaced by 0.25 mag for display purposes).
Warner 1995a). Using the correction formula (Eqn. 2.63 of
Warner 1995a) for disc inclinations, removes the 3.5 mag
excess if i ∼ 88◦. We conclude that V630 Sgr is a very high
inclination system.
Figure 9. CCD image of YY Tel (the brighter star of the two
indicated by markers) taken on 21 May 2001. The field of view is
109′′ by 74′′, north is up and east is to the left.
2.6 YY Telescopium
YY Tel is a dwarf nova with a range V ∼ 14.4−19.3 (Downes
et al. 1997) suspected to be an SU UMa star of long out-
burst interval (O'Donoghue et al. 1991). The last reported
outburst was in October 1998, and no high speed photom-
etry has been reported, either in outburst or at quiescence.
The candidate identified at quiescence (Downes et al. 1997)
was observed spectroscopically by Zwitter & Munari (1996)
and showed only a continuum with no emission lines char-
acteristic of a CV. Their photometry gave V = 19.26, (B-V)
= 0.47, (U-B) = -0.28, placing the candidate near the black
body line at T ∼ 6700 K in the two colour diagram. This is
also a region frequented by late F type subdwarfs.
Our observation on 5 June 2000 showed the YY Tel
candidate to be constant in brightness (see Fig. 8). Re-
examination nearly one year later again showed absence of
any rapid variation but the star changed in brightness by
0.15 mag over three to five days. A nearby, fainter star,
marked in Fig. 9 (YY Tel is the brighter of the two stars
that are marked), shows marginal evidence for variation and
should be examined spectroscopically.
2.7 CQ Velorum
CQ Vel was Nova Velorum 1940, reaching V ∼ 9.0. It has
been identified with a 21st magnitude star (Duerbeck 1987)
but no confirmatory spectra or photometry have been ob-
tained. In our observations on 10 March 2000 we found
the candidate nova remnant to be of constant brightness
as were all the stars in the CCD frame, except for a star
slightly fainter than the candidate, 9′′ to the northwest
(Fig. 10). This star has the flickering characteristic of a CV,
see Fig. 11. These observations provide a correct identifica-
tion for CQ Vel.
2.8 CE-315
CE-315 is a recently discovered member of the AM CVn
class of CVs: systems transferring helium rather than hy-
drogen (Ruiz et al. 2001). It is only the second of this class
to be found in a long-term state of low M , the other being
GP Com (alias G61-29). Ruiz et al. found a period of 65.1
± 0.7 min from an S-wave in the spectra, which by analogy
with GP Com is thought to be the orbital period.
We have obtained two light curves for CE-315, which are
displayed in Fig. 12. There is no sign in the light curve, or
its Fourier Transform, of the orbital periodicity. CE-315 re-
sembles GP Com in the general appearance of its light curve
(see Warner 1972; Harrop-Allin 1996): much short time scale
flaring or flickering set on a slowly varying background. In
GP Com there are occasions when a few cycles at the 46.5
min orbital period appear in the light curve (Harrop-Allin
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
High Speed photometry of faint Cataclysmic Variables: I.
7
result of radiation of gravitational waves (Warner 1995a,b).
According to Eqn. 9.61 of Warner (1995a) the mass transfer
M ∼ 1.0 × 10−11 M⊙ y−1 for GP Com
rates should be
and M ∼ 3 × 10−12 M⊙ y−1 for CE-315. GP Com has X-
rays modulated at the orbital period (Beuermann & Thomas
1993). At B ∼ 17.2, CE-315 is ∼ 1.4 mag fainter than GP
M , so they are
Com, but this is accounted for by the lower
at similar distances and it would be worth seeking an X-ray
detection.
3 CONCLUSIONS
The results presented here cover a miscellany of CV types
and demonstrate the rewards waiting among the fainter
members of this class. For those objects where new phe-
nomena have been found (e.g., the possible quiescent super-
humps in V359 Cen), or where rarity is an incentive (e.g.,
the eclipsing superhumping nova remnant V630 Sgr), further
observations with larger telescopes are required. V351 Pup
may show circular polarization or other magnetic signatures
when it has descended the nova light curve further.
ACKNOWLEDGMENTS
We thank Dr. D. O'Donoghue for the use of his EAGLE
program for Fourier analysis of the light curves. PAW is
funded partly through strategic funds made available to BW
by the University of Cape Town. BW's research is funded
entirely by that university.
REFERENCES
Abbott T.M.C., Shafter A.W., 1997, IAU Colloq. No. 163, 679
Baptista R., Jablonski F.J., Cieslinski D., Steiner J.E., 1993, ApJ,
406, L67
Beuerman K., Thomas H.-C., 1993, Adv.Sp.Res., 13, 115
Diaz M.P., Steiner J.E., 1991, PASP, 103, 964
Diaz M.P., Steiner J.E., 1994, ApJ, 425, 252
Downes R.A., Duerbeck H.W., 2000, AJ, 120, 2007
Downes R.A., Webbink R.F., Shara M.M., 1997, PASP, 109, 345
Duerbeck H.W., 1987, Sp.Sci.Rev., 45, 1
Harrop-Allin M.K., 1996, MSc thesis, University of Cape Town
Horne K., 1985, MNRAS, 213, 129
Howell S.B., Szkody P., Kreidl T.J., Dobrzycka D., 1991, PASP,
103, 300
Kaluzny J., Chlebowski T., 1988, ApJ, 332, 287
Koen C., O'Donoghue D., 1995, ApJS, 101, 347
McNaught R., 1992, IAU Circ. 5422
McNaught R., Garradd G.J., 1993, IAU Circ. 5868
Munari U., Zwitter T., 1998, A&AS, 128, 277
O'Donoghue D., 1995, Baltic Astr, 4, 517
O'Donoghue D., Chen A., Marang F., Mittaz J.P.D., Winkler H.,
Warner B., 1991, MNRAS, 250, 363
Orio M., Balman S., Della Valle M., Gallagher J., Ogelman H.,
1996, ApJ, 466, 410
Pachoulakis I., Saizar P., 1995, ApSpSciLibr, 205, 303
Patterson J., Augustein T., Harvey D.A., Skillman D.R., Abbott
T.M.C., Thorstensen J., 1996, PASP, 108, 748
Patterson J., Kemp J., Skillman D.R., Harvey D.A., Shafter
A.W., Vanmunster T., Jensen L., Fried R., Kiyota S.,
Thorstensen J.R., Taylor C.J., 1998, PASP, 110, 1290
Ruiz M.T., Rojo P.M., Garay G., Maza J., 2001, ApJ, 552, 679
Figure 10. CCD image of CQ Vel (indicated by the markers)
taken on 10 March 2000. The field of view is 50′′ by 34′′, north
is up and east is to the left.
Figure 11. The light curve of CQ Vel (V = 21.1) obtained on 10
March 2000.
1996); if the same occurs in CE-315 it would need extensive
photometry to detect.
The fact that the two longest period AM CVn stars, GP
Com and CE-315, are low M systems (equivalent to dwarf
novae in hydrogen transferring CVs) is in accord with the
evolutionary model where orbital angular momentum is the
Figure 12. The light curves of CE-315 obtained on 15 and 16
May 2001. The data in the top panel have been binned along the
horizontal axis by a factor of four, the data in the bottom panel
by a factor of two.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
8
Patrick A. Woudt and Brian Warner
Saizar P., Pachoulakis I., Shore S.N., Starrfield S., Williams R.E.,
Rothschild E., Sonneborn G., 1996, MNRAS, 279, 280
Shapley H., Hughes E.M., 1934, HA, 90, No. 4
Stockman H.S., Schmidt G.D., Lamb D.Q., 1988, ApJ, 332, 282
Szkody P., Howell S.B., 1992, ApJS, 78, 537
Vanlandingham K.M., Schwarz G.J., Shore S.N., Starrfield S.,
2001, AJ, 121, 1126
Warner B., 1972, MNRAS, 159, 315
Warner B., 1986, MNRAS, 222, 11
Warner B., 1995a, Cataclysmic Variable Stars, Cambridge Univ.
Press, Cambridge
Warner B., 1995b, ApSpSci, 225, 249
Warner B., 2000, PASP, 112, 1523
Warner B., van Zyl L., 1998, IAU Symp. No. 185, 321
Wood J., Horne K., Berriman G., Wade R.A., O'Donoghue D.,
Warner B., 1986, MNRAS, 219, 629
Zwitter T., Munari U., 1996, A&AS, 117, 449
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
|
astro-ph/9810060 | 1 | 9810 | 1998-10-06T00:00:37 | The Ultraviolet Luminosity Density of the Universe from Photometric Redshifts of Galaxies in the Hubble Deep Field | [
"astro-ph"
] | Studies of the Hubble Deep Field (HDF) and other deep surveys have revealed an apparent peak in the ultraviolet (UV) luminosity density, and therefore the star-formation rate density, of the Universe at redshifts 1<z<2. We use photometric redshifts of galaxies in the HDF to determine the comoving UV luminosity density and find that, when errors (in particular, sampling error) are properly accounted for, a flat distribution is statistically indistinguishable from a distribution peaked at z~1.5. Furthermore, we examine the effects of cosmological surface brightness (SB) dimming on these measurements by applying a uniform SB cut to all galaxy fluxes after correcting them to redshift z=5. We find that, comparing all galaxies at the same intrinsic surface brightness sensitivity, the UV luminosity density contributed by high intrinsic SB regions increases by almost two orders of magnitude from z~0 to z~5. This suggests that there exists a population of objects with very high star formation rates at high redshifts that apparently do not exist at low redshifts. The peak of star formation, then, likely occurs somewhere beyond z>2. | astro-ph | astro-ph | The Ultraviolet Luminosity Density of the Universe from
Photometric Redshifts of Galaxies in the Hubble Deep Field1,2
Sebastian M. Pascarelle and Kenneth M. Lanzetta
Department of Physics and Astronomy, State University of New York at Stony Brook,
Stony Brook, NY 11794-3800
and
Alberto Fern´andez-Soto
School of Physics, University of New South Wales, Sydney, NSW2052, Australia
8
9
9
1
t
c
O
6
1
v
0
6
0
0
1
8
9
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
1Based on observations with the NASA/ESA Hubble Space Telescope obtained at the Space Telescope
Science Institute, which is operated by AURA, Inc., under NASA Contract NAS 5-26555.
2Based in part on observations with the KPNO 4-meter Mayall Telescope of NOAO, which is operated
by AURA, Inc., under cooperative agreement with the NSF.
-- 2 --
ABSTRACT
Studies of the Hubble Deep Field (HDF) and other deep surveys have revealed
an apparent peak in the ultraviolet (UV) luminosity density, and therefore the
star-formation rate density, of the Universe at redshifts 1 < z < 2. We use photo-
metric redshifts of galaxies in the HDF to determine the comoving UV luminosity
density and find that, when errors (in particular, sampling error) are properly
accounted for, a flat distribution is statistically indistinguishable from a distri-
bution peaked at z ≃ 1.5. Furthermore, we examine the effects of cosmological
surface brightness (SB) dimming on these measurements by applying a uniform
SB cut to all galaxy fluxes after correcting them to redshift z = 5. We find that,
comparing all galaxies at the same intrinsic surface brightness sensitivity, the UV
luminosity density contributed by high intrinsic SB regions increases by almost
two orders of magnitude from z ≃ 0 to z ≃ 5. This suggests that there exists a
population of objects with very high star formation rates at high redshifts that
apparently do not exist at low redshifts. The peak of star formation, then, likely
occurs somewhere beyond z > 2.
Subject headings: galaxies: evolution -- galaxies:
universe
formation -- cosmology: early
-- 3 --
1.
Introduction
Important properties of the Universe as a whole can be determined by analyzing com-
plete samples of galaxy redshifts to very faint magnitude limits. The comoving ultraviolet
(UV) luminosity density is one such property. Because it is dominated by massive, short-
lived stars, the UV emission of an actively star forming galaxy is nearly independent of star
formation history. For this reason, the comoving UV luminosity density is directly related to
the comoving star formation and metal production rate densities of the Universe (e.g., Cowie
1988; Fall, Charlot, & Pei 1996; Madau et al. 1996, hereafter M96). Lilly et al. (1996, here-
∼ 1 using the Canada -- France
after L96) measured the UV luminosity density at redshifts z <
Redshift Survey (CFRS) and found that it rises rapidly by a factor of ∼ 15 from z = 0 to
z = 1. Similar results were found in this redshift range by Cowie, Hu, & Songaila (1997).
M96 (later updated by Madau, Pozzetti, & Dickinson 1998, hereafter M98) measured
the UV luminosity density at redshifts z ≃ 2.75 and z ≃ 4.00 using the U- and B-band
"dropout" technique applied to the Hubble Deep Field (HDF) and found that it might be
lower than the L96 measurements at z <
∼ 1 by a factor of ∼ 2. This suggested that there
might be a peak in the UV luminosity density -- and therefore the star formation and metal
production rate densities -- at some redshift between z = 1 and 2, prompting a spate of
theoretical and observational interpretation (e.g., Shaver et al. 1996; Baugh et al. 1998;
Ferguson & Babul 1998; M98; Madau, Della Valle, & Panagia 1998; Silk & Rees 1998).
More recently, Connolly et al. (1997, hereafter C97) measured the UV luminosity density
from the HDF in the previously unsurveyed redshift range 1 <
∼ 2 and found that it appears
to peak at z ≃ 1.5, which is consistent with the results of L96 and M98. However the
photometric redshifts of galaxies in the HDF from Sawicki, Lin, and Yee (1997) result in
∼ 2.5, indicating that the peak at
a UV luminosity density that continues to increase to z >
z ≃ 1.5 may be questionable.
∼ z <
In this paper, we apply our photometric redshifts of galaxies identified in the HDF
(Lanzetta, Yahil, & Fern´andez-Soto 1996, hereafter LYF96; Lanzetta, Fern´andez-Soto, &
Yahil 1998; Fern´andez-Soto, Lanzetta, & Yahil 1998, hereafter FLY98) to measure the UV
luminosity density of the Universe at redshifts 0 < z < 6. Our analysis differs from previous
analyses in three important ways. First, our photometric redshifts are determined from
spectral template fits to optimal photometry of optical (Williams et al. 1996) and infrared
(Dickinson et al. 1998) images of the HDF. In contrast to the U- and B-band "dropout"
technique of M98, our analysis determines the most likely redshift of every galaxy in the
HDF, and in contrast to the photometric redshifts of C97, our analysis makes use of the J-,
H-, and K-band infrared photometry (instead of only the J-band photometry). Second, we
determine realistic uncertainties of the luminosity density measurements, including the effects
-- 4 --
of systematic and photometric error on the photometric redshifts and of sampling error.
Previous analyses have neglected sampling error, which in fact dominates the uncertainty of
the luminosity density measurements in the HDF. Third, we explicitly consider the effects
of cosmological (1 + z)3 surface brightness dimming on the observed luminosity density.
Previous analyses have neglected cosmological (1 + z)3 surface brightness dimming, which
varies by more than two orders of magnitude over the redshift range of galaxies identified in
the HDF image.
In § 2 we briefly describe our photometric redshift technique. In § 3 we present the results
of our measurement of the UV luminosity density, in § 4 we show how this determination is
affected by cosmological surface brightness dimming, and in § 5 we discuss these results and
present our conclusions.
2. Photometric Redshifts
The starting point of our analysis is the photometric redshift estimates of LYF96 and
FLY98. Because details of the photometric redshift estimation technique have been and
will be presented elsewhere (Lanzetta, Fern´andez-Soto, & Yahil 1998; FLY98), we simply
summarize the method here.
Galaxy photometry is determined from the optical F300W, F450W, F606W, and F814W
(Williams et al. 1996) and infrared J, H, and K (Dickinson et al. 1998) images of the HDF.
To measure fluxes and flux uncertainties in the optical images, we directly integrate within
the aperture mask of every object detected in the F814W image. To measure fluxes and flux
uncertainties in the infrared images, we (1) model the spatial profile of every object detected
in the F814W image as a convolution of the portion of the F814W image containing the object
with the appropriate point spread function of the infrared image and (2) determine a least-
squares fit of a linear sum of the model spatial profiles to the infrared image. The advantages
of this method over simple aperture photometry are that the flux measurements correctly
weight signal-to-noise ratio variations within the spatial profiles, and the flux uncertainty
measurements correctly include the contributions of nearby, overlapping neighbors.
Galaxy redshifts are determined from fits to the spectral templates of E/S0, Sbc, Scd,
and Irr galaxies, including the effects of intrinsic and intervening neutral hydrogen absorp-
tion. These effects are included as a function of redshift, with mean values taken from
Madau (1995) and Webb (1997). First, we integrate the redshifted spectral templates with
the throughputs of the F300W, F450W, F606W, F814W, J, H, and K filters, at redshifts
spanning z = 0 − 7. Next, we construct the "redshift likelihood function" of every object
-- 5 --
detected in the F814W image by calculating the relative likelihood of obtaining the mea-
sured fluxes and uncertainties given the modeled fluxes at an assumed redshift, maximizing
with respect to galaxy spectral type and arbitrary flux normalization. Finally, we determine
the maximum-likelihood redshift estimate of every object detected in the F814W image by
maximizing the redshift likelihood function with respect to redshift. The result of the most
recent application of this method is presented in the catalog of FLY98, which lists photomet-
ric redshift estimates of 1067 galaxies to a limiting magnitude threshold of AB(814) = 28.0
over an angular area of ∼4 arcmin2.
Spectroscopic redshifts of more than 100 galaxies in the HDF have been obtained using
the Keck telescope (Cohen et al. 1996; Cowie 1997; Steidel et al. 1996; Zepf et al. 1997;
Lowenthal et al. 1997). Comparison between the spectroscopic and photometric redshifts
indicates the following results: (1) At redshifts z < 2, the residuals between the spectro-
scopic and photometric redshifts are characterized by an RMS dispersion of σ = 0.09 and
a discordant fraction (> 3σ discrepant after sigma clipping) of 0%; (2) at redshifts z > 2,
the residuals between the spectroscopic and photometric redshifts are characterized by an
RMS dispersion of σ = 0.45 and a discordant fraction of 7%; and (3) these residuals between
the spectroscopic and photometric redshifts arise from cosmic variance with respect to the
spectral templates (rather than from photometric error), so a proper assessment of the er-
rors of the photometric redshifts of faint galaxies must include the effects of systematic and
photometric error.
3. Ultraviolet Luminosity Density
We determined the luminosity (per unit wavelength interval) of each galaxy at a rest-
frame wavelength ∼1500A by applying an empirical K-correction derived from the best-fit
spectral template to the measured galaxy photometry. The K-corrections are interpolated
from the measured photometry at redshifts z > 0.6, although they are extrapolated from the
measured photometry (by up to a factor of two in wavelength) at redshifts z < 0.6. Next,
we determined the comoving luminosity density versus redshift by arranging the galaxies
into redshift bins, summing the luminosities within the bins, and dividing by the appropri-
ate comoving volumes. Finally, we determined the uncertainty of the comoving luminosity
density versus redshift by applying a bootstrap resampling technique. For each iteration of
the bootstrap technique, we resampled the photometric catalog and redetermined the pho-
tometric redshift of each resampled galaxy, perturbing the photometry by random deviates
according to the measured photometric error and perturbing the redshift by a random de-
viate according to the RMS residuals described in §2. We then determined the comoving
-- 6 --
luminosity density versus redshift using the resampled, perturbed redshift catalog. We re-
peated the procedure 1000 times in order to determine the range of values compatible with
the observations. This procedure explicitly allows for sampling error, photometric error, and
cosmic variance with respect to the spectral templates.
The results are shown in the top panel of Figure 1, which plots the comoving UV
luminosity density versus redshift, and are given in the second column of Table 1. For
comparison, the data points from L96, C97, and M98 are also plotted in Figure 1a. Note
that our data are entirely consistent with the previous measurements to well within ≃ 1.5σ
(and all but our first data point are within <
∼ 1σ of previous data). The fact that we do not
appear to reproduce the steep rise in the UV luminosity density from z = 0 − 1 of L96 is
attributed only to the lowest redshift data point of L96, which is <
∼ 1.5σ discrepant. While
this is still consistent with our measurement, the small difference is likely due to the much
larger sample of galaxies and the much larger surface area covered by the L96 sample as
compared to that of the relatively small HDF field.
Table 1: Comoving UV Luminosity Density
Redshift
0.00 -- 0.50
0.50 -- 1.00
1.00 -- 1.50
1.50 -- 2.00
2.00 -- 3.00
3.00 -- 4.00
4.00 -- 5.00
5.00 -- 6.00
Luminosity Densitya Luminosity Density HDF Galaxies HDF Galaxies
above SB Cut
above SB Cut
0.12
0.22
0.14
26.29± 0.31
26.43± 0.23
26.62± 0.24
26.65± 0.24
26.44± 0.28
26.41± 0.34
26.52± 0.44
26.16± 0.59
0.37
0.12
0.27
0.21
0.38
0.13
0.25
0.26
24.07± 0.24
24.98± 0.25
25.69± 0.11
25.61± 0.10
26.24± 0.17
26.15± 0.15
26.44± 0.19
25.77± 0.28
0.10
0.17
0.14
0.20
0.28
152
241
228
205
141
62
50
8
1
7
23
49
85
60
50
8
aValues are log luminosity density at restframe 1500A in units of h−2
100 erg s−1 Hz−1 Mpc−3 (qo=0.5). Errors
include the effects of systematic and photometric error on the photometric redshifts and of sampling error.
With sampling errors properly accounted for, the errorbars generated from our bootstrap
code are more than a factor of two larger than those of C97 or M98 for the z > 2 UV
luminosity density. It can be seen that, after an increase at redshifts z ≃ 0 − 2, the UV
luminosity density remains constant to within errors to redshifts z ≃ 6. In other words, we
find no convincing evidence that the UV luminosity density decreases with redshift for z > 2.
-- 7 --
Fig. 1. -- (a), The UV luminosity density as a function of redshift as measured from galaxies
in the HDF using our photometric redshifts (filled circles). The errorbars were calculated
from a bootstrapping code which takes into account the effects of systematic and photometric
error on the photometric redshifts and the effects of sampling error. For comparison, we have
included data points from Lilly et al. (1996, open triangles), Connolly et al. (1997, open
circles), and Madau et al. (1998, open squares). Note that while our data appear to show
a possible increase in the UV luminosity density from z = 0 − 2, there is little evidence for
a decrease at higher redshifts to within the errors. (b), The UV luminosity density arising
from intrinsically high surface brightness (SB) regions, after applying a uniform SB cut so
that z < 5 galaxies are considered down to the same SB level as z >
∼ 5 galaxies. Comparing
the HDF galaxies at the same intrinsic SB sensitivity shows that the UV luminosity density
of high SB regions increases by almost two orders of magnitude from z ≃ 0 to z ≃ 5.
-- 8 --
4. Effects of Cosmological (1 + z)3 Surface Brightness Dimming
To meaningfully compare the comoving UV luminosity density of the local, low-redshift
Universe with that of the distant, high-redshift Universe, it is necessary to account for the
very large effect of cosmological (1 + z)3 surface brightness (SB) dimming. Because low-
redshift galaxies are viewed to much lower intrinsic SB thresholds than are high-redshift
galaxies, certain corrections must be applied in order to view all galaxies at a common
intrinsic SB threshold.
We applied these corrections to our catalog of 1067 galaxies. First, we corrected the flux
of each galaxy to the value that would be observed if the galaxy were placed at redshift z = 5.
Specifically, we applied monochromatic SB corrections and K-corrections on a pixel-by-pixel
basis, as in Bouwens, Broadhurst, & Silk (1998), and we also applied small corrections
to account for the different WFPC2 passbands that sample rest-frame 1500 A at different
redshifts.
Next, we applied a uniform SB cut on a pixel-by-pixel basis, excluding pixels that
failed to meet a minimum SB threshold. The threshold was determined by assuming that
objects are detected in the F814W image to within ∼ 1σ of sky, which corresponds to a SB of
1.27×10−33 erg s−1 cm−2 Hz−1 pixel−1. Next, we reversed the monochromatic SB corrections
and K-corrections on a pixel-by-pixel basis to bring each galaxy back to its actual redshift
and again calculated the comoving luminosity density versus redshift. In this way, only those
parts of the galaxies that are of high enough intrinsic SB to be detected at all redshifts up
to z = 5 -- given the actual sensitivity of the HDF F814W image -- are included into the
measurement.
Results are shown in the bottom panel of Figure 1, which plots the comoving UV
luminosity density versus redshift of high intrinsic surface brightness regions, and in the
third column of Table 1. Also listed in Table 1 are the number of galaxies within each
redshift bin and the number of galaxies in each bin that have at least one pixel above the
SB cut. It is evident that (1) the comoving UV luminosity density of high intrinsic surface
brightness regions increases by two orders of magnitude from z ≃ 0 to z ≃ 5, and (2)
star-forming objects seen at z > 2.5 are relatively rare at z < 2.5.
In other words, the
comoving UV luminosity density contributed by high intrinsic SB regions appears to increase
monotonically with redshift, at least out to z ≃ 5.
-- 9 --
5. Discussion and Conclusions
Previous determinations of the comoving UV luminosity density at z > 2 rely primarily
on the M98 Lyman break galaxy sample. The stated uncertainties of these measurements
include contributions from incompleteness at the faint end of the luminosity function (LF)
as well as from the volume normalization and color selection region. While these effects do
indeed contribute to the total uncertainty, they are by no means the dominant factors. For
any LF with power-law index α < 2, the luminosity density is dominated by the bright end
of the LF. Because the bright end of the LF is inherently poorly sampled, this sampling
error at the bright end of the LF in fact dominates the uncertainty of the luminosity density.
Indeed, at redshifts z > 2, our uncertainties -- which include the effects of sampling error --
are more than two times larger than those quoted by C97 and M98. This calls into question
the statistical significant of the "peak" in the comoving UV luminosity density at z ≃ 1.5.
Ignoring errors, one can see from Figure 1a that our z = 4 measurement of the UV
luminosity density is higher than that of M98. It is possible that this arises because the
M98 method of finding galaxies at that redshift by their definition does not find all galaxies
at z ≃ 4. Rather, their color-color polygon was designed to find objects that are almost
certainly at z ≃ 4 with little contamination from low-redshift objects, which is why the data
points of M96 were plotted as lower limits. In Figure 2, we plot the z ≃ 4 objects from our
photometric redshift catalog with the M98 polygon. One can see that even with the large
(B − V ) errors involved, the M98 technique may be missing half of the high-redshift galaxies
in the HDF.
Perhaps the single most important factor that must be included in such a study is the
enormous effect of the cosmological surface brightness dimming at very high redshifts. At
z < 0.1 this effect is less than a factor of 1.3, but at z = 5 it becomes a factor of 216.
Therefore, if we are to compare the UV luminosity density at z = 5 to that of the local
Universe, we must consider local galaxies only down to the same SB level as that sampled
at high-redshift.
At the highest redshifts, the severe cosmological SB dimming affecting the HDF observa-
tions allows only the very highest levels of UV luminosity "column density" (or equivalently,
star formation rate "column density") to be sampled. At progressively lower redshifts, we
are able to observe galaxies down to lower and lower SB thresholds. The uniform SB cut
that we applied to our galaxy sample to allow for this discrepancy corresponds to a star
formation rate column density of ≃ 0.1h2
100 M⊙ yr−1 kpc−2 at z = 5 (assuming qo = 0.5 and
using the relation of UV luminosity to star formation rate with a Salpeter IMF from M98).
As seen in Table 1 and the bottom panel of Figure 1, almost no z < 1 objects have star
formation rates above this level, while at high redshifts, there are many objects exceeding
-- 10 --
Fig. 2. -- Color-color diagram similar to that of Madau et al. (1996, 1998) containing z ≃ 4
galaxies selected from our photometric redshift catalog of HDF galaxies and plotted with
their measured photometric errors.
It can be seen that, while this technique does locate
z ≃ 4 galaxies, almost 50% of them may lie outside the selection polygon despite the large
(B − V ) errors involved.
-- 11 --
this cut. For example, only 7 out of 241 galaxies from the z = 0.5 − 1.0 redshift bin would
be visible at z = 5, and then only the brightest few image pixels of those objects would peak
above the SB cutoff.
Figure 1 provides us with another piece of evidence in favor of a UV luminosity density
that increases with redshift. In the lowest redshift bin (z = 0.0 − 0.5) of Figure 1a and b,
where the SB effect is smallest, the intrinsically high SB regions make up only 0.5% of the
total UV luminosity density. If this ratio is the same at high redshifts as it is at low redshifts
then our measurements of the UV luminosity density at high redshifts in Figure 1a may need
to be increased by up to a factor of 150. Although this ratio may be quite different at high
redshifts than it is at low redshifts, this suggests that the UV luminosity density could be a
strongly increasing function of redshift.
We have shown that when the low- and high-redshift Universes are observed on equal
footing, one gets a completely different impression of the global UV luminosity density than
previously thought. We find that objects with star formation rates comparable to those at
∼ 3 are very rare in the nearby Universe. This implies that a majority of the star formation
z >
may have occurred at very high redshifts, and therefore that a peak in the star formation
rate density of the Universe has not yet been observed and likely lies somewhere at z > 2.
SMP and KML acknowledge support from NASA grant NAGW-4422 and NSF grant
AST-9624216. AF acknowledges support from a grant from the Australian Research Council.
REFERENCES
Baugh, C. M., Cole, S., Frenk, C. S., & Lacey, C. G. 1998, ApJ, 498, 504
Bouwens, R., Broadhurst, T., & Silk, J. 1998, ApJ (submitted, preprint astro-ph/9710291)
Cohen, J. G., Cowie, L. L., Hogg, D. W., Songaila, A., Blandford, R., Hu, E. M., & Shopbell,
P. 1996, ApJ, 471, L5
Connolly, A. J., Szalay, A. S., Dickinson, M., SubbaRao, M. U., & Brunner, R. J. 1997, ApJ,
486, L11
Cowie, L. L. 1988, in The Post-Recombination Universe, eds Kaiser, N. & Lasenby, A.
(Dordrecht, Kluwer) 1
Cowie, L. L., Hu, E. M., & Songaila, A. 1997, ApJ, 481, L9
Cowie, L. L. 1997, http://www.ifa.hawaii.edu/∼cowie/tts/tts.html
Dickinson, M. et al. 1998 (in preparation)
-- 12 --
Fall, S. M., Charlot, S., & Pei, Y. 1996, ApJ, 464, L43
Ferguson, H. C., & Babul, A. 1998, MNRAS, 296, 585
Fern´andez-Soto, A., Lanzetta, K. M., & Yahil, A. 1998, ApJ (in press)
Lanzetta, K. M., Yahil, A., & Fern´andez-Soto, A. 1996, Nature, 381, 759
Lanzetta, K. M., Fern´andez-Soto, A., & Yahil, A. 1998, in "The Hubble Deep Field," pro-
ceedings of the Space Telescope Science Institute 1997 May Symposium, ed. M. Livio,
S. M. Fall, and P. Madau (in press, preprint astro-ph/9709166)
Lilly, S. J., & Le F`evre, O., Hammer, F., & Crampton, D. 1996, ApJ, 460, L1
Lowenthal, J. D. et al. 1997, ApJ, 481, 673
Madau, P. 1995, ApJ, 441, 18
Madau, P., Ferguson, H. C., Dickinson, M., Giavalisco, M., Steidel, C. C., & Fruchter, A.
1996, MNRAS, 283, 1388
Madau, P., Pozzetti, L., & Dickinson, M. 1998, ApJ, 498, 106
Madau, P., Della Valle, M., & Panagia, N. 1998, MNRAS, 297, 17
Sawicki, M. J., Lin, H., & Yee, H. K. C. 1997, AJ, 113, 1
Shaver, P. A., Wall, J. V., Kellermann, K. I., Jackson, C. A., & Hawkins, M. R. S. 1996,
Nature, 384, 439
Silk, J., & Rees, M. J. 1998, A&A, 331, L1
Steidel, C. C., Giavalisco, M., Dickinson, M., & Adelberger, K. L. 1996, AJ, 112, 352
Webb, J. K. 1997, (private communication)
Williams, R. E., et al. 1996, AJ, 112, 1335
Zepf, S. E., Moustakas, L. A., & Davis, M. 1997, ApJ, 474, L1
This preprint was prepared with the AAS LATEX macros v4.0.
|
astro-ph/0308027 | 1 | 0308 | 2003-08-02T22:39:26 | IRAS PSCz v.s. 1.2-Jy Velocity and Density Fields: A Spherical Harmonics Comparison | [
"astro-ph"
] | We have used the two IRAS redshift surveys, 1.2-Jy (Fisher et al 1995) and PSCz (Saunders et al 2000), to model the linear velocity fields within a redshift of 8,000 km/s and have compared them in redshift space. The two velocity fields only differ significantly in their monopole components. The monopole discrepancy cannot be solely ascribed to shot-noise errors and incomplete sky coverage. The mismatch seems to arise from incompleteness of the PSCz catalog at fluxes $\le$ 1.2 Jy. The 1.2-Jy and PSCz higher order velocity multipoles, particularly the dipole and quadrupole components, appear to be consistent, suggesting that the dipole residuals found by Davis, Nusser and Willick (1996) when comparing 1.2-Jy and MarkIII velocity fields probably originates from the MarkIII velocities calibration procedure rather than from uncertainties in the model velocity field. Our results illustrate the efficiency of the spherical harmonics decomposition techniques in detecting possible differences between real and model velocity fields. Similar analyses shall prove to be very useful in the future to test the reliability of next generation model velocity fields derived from new redshift catalogs like 2dFGRS (Colless et al 2001), SDSS (York et al 2000), 6dF and 2MRS. | astro-ph | astro-ph |
DRAFT VERSION NOVEMBER 12, 2018
Preprint typeset using LATEX style emulateapj v. 26/01/00
IRAS PSCZ V.S. IRAS 1.2-JY MODEL VELOCITY FIELDS: A SPHERICAL HARMONICS COMPARISON.
LUÍS TEODORO1, ENZO BRANCHINI2 AND CARLOS S. FRENK3
1T-8, Theoretical Division, Los Alamos National Laboratory, Los Alamos, New Mexico, 87545, USA
2Dipartimento di Fisica, Università degli Studi Roma TRE, Roma, Italia
3Department of Physics, University of Durham, Science Laboratories, Durham DH1 3LE, England.
Draft version November 12, 2018
ABSTRACT
We have used the two IRAS redshift surveys, 1.2-Jy (Fisher et al. 1995) and PSCz (Saunders et al. 2000), to
model the linear velocity fields within a redshift of 8 000 km s- 1 and have compared them in redshift space. The
two velocity fields only differ significantly in their monopole components. The monopole discrepancy cannot be
solely ascribed to shot-noise errors and incomplete sky coverage. The mismatch seems to arise from incomplete-
ness of the PSCz catalog at fluxes ≤ 1.2 Jy. The 1.2-Jy and PSCz higher order velocity multipoles, particularly
the dipole and quadrupole components, appear to be consistent, suggesting that the dipole residuals found by
Davis, Nusser and Willick (1996) when comparing 1.2-Jy and Mark III velocity fields probably originates from
the Mark III velocities calibration procedure rather than from uncertainties in the model velocity field. Our results
illustrate the efficiency of the spherical harmonics decomposition techniques in detecting possible differences
between real and model velocity fields. Similar analyses shall prove to be very useful in the future to test the
reliability of next generation model velocity fields derived from new redshift catalogs like 2dFGRS (Colless et
al. 2001), SDSS (York et al. 2000), 6dF and 2MRS.
Subject headings: cosmology -- theory -- galaxies large-scale structure of the Universe -- large-scale dynamics.
1. INTRODUCTION
In the framework of linear gravitational instability theory and
linear biasing a comparison between the predicted and observed
peculiar velocities allows us to measure the so called β param-
eter
β ≡
.
(1)
Ω0.6
m
b
i.e. a combination of the mass density parameter, Ωm, and the
linear bias parameter, b, which relates the mass overdensity
field δ to the density fluctuations in galaxy counts δg, through
δg = bδ. Several different techniques have been used to compare
measured peculiar velocities (taken from all available catalogs:
Mark III Willick 1997a, SFI Haynes 1999, ENEAR da Costa et
al. 2000, SEcat Zaroubi 2002) to predict velocities that, in most
cases, have been obtained from two redshift surveys of IRAS
galaxies: the 1.2-Jy and the PSCz catalogs.
Most analyses have investigated the consistency between
model and observed velocities, while only a few of them have
been devoted in comparing different datasets or in studying the
consistency of model velocity fields. All recent works show
a general consistency between models and data (Willick et al.
1997, hereafter WSDK; Willick and Strauss 1998; Sigad et
al. 1998; Dekel et al. 1999; Branchini et al. 1999, hereafter
B99; Nusser et al. 2000; Zaroubi et al. 2002) with one no-
ticeable exception represented by the work of Davis, Nusser
& Willick (1996, hereafter DNW). Their analysis showed that
the velocity residuals between the Mark III peculiar veloci-
ties and the 1.2-Jy predictions display a significant spatial cor-
relation. Subsequent analyses by WSDK and Willick and
1
Strauss (1998) were performed using the same dataset, a sim-
ilar model velocity field but a different comparison technique,
called VELMOD, which allows for independent calibrations of
Mark III velocities in each of the Mark III sub-catalogs. As
a result the observed and model velocity fields were brought
into better agreement, at least within a redshift distance s =
7 500 km s- 1 and returned a calibration inconsistent with the
original one performed by Willick et al. (1997a).
While these results suggest that the DNW mismatch orig-
inates from calibration problems in the Mark III catalog, the
possibility that systematic errors in the 1.2-Jy model velocities
also contribute to the discrepancy has been completely over-
looked. In particular, we ignored whether the velocity fields
predicted from the deeper IRAS PSCz catalog provides a better
match to the Mark III velocities than the shallower 1.2-Jy cata-
log, considered by DNW.
The main goal of this work is to check whether the discrep-
ancies between the 1.2-Jy and Mark III velocities can be as-
cribed, at least to some extent, to model uncertainties. For
this purpose we compare both the overdensity and the radial
velocity fields modeled from the redshift space distribution of
1.2-Jy and PSCz galaxies. This work is meant to stress the
usefulness of spherical harmonics decomposition techniques in
assessing the adequacy of a model velocity (or density) field
through velocity-velocity (or density-density) comparisons.
In § 2 we describe how to compare velocity and overden-
sity fields inferred from different redshift surveys. The galaxy
redshift catalogs used in our analyses are described in § 3. In
sections 4 and 5 we analyse the two model overdensity fields
and radial velocity fields. Our main results are discussed in § 6
2
PSCz v.s. 1.2-Jy Velocity and Density Fields
and our main conclusions are presented in § 7.
2. SPHERICAL HARMONIC COEFFICIENTS DECOMPOSITION
The relations between an arbitrary scalar field ψ(s) and the
real -- valued spherical harmonic coefficients ψlm(s) are
ψlm(s) =Z ψ(s)Ylm(s)dΩ
∞
l
ψ(s) =
ψlm(s)Ylm(s),
Xl=0
Xm=- l
(2)
(3)
where Ylm(s){l = 0, .., ∞; m ≤ l} denote the well known real-
valued spherical harmonics, defined as in Baker et al. (1999,
hereafter BDSLS) and Bunn (1995) (see also Jackson 1999).
For a given l these functions are normalized to the value of Yl0
at the North Galactic Pole.
in quadrature over the (2l + 1)-ψ lm's, where {m = - l, ...,l}:
The amplitude of the l-multipole ψl(s) is defined by the sum
ψl(s) =" m=l
Xm=- l
1
2
.
ψ2
lm(s)#
2.1. Peculiar velocity and overdensity fields from the
distribution of galaxies in redshift space
(4)
FIG. 1. -- The 1.2-Jy smoothing width as a function of redshift. The dots
represent the value of σn at centre of the 52 redshift bins.
Nusser & Davis (1994) show that in the linear regime the
peculiar velocity field is irrotational not only in real space but
also in redshift space and thus can be expressed as the gradient
of a scalar velocity potential: v(s) = - ∇Φ(s). At a given redshift
s, Φ(s) and δ(s) can be expanded in spherical harmonics and
related to each other through a modified Poisson equation:
1
lm(cid:1)′ -
s2(cid:0)s2Φ′
1
1 + β
l(l + 1)Φlm
s2
=
β
1 + β(cid:18)δlm - 1
s
d lnφ
d lns
Φlm
′(cid:19) ,
(5)
where φ(s) and δlm(s) denote the sample selection function
and the spherical components of the overdensity field, respec-
tively. Here prime expresses d/ds. Solving this differential
equation requires computing the redshift space density field
from the observed galaxy distribution on a spherical grid us-
ing Gaussian cells of approximately equal solid angle. These
cells are equally distributed in longitude (64 bins) and are cen-
tered at the Gaussian-Legendre 32 point quadrature formula in
the range - 1 ≤ cos(b) ≤ 1, where b is the Galactic latitude.
In this work we measure redshifts in the Local Group frame,
and use 52 Gaussian radial redshift bins out to a distance of
s = 18 000 km s- 1. When comparing two or more galaxy cata-
logs the size of the radial bin is set by requiring equal spatial
resolution and the minimal shot noise constant throughout the
volume. We do this setting the distance between the centers
of the Gaussian cells at distance s equal to the average galaxy-
galaxy separation, σ(s) = [¯nφ(s)]- 1/3, in the sparser catalog. The
mean number density is estimated as
1
1
(6)
¯n =
V Xi
,
φ(si)
where the sum is over the galaxies contained within the volume
V , which, in this work, is a sphere of radius 8 000 km s- 1.
1 + δ(sn) =
Hernquist and Katz (1989) pointed out that there are two pos-
sible ways to estimate the smoothed density field: the "scatter"
and "gather" approches. In the former approch each particle is
distributed in space, and the density estimate at a given point is
the superposition of individual smoothing spheres. In the lat-
ter case one defines a smoothing lenght at a given point and
weights the particle in its neighborhood by the resulting kernel.
The Gaussian-smoothed galaxy overdensity field at a grid cell
centred around sn is given by:
N
1
(2π)3/2σ3
n
1
φ(si)
exp(cid:20)-
Xi
(sn - si)2
2σ2
n
(cid:21) ,
(7)
where the sum is over the N galaxies in the catalog, which
clearly is a "gather" approach. The smoothing width at a red-
shift s, used in this paper is σn = max[320, σ(s)] km s- 1. Since
in this work we compare the PSCz catalog with the sparser
1.2-Jy catalog, we use the σn obtained from the 1.2-Jy catalog
shown in Fig. (1). This smoothing scheme is tailored to keep the
shot-noise uncertainty roughly constant throughout the sample
volume and is similar to the optimal Wiener filtering procedure
used by Lahav et al. (1994) and Fisher et al. (1995b). More-
over, the our kernel has a compact support, which simplifies
the computational task.
Eqn. (5) is valid in the linear regime, where a one-to-one
mapping between distance and redshift is guaranteed. This as-
sumption is not valid in high density regions, such as cluster of
galaxies, where shell crossing may occur triple valued-regions
may apear. To deal with the latter effect, we adopt the same
procedure as Yahil et al. (1991) and collapse the fingers of God
associated with the six richest clusters in the sample listed in
Table 2 of that work.
In this work we will consider galaxies within smax = 20 000
Teodoro, Branchini & Frenk
3
km s- 1 with the spherical harmonics decomposition limited to
lmax = 15. Outlying smax galaxies are assumed to be distributed
uniformly according to the mean number density of the parent
catalogs.
3. DATASETS
The IRAS PSCz catalog contains ≈ 15 500 IRAS PSC galax-
ies with a 60 µm flux larger than 0.6 Jy. The average depth of
this survey is ≈ 100 h- 1 Mpc . We restrict our analysis to a sub-
sample of 11 206 galaxies within 20 000 km s- 1 from the Local
Group (LG, hereafter). The areas not covered by the survey
amount to 16.0% of the sky and are preferentially located near
the galactic plane, in the so-called Zone of Avoidance (Saun-
ders et al. 2000). The empty areas have been filled-in with the
cloning procedure described in Branchini et al. (1999) .
The 1.2-Jy catalog (Fisher et al. 1995a) contains 5 331 IRAS
galaxies with a 60 µm flux larger than 1.2-Jy within 20 000
km s- 1. This catalog has a slightly larger sky coverage of (≈
- 1 Mpc). The
87.6%) and a smaller median redshift (≈ 84 h
same B99 filling technique has been used to restore all sky cov-
erage.
Fisher et al. (1995b) have found that uncertainties in filling-
in the empty areas of the 1.2-Jy survey do not appreciably affect
the spherical harmonics coefficients with l . 15. Similar con-
clusions are also valid for the IRAS PSCz survey (Teodoro et
al. 1999).
The selection functions of the two surveys, that quantify the
probability of a galaxy at distance s, have been determined as
maximum-likelihood fits to the smooth function
φ(s) =( (cid:0) rr
r(cid:1)2α(cid:16) r2
1
⋆
r2
⋆
+r2
s
+r2(cid:17)β
, s ≥ H0 rs,
, otherwise,
(8)
1 is expressed in h- 1 Mpc (Yahil et al. 1991).
where r = s/H0
The best-fit parameters, determined using the galaxies within
8 000 km s- 1 and assuming a flat universe with Ωm = 1, are
shown in Tab. (1). The quantity rs in Eqn. (8) accounts for
the incompleteness of faint galaxies in the innermost regions of
the catalogs (Rowan-Robinson et al. 1990, Yahil et al. 1991).
4. THE IRAS PSCz AND 1.2-Jy FIELDS OVERDENSITY
The first step of our analysis consists of applying Eqn. (7) to
compute the IRAS 1.2-Jy and PSCz overdensities onto a spheri-
cal grid of radius 18 000 km s- 1 and lmax = 15 with the smooth-
ing scheme given by Eqn. (7).
4.1. Overdensity Maps
In Figs. (2 -- 4) we show the Aitoff projections of the PSCz (top
panels) and 1.2-Jy (middle panels) galaxy overdensity fields in
three spherical shells centred at s = 1 000, 3 000 and 5 000 km s- 1,
respectively. The thickness of each shell is defined by the Gaus-
sian width at the distance of the shell. All maps are plotted in
Galactic coordinates. The bottom panels show the map of the
this article we write Hubble's constant as H0 = 100h
1Throughout
km s- 1 Mpc- 1
FIG. 2. -- PSCz (top), 1.2-Jy (middle) and residual (bottom) overdensity
fields in the radial shell centred at s = 1 000 km s- 1. The thick-line contour
indicates δ = 0 or δres = 0 (bottom). The light- and dark-gray contours show
over- and under-dense regions, respectively. In the top and middle panels the
contours are equally -- spaced by ∆δ = 1.0. The bottom panel the density resid-
uals are shown in steps of ∆δres = 0.125.
overdensity residuals δres(s) = δ0.6(s) - δ1.2(s), where the sub-
scripts 0.6 and 1.2 refer to PSCz and 1.2-Jy surveys, respec-
tively. In all plots the thick line indicates the zero overdensity
(or residual) contour. The overdensity (residual) contour spac-
ing ∆δ (∆δres) depends on the shell's depth and is indicated in
the Figure captions.
Detailed cosmographic studies of the
structures in the nearby universe have already been performed
using both the distribution of 1.2-Jy (Strauss and Willick 1995)
and PSCz galaxies2 and will not be repeated here. We only
stress that the main cosmic structures like the Virgo Cluster (l ≈
240◦, b ≈ 75◦, s = 1000 km s- 1), the local void spanning the re-
gion 330◦ . l . 120◦, - 90◦ . b . 30◦ at s = 1000 km s- 1, the
Hydra-Centaurus complex (l ≈ 270◦, b ≈ 20◦, s = 3000 km s- 1)
and (l ≈ 300◦, b ≈ 20◦, s = 3000 km s- 1) and the Perseus-
Pisces supercluster (l ≈ 140◦, b ≈ - 25◦, s = 3000 km s- 1) are
detected in both surveys.
The residuals, calculated at the same gridpoints as the den-
2See http://www-astro.physics.ox.ac.uk/∼wjs/pscz.html
4
PSCz v.s. 1.2-Jy Velocity and Density Fields
SELECTION FUNCTION PARAMETERS FOR IRAS CATALOGS.
TABLE 1
60µm Ngal a
Catalog f lim
[Jy]
α
β
γ
r b
s
r⋆
¯n
[ h- 1 Mpc] [ h- 1 Mpc] [h3Mpc- 3]
0.6 13 364 0.54 1.80 ...
PSCz
PSCz 0.7 0.7 11 170 0.57 1.80 ...
PSCz 1.2 1.2
5 519 0.43 1.86 ...
1.2
1.2-Jy
6 010 0.43 1.86 ...
c
0.6 13 364 0.99 3.45 1.93
PSCz evo.
6.0
6.0
5.0
5.0
6.0
87.00
76.35
50.40
50.60
76.02
6.12 ·10- 2
6.38 ·10- 2
4.62 ·10- 2
4.67 ·10- 2
5.76 ·10- 2
a Includes synthetic objects.
b Parameter kept constant in the likelihood maximization.
c The selection function parameters have been taken from Canavezes et al. (1999)
FIG. 3. -- Same as in Fig. (2). The shell is now centred at s = 3 000
km s- 1 and the contour spacings are ∆δ = 0.5 and ∆δres = 0.125.
FIG. 4. -- Same as in Fig. (2). The shell is now centred at s = 5 000
km s- 1 and the contour spacings are ∆δ = 0.5 and ∆δres = 0.125.
Teodoro, Branchini & Frenk
5
THE PARAMETERS OF THE δ1.2 - δ0.6 AND u1.2 - u0.6 LINEAR REGRESSIONS.
TABLE 2
Regression Ngrid
Ndof
A a
B
χ2
eff/Ndof
δ1.2 - δ0.6
u1.2 - u0.6
112 552 156.36
112 552 . 156.36
- 0.011 ± 0.014
60.0 ± 5.9
1.025 ± 0.036
1.040 ± 0.019
0.78
0.88
Notes:
a For the u - u comparison A is expressed in km s- 1.
Column 2: Ngrid, the number of gridpoints used for the regression;
column 3: Ndof, the number of independent volumes;
column 4: A, the zero -- point offset in the linear regression, and its 1 - σ error;
column 5: B, the slope of the linear regression, and its 1 - σ error;
column 6: χ2
eff/Ndof from the linear regression.
i. This expression is a generalization of Eqn. (12) in Dekel et
al. (1993) to the case of a grid with variable mesh-size since
Rs,i depends on the gridpoint position. The new statistics χ2
eff =
(Ngrid/Neff)χ2 is approximately distributed as a χ2 with Ndof =
Neff - 1 degrees of freedom and can be applied to infer the values
of A, B and their errors.
Fig. (5) shows the δ1.2 vs. δ0.6 scatter diagram obtained us-
ing all 112 552 gridpoints with s < 8 000 km s- 1 and b > 8o.
Within these limits the estimated overdensities are little affected
by uncertainties in the filling-in procedure, and we expect that
sity fields, seem to indicate that, as expected, the largest residu-
als are located near the galactic plane where uncertainties in the
filling-in procedure dominate the errors. In the inner shell, over-
densities in the 1.2-Jy galaxy distribution appear to be larger
than the PSCz ones while in the most external shell the situa-
tion is reversed.
4.2. Density Residuals
A way of quantifying possible systematic differences be-
tween the two fields is that of performing a point-to-point com-
parison.
In absence of systematic errors we expect the two
IRAS overdensity fields to be linearly related through
δ1.2 = Bδ0.6 + A
(9)
with A = 0 and B = 1. δ0.6 and δ1.2 are estimated at Cartesian
gridpoint positions. A zero -- point offset A 6= 0 indicates discrep-
ancies in the mean densities of the two catalogs while a slope
B 6= 1 reveals the presence of systematic errors. If the values of
δ0.6 and δ1.2 are independent and their errors are Gaussian then
we can compute the χ2 statistics
χ2 =
Ngrid
Xi=1
(δ1.2 - A - Bδ0.6)2
+ B2σ2
(cid:0)σ2
0.6(cid:1)
1.2
,
(10)
where σ0.6 and σ1.2 indicate the typical errors in δ and Ngrid is
the total number of points in the comparison. However, since
the gridspacing (250 km s- 1) is smaller than the smoothing length
(≥ 320 km s- 1), not all values of δ0.6 (or δ1.2) are independent.
Indeed, the effective number of independent points used in the
χ2 statistics, Neff is given by
N- 1
eff = Ngrid
- 2
exp(- r2
i j/2R2
s,i),
(11)
Ngrid
Ngrid
Xi=1
Xj=1
where ri j is the separation between gridpoints i and j and Rs,i
is the smoothing radius of the Gaussian kernel at the gridpoint
FIG. 5. -- 1.2-Jy vs. PSCz overdensity. Only 1883 randomly selected grid-
points with b > 8◦ and within s = 8 000 km s- 1 are shown in the scatter plot.
The continuous line represents the linear best fit characterized by the zeropoint
A and slope B indicated in the panel.
6
PSCz v.s. 1.2-Jy Velocity and Density Fields
consistent with recent determinations based on several velocity-
velocity comparisons (e.g. Nusser et al. 2000), a recent density-
density analysis (Zaroubi et al. 2002) and the study of mean
relative velocities of galaxy pairs (Juszkiewicz et al. 2000).
5.1. Radial Peculiar Velocity Maps
To obtain maps of the radial velocity field we have computed
the gradient of the velocity potential projected along the line of
sight: u(s) = s · ∇Φ at the same gridpoint positions used for the
analysis of the density field.
The top end mid panels of Figs. (7-9) are similar to those of
figs. 2 -- 4 and show the radial velocity maps in the same shells.
The bottom panels show the maps of the radial velocity residu-
als ures(s) ≡ u0.6(s) - u1.2(s). The thick line represents the zero
radial velocity (residual) contour. The velocity (residual) con-
tour spacing ∆u (∆ures) varies with the redshift and is indicated
in the Figure captions.
The main feature seen at all redshifts in both 1.2-Jy and
PSCz radial velocity maps is the dipole pattern resulting from
the LG reflex motion toward the CMB apex (l ≈ 276◦,b ≈ 30◦,
Kogut et al.1993).
Local deviations from this pattern arise from peculiar mo-
tions, like the infall onto Perseus-Pisces superclusters at (s ≈
3000 km s- 1, l ≈ 120◦, b ≈ - 40◦). The largest velocity resid-
uals are located near the galactic plane, but their contours are
broader due to the intrinsic non-locality of the peculiar velocity
field. The most striking feature is perhaps the fact that velocity
residuals, that are positive across a large fraction of the sky in
the first shell, becomes mostly negative in the remaining two
shells, suggesting possible systematic errors in either PSCz or
1.2-Jy velocity models.
shot-noise dominates the total error budget. To estimate the
shot-noise errors affecting both PSCz and 1.2-Jy overdensity
fields we have generated 100 bootstrap realizations of the ob-
served PSCz and 1.2-Jy surveys. These are obtained by replac-
ing each galaxy (including the 'synthetic' galaxies used to fill-
in the unobserved regions) with a number of objects drawn from
a Poisson deviate with mean unity. Shot noise errors in δ at the
generic gridpoint were set equal to the dispersion over the 100
realizations. The solid line in the plot is the best linear fit ob-
tained from the minimization of χ2
eff. As shown in Tab. (2), the
slope of the line, B = 1.025 ± 0.036, is consistent with unity,
indicating no systematic mismatch. On the other hand the zero-
point value A = - 0.011 ± 0.014 shows no systematic difference
in the mean density of the two IRAS catalogs.
The scatter around the best -- fitting line, σδ = 0.16 is similar to
average shot-noise error in the 1.2-Jy density field (σδ1.2
SN = 0.15).
The parameter S = χeff/Ndof = 0.78 indicates that bootstrap er-
rors slightly overestimate the true uncertainties.
4.3. Radial Density Contrast
An alternative way to characterize possible discrepancies be-
tween the two IRAS density fields is that of computing the dif-
ferential radial density contrast in shells with identical thick-
ness ∆s for both galaxy distributions:
ρ(s)
¯ρ
=
1
V ¯n
W (s,si, ∆s)
φ(si)
,
(12)
Np
Xi=1
where the window associated to the spherical shell Wδρ/ ¯ρ(s,si,
∆s) is the Heaviside function H(s + ∆s/2 - si)H(si + ∆s/2 - s),
si is the redshift of the object i, V is the volume used to com-
pute the mean number density ¯n and the sum runs over the Np
galaxies of the catalog. Fig. (6) shows the quantity ρ(s)/ ¯ρ for
1.2-Jy (light, dotted histogram) and PSCz (thick, continuous
histogram) galaxies within s = 15 000 km s- 1, computed in in
spherical shells of thickness ∆s = 200 km s- 1. Although the
general shapes of two radial density contrasts are similar (with
a bump at s ∼ 1 100 km s- 1 due to Virgo and Fornax clusters,
a dip at s ∼ 3 500 km s- 1 induced by the Sculptor Void and a
second prominent peak at s ∼ 5 000 km s- 1, at the same red-
shift as the Perseus-Pisces supercluster and the Great Attrac-
tor), a discrepancy between the two curves exists in the range
2 000 . s . 6 000 km s- 1 where the PSCz density is larger than
the 1.2-Jy one.
The two smoothed curves in Fig. (6) show the monopole
terms of the two overdensity fields δl=0,m=0 as a function of
radius, obtained from the spherical harmonics decomposition.
Both curves interpolate the histograms rather well (hence show-
ing the adequacy of our spherical harmonics analysis) and con-
firm the existence of a PSCz vs. 1.2-Jy density mismatch be-
tween the two catalogs.
5. THE IRAS PSCz AND 1.2-Jy PECULIAR VELOCITY FIELDS
To perform a similar analysis on the 1.2-Jy and PSCz model
velocity fields we have integrated Eqn. (5) using β = 0.6, a value
FIG. 6. -- Differential radial density contrast for PSCz (thick line histogram)
and 1.2-Jy galaxies (thin line histogram). The monopole of the PSCz and 1.2-
Jy density fields are shown with smooth solid and dashed lines, respectively.
Teodoro, Branchini & Frenk
7
FIG. 7. -- PSCz (top), 1.2-Jy (middle) and residual (bottom) radial velocity
fields in the spherical shell centred at s = 1 000 km s- 1. The thick-line contour
indicates δu = 0 or ∆ures = 0. The light- and dark-gray contours show regions
characterized by outflows or infall motions. In the top and middle panels the
contours are equally -- spaced by ∆u = 200 km s- 1. The bottom panel shows
the velocity residuals in steps of ∆δres = 25 km s- 1.
5.2. Velocity Residuals
To perform a more quantitative comparison between the two
radial velocity fields we have repeated the analysis of § 4.2 and
performed a linear regression between u1.2 and u0.6 by minimiz-
ing
Ngrid
Xi=1 (cid:0)u1.2 - A - Bσu,0.6(cid:1)2
u,0.6(cid:17)
(cid:16)u2
+ B2σ2
1.2
χ2 =
,
(13)
where Ngrid runs over the same points used for the density-
density comparison. Linear velocities are obtained by integrat-
ing the mass density fields over large scales causing them to
be correlated on scales larger than the radius of the Gaussian
kernel. This means that the number of independent gridpoint
Ne f f is smaller than the estimate given by Eqn. (11). Intrinsic
large scale smoothing also affects the velocity error estimate,
σu since uncertainties in modeling the galaxy distribution in the
unobserved areas now add to shot noise errors.
FIG. 8. -- Same as in Fig. (7). The shell is now centred at s = 3 000 km s- 1and
the contour spacings are ∆u = 200 km s- 1and ∆ures = 50 km s- 1
The shot noise errors have been computed using the same
bootstrap resampling technique described in § 4.2. Uncertain-
ties in the filling-in procedure have been evaluated from 20 in-
dependent PSCz and 1.2-Jy mock catalogs extracted from the
N-body simulations performed by Cole et al. (1997) simulating
a volume of ≈ 170 h- 1 Mpc in an Einstein-de Sitter universe
with a non-vanishing (ΩΛ = 0.7) cosmological constant and
density fluctuations normalized to the observed cluster abun-
dance (Eke, Cole & Frenk 1996 ). Model velocity fields from
each IRAS mock catalog were compared with velocities ob-
tained from ideal, all-sky IRAS mock catalogs. The variance
over 20 mocks at each gridpoint quantifies the uncertainties
in the filling procedure. Total uncertainties were obtained by
adding in quadrature the two errors.
The results of the linear regression are summarized in Tab. (2)
which shows the slope of the best fitting line B = 1.040 ± 0.019
and the value of the zero-point, A = 60.0 ± 5.9 km s- 1. While
the slope is still consistent with unity, at the 2.1-σ confidence
level, the zero point is significantly different from zero, hence
corroborating the evidence for a mismatch in the average densi-
ties of PSCz and 1.2-Jy catalogs. The dispersion around the fit
8
PSCz v.s. 1.2-Jy Velocity and Density Fields
FIG. 10. -- 1.2-Jy vs. PSCz radial velocities computed at 1883 randomly
selected gridpoint positions with b > 8◦ and s ≤ 8 000 km s- 1 inferred radial
velocity field (x-axis) versus that inferred from the 1.2-Jy survey (y-axis). The
parameters of the best fitting line (A, B) the scatter σu are indicated.
The monopole terms of PSCz (continuous line) and 1.2-
Jy (dashed line) velocity fields are plotted in Fig. (11) as a func-
tion of the redshift. The hatched and dashed regions represent
the 1-σ uncertainty strip, computed from the mock catalogs and
the bootstrap resampling procedure.
FIG. 11. -- Monopole coefficients of 1.2-Jy (dashed) and PSCz (continuous)
radial velocity fields. 1 - σ errors are represented by the hatched (shaded)
areas.
FIG. 9. -- Same as in Fig. (7). The shell is now centred at s = 5 000 km s- 1and
the contour spacings are ∆u = 200 km s- 1and ∆ures = 50 km s- 1.
is, σu = 88 km s- 1 is similar to the typical velocity error com-
puted from the IRAS mocks and the bootstrap resampling cat-
alogs (σu0.6 = 71 km s- 1 and σu1.2 = 140 km s- 1). The parameter
χeff/Ndof is less than unity which either indicates that errors
are overestimated or that the value Ne f f , taken from Eqn. (11),
does significantly overestimate the true number of degrees of
freedom.
5.3. Velocity Multipoles
The spherical harmonics decomposition allows us to bet-
ter investigate and characterize possible discrepancies between
1.2-Jy and PSCz velocity fields. Here we are only interested
in the first three (l = 0,1,2) components, corresponding to the
monopole, dipole and quadrupole terms, which have already
been investigated in previous analyses (e.g. BDSLS). It is worth
stressing that the monopole and dipole components at a given
redshift are only sensitive to the mass distribution within the
corresponding distance and thus can be directly related to the
structures within the volume considered, while the quadrupole
components are also sensitive to the mass distribution beyond
such a distance [see Eqn. (9) in Nusser & Davis 1994].
Teodoro, Branchini & Frenk
9
FIG. 12. -- The velocity dipole coefficients, u1m(s){m = - 1, 0, 1}, inferred
from PSCz (continuous) and 1.2-Jy (dashed) and their associated 1-σ uncer-
tainties. The three Galactic Cartesian components GX, GY , GZ are shown in
the top-left, top-right and bottom-right panel, respectively. The total ampli-
tude is shown in the bottom-left panel. The dotted line refers to the dipole of
PSCz galaxies with flux f60µm ≥ 1.2 Jy.
Both velocity monopoles display the same general features,
as expected given the similarities between the density monopoles
δ00(s) shown in Fig. (6). There is a significant radial in-
flow around s = 1 800 km s- 1 corresponding to the infall mo-
tion toward the Virgo-Fornax complex, an outflow in the range
1 800 . s . 4 000 km s- 1 due to the presence of the local void
and another infall beyond this radius, due to the Hercules, Hy-
dra, Centaurus and Perseus-Pisces superclusters. Differences
between the two velocity monopoles are detected with a signif-
icance level larger than 1-σ at very small radii (s < 1200 km s- 1)
and in the range 3500 km s- 1< s < 7500 km s- 1.
Fig. (12) shows the total amplitude (lower left panel) and
the three Cartesian components (remaining panels) of the two
IRAS velocity dipoles. The total amplitude of the PSCz dipole
(continuous line) is systematically smaller that the 1.2-Jy one
(dashed line) but the discrepancy is well within the expected
errors, in agreement with previous studies (e.g. Schmoldt et
al. 1999).
Finally, Fig. (13) shows the five components of the PSCz (con-
tinuous line) and 1.2-Jy (dashed line) velocity quadrupoles,
u 2m(s) (where m = - 2, ...,2) along with total amplitude (bottom-
left panel). As for the dipole case, the two sets of quadrupole
components agree to within 1-σ apart from a small disagree-
ment in the the m = +2 and m = - 2 components beyond s = 7000
km s- 1. Indeed, beyond s ≈ 6 000 km s- 1 the amplitude of the
PSCz quadrupole decreases, while the 1.2-Jy one increases
monotonically.
It is worth noticing that our 1.2-Jy velocity multipoles agree
FIG. 13. -- Quadrupole coefficients u 2m(s){m = - 2, ..., 2} derived from
PSCz (solid line) and 1.2-Jy (dashed) model velocity fields. Shaded regions in-
dicate the 1 - σ uncertainty regions. The dotted line represents the quadrupole
components for PSCz galaxies with flux f60µm ≥ 1.2 Jy.
with those computed by BDSLS except that their multipole
components exhibit sharper features at small radii (s < 500
km s- 1). This mismatch derives from our use of a Gaussian
smoothing kernel of radius 330 km s- 1 which is somewhat larger
than that applied by BDSLS.
6. DISCUSSION.
Perhaps the most surprising result of our analysis is the dis-
agreement between the PSCz and 1.2-Jy velocity monopoles in
the 1 800 . s . 4 000 km s- 1 range. Before investigating the
possible causes of this mismatch, it is worth noticing that none
of our IRAS monopoles is consistent with an outward flow of
magnitude ≈ 400 km s- 1 at a distance of 7 000 km s- 1 like the
one detected by Zehavi et al. (1998) using a sample of 44 Type
Ia supernovae. That result is also at variance with the velocity
model obtained from the distribution of ORS galaxies obtained
by BDSLS which means that, if confirmed, it would be difficult
to reconcile with the Gravitational Instability scenario.
A possible clue to understand the origin of the monopole
mismatch is provided by the analysis of BDSLS that returns
a velocity monopole somewhere between the PSCz and 1.2-
10
PSCz v.s. 1.2-Jy Velocity and Density Fields
Jy ones in the range of interest. This seems to suggest that
our monopole discrepancy originates from systematic errors in
one of the IRAS catalogs, possibly some incompleteness of the
largest catalog at faint fluxes.
The results of Tadros et al. (1999) seem to confirm that
PSCz catalog might be statistically incomplete at fluxes f60µm ≤
0.7 Jy. To test whether this is indeed the case we have extracted
a brighter PSCz subsample (called PSCz0.7) by discarding all
object with f60µm ≤ 0.7 Jy and have repeated our spherical har-
monics decomposition analysis. The results, shown in Fig. (14)
indicate that the PSCz0.7 velocity monopole (dotted line) basi-
cally coincide with that of the complete sample. It is only when
discarding galaxies with fluxes smaller than 1.2 Jy that the ve-
locity monopole (dashed line) agrees with the 1.2-Jy one.
Another possibility of understanding the monopole mismatch
is that of invoking a strong evolution of IRAS galaxies as a
function of the redshift, which we have neglected in our calcu-
lations so far. Roughly speaking evolution can be distinguish
into two different types. Pure density evolution changes only
the normalization of the luminosity function while its shape is
invariant with redshift. Pure luminosity evolution on the other
hand describes a possible change of the intrinsic luminosities of
the sources but not their number density. Both forms of evolu-
tion occur simultaneously in the Universe. For instance, galaxy
merging reduces the number density of galaxies causing thus
evolution. Besides, such events also cause different luminosi-
ties of the final systems and hence leading to luminosity evo-
lution. Other physical mechanism for galaxy evolution can be
devised. Some evidences for a sizable evolution of IRAS galax-
ies have been reported by a number of authors, although there
seem to be much controversy about the amplitude of the effect
(see Springel 1996 and references therein).
Canavezes et al. (1998) expressed evolution in terms of a
generalization of the selection function introduced by Yahil et
al. (1991) given by
φ(s)evol =( (cid:0) rs
1
⋆
rγ
⋆
r(cid:1)α(cid:16) rγ
+rγ
s
+rγ(cid:17)β/γ
, s ≥ H0rs,
, otherwise.
(14)
Springel (1996) showed that it indeed provides a rather accurate
modeling of both IRAS data-sets. The estimated parameters us-
ing a maximum-likelihood procedure are listed in Tab. (1) and
are denoted by PSCzevol. The effect of evolution are shown in
Fig. (14) in which the velocity monopole term derived from
the PSCz galaxy distribution in which evolutionary effects have
been accounted for through the new selection function φ(s)evol
(dot -- dashed line) is plotted against the unevolved model. The
evolutionary effects are very minor, as was somewhat expected
given the locality of our galaxy sample, and do not help in re-
ducing the PSCz vs. 1.2-Jy velocity monopole mismatch.
Overall, our analysis seems to indicate that the discrepancy
between 1.2-Jy and PSCz velocity monopoles cannot be as-
cribed to errors in the modeling of the selection function which,
as we have seen, would not change the monopole terms appre-
ciably. Catalog incompleteness at very low fluxes has to be
ruled out too. We can only conclude that either the PSCz in-
FIG. 14. -- Velocity monopoles for the 1.2-Jy (dark gray line) PSCz (light-
gray), PSCz0.7 (dot-dashed), PSCz1.2 (long-dashed) PSCzevol (dotted) sam-
ples.
completeness extend at objects with fluxes comparable to 1.2
Jy or that the monopole mismatch does reflect a genuine differ-
ence in the average density of faint vs. bright IRAS galaxies at
s ≤ 4000 km s- 1.
A second important result found in our analysis is that the
1.2-Jy and PSCz dipoles agree within the 1-σ errors. This
means that the use of the PSCz rather than the 1.2-Jy catalog
does not help in reducing the Mark III-IRAS dipole residuals
found by Davis, Nusser & Willick (1996). Indeed, both GX
and GY components of the PSCz velocity dipole are much too
large with respect to Mark III ones (see Fig 15 of Davis, Nusser
& Willick 1996). Similar results were also obtained from the
ORS vs. Mark III dipole velocity comparison (BDSLS). We
conclude that the discrepancy between the Mark III and 1.2-
TABLE 3
SELECTION FUNCTION PARAMETERS IN THE IRAS MOCK
CATALOGS.
Mock catalog
α
β
rs
[ h- 1 Mpc]
r⋆
[ h- 1 Mpc]
PSCz
1.2-Jy
0.53 1.90
0.48 1.79
10.90
5.00
86.40
50.40
Teodoro, Branchini & Frenk
11
ple in redshift space. Our model predictions are based on
the Nusser & Davis (1994) spherical harmonics decomposition
technique and assumes Zeld'ovich approximation and linear bi-
asing.
The IRAS overdensity fields are reconstructed with a smooth-
ing proportional to the 1.2-Jy inter-galaxy separation and ex-
hibit the same general features corresponding to the known cos-
mic structures in the local universe. The model velocity field,
computed in the LG frame, has a dipole -- like appearance, with
galaxies moving away from us toward Perseus-Pisces region
and approaching the LG from the opposite direction.
The analysis of the velocity multipoles has revealed a signif-
icant discrepancy between the PSCz and 1.2-Jy velocity mono-
poles in the range 1 800 . s . 4 000 km s- 1 in which PSCz radial
velocities are systematically larger than 1.2-Jy ones, that cannot
be explained by uncertainties in the IRAS selection functions
or incompleteness of the PSCz catalog at faint fluxes. Also,
the mismatch cannot be ascribed to errors derived from shot --
noise and treatment of unobserved areas since both have been
accounted for with the help of realistic mock IRAS catalogs and
the extensive application of bootstrap resampling techniques to
both IRAS datasets.
Both IRAS dipole and quadrupole moments are in good
agreement within the sampled volume apart from a small mis-
match in the u10, u2,- 2 and u22 quadrupole components beyond
s = 7 000 km s- 1. The agreement between the dipole moments
and their consistency with the velocity dipole computed by BD-
Jy velocity fields cannot be alleviated by reducing the shot noise
errors of the model velocity field.
The agreement between the PSCz, 1.2-Jy and ORS dipoles
is reassuring since it guarantees that uncertainties in model ve-
locity fields arising from within the galaxy samples that derive
from incompleteness in one (or more) catalogs, incorrect treat-
ment of non-linear velocities or non-linear non-uniform biasing
do not generate systematic errors apart from a the monopole
mismatch in the range 1 800 km s- 1≤ s ≤ 4 000 km s- 1.
Unlike the case of monopole and dipole moments, possible
differences between the 1.2-Jy and PSCz quadrupole compo-
nents can also be attributed to incorrect determination and di-
lute sampling of the density field beyond the sample. The fact
that 1.2-Jy and PSCz velocity quadrupole agree within the ex-
pected errors suggests that differences in the quadrupole com-
ponents of the two fields can be understood in terms of shot
noise errors and uncertainties in filling the empty regions.
With this respect it is worth investigating whether the denser
sampling of the PSCz catalog at large radii allows a better
modeling of the large scale contribution to the model veloc-
ity field. Indeed, when comparing the model 1.2-Jy velocity
field to Mark III velocities in the framework of the VELMOD
analysis, WSDK have found that velocity residuals exhibit a
quadrupole pattern of the form uQ = VQ r · r, where VQ is a
traceless, symmetric 3 × 3 tensor. They have attributed this
"VELMOD quadrupole" residuals to uncertainties in modeling
the IRAS density field beyond 3 000 km s- 1. In particular they
found that the most important sources of uncertainties were the
smoothing procedure (based on a Wiener Filtering technique)
and the shot noise errors, with the errors deriving from having
ignored mass inhomogeneities beyond 12,000 km s- 1 playing
a minor role. Using PSCz instead of 1.2-Jy catalog reduces the
shot noise errors in the density field beyond 3000 km s- 1 and
thus should improve the agreement with the Mark III velocities.
In other words one would expect the quadrupole of the resid-
ual velocity field u>
1.2, where one only takes in account
0.6
the mass beyond 3000 km s- 1, to be similar to the "VELMOD
quadrupole" of WSDK.
- u>
- u>
In Fig. (15) we compare the VELMOD quadrupole at s =
2000 km s- 1 (rescaled from β = 0.492 to β = 0.6, upper panel) to
the u>
1.2 quadrupole residual (lower panel). The VELMOD
0.6
quadrupole reaches its maximum value at (l ≈ 165◦, b ≈ 55◦)
and in the opposite direction of the sky. The u>
1.2 resid-
0.6
ual quadrupole exhibits a rather different pattern, reaching its
maximum amplitude at (l ≈ 156◦, b ≈ - 74◦). The value of the
quadrupole components of 1.2-Jy and PSCz as well as those of
1.2 quadrupole residuals computed at
the VELMOD and u>
0.6
a distance of 2000 km s- 1 are shown in Tab. (4). This com-
parison is at best qualitative since (i) VELMOD quadrupole is
computed in real space while both IRAS quadrupoles refer to
redshift space insted, and (ii) the smoothing scheme used in
WDKS is different from with ours.
- u>
- u>
7. CONCLUSIONS.
We have compared the model velocity and density fields of
the PSCz survey to those derived from the sparser 1.2-Jy sam-
- u>
FIG. 15. -- VELMOD (upper panel) and u>
1.2 (lower panel) quadrupole
0.6
residuals at s = 2 000 km s- 1. Stars denote outflowing from us (LG frame); cir-
cles denote infall to us. The size of the symbol is proportional to the amplitude
of the flow (notice the different velocity scales in the two panels). The max-
imum amplitude of the "VELMOD quadrupole" (top panel) is ≈ 181 km s- 1,
while for u>
0.6
1.2 is ≈ 23 km s- 1.
- u>
12
PSCz v.s. 1.2-Jy Velocity and Density Fields
SLS from the ORS catalog suggests that deviations from ho-
mogeneous linear biasing prescriptions have to be small and
cannot be advocated to explain the aforementioned mismatch
between the IRAS monopoles.
Another remarkable consequence of the agreement among
the various dipoles is that the 1.2-Jy-Mark III residuals cannot
be alleviated by using the PSCz catalog instead of the 1.2-Jy or
the ORS ones. In fact, this suggests that the origin of the 1.2-
Jy-Mark III dipole resides in the Mark III calibration procedure;
a suggestion corroborated by the fact that the dipole mismatch
disappears when adopting the alternative VELMOD procedure
to calibrate -Mark III velocities (WDSK). The VELMOD anal-
ysis requires a sizable external quadrupole component within
3000 km s- 1. This quadrupole accounts for uncertainties er-
rors in the 1.2-Jy model density field beyond that distance. We
found that using the PSCz sample provides part of the required
external tidal field.
As a final remark, we would like to stress that the present
analysis illustrates once more the importance of spherical har-
monics decomposition techniques to compare observed and
model density and velocity fields obtained from all-sky sam-
ples, to reveal possible data-model or model-model inconsisten-
cies and to assess model adequacy. These techniques will prove
very useful when the next generation of all-sky redshift surveys
and peculiar velocity catalogs [e.g. the 6dF (http:www.mso.anu.
edu.au/6dFGS) and 2MRS (http:cfa-www.harvard.edu/huchra/
2mass) datasets] will become available.
ACKOWLEDGEMENTS
We thank Adi Nusser for enlightening discussions. LT was
partly funded by FCT (Portugal) under the grants PRAXIS XXI
/BPD/16354/98, PRAXIS/C/FIS/13196/98 and POCTI/1999/
FIS/36285. LT aknowledges the hospitality of the Aspen Cen-
ter for Physics and the Kvali Institute for Theoretical Physics
where much of the work was completed.
TABLE 4
QUADRUPOLE COMPONENTS AT 2 000 km s- 1
VELMOD PSCz
[ km s- 1]
[ km s- 1]
1.2-Jy
[ km s- 1]
PSCz - 1.2-Jy
[ km s- 1]
VQ (x,x)...
VQ (y,y)...
VQ (z,z)...
VQ (x,y)...
VQ (x,z)...
VQ (y,z)...
45
44
-89
18
138
-29
57
-22
-35
-150
57
-106
69
-15
-54
-156
48
-102
-13
-7
20
6
10
-5
NOTE. -- (1) We have rescaled the "VELMOD quadrupole" compo-
nents presented in WSDK's Tab. (2) to the numerical value of β = 0.6.
REFERENCES
Baker, J. and Davis, M. and Strauss, M. A. and Lahav, O. and Santiago, B.,
1999, ApJ, 508, 6
Branchini, E. and Teodoro, L. and Frenk, C. S. and Schmoldt, I. and Efstathiou,
G. and White, S. D. M. and Saunders, W. and Sutherland, W. and Rowan-
Robinson, M. and Keeble, O. and Tadros, H. and Maddox, S. and Oliver, S.,
1999, MNRAS,308, 1
Bunn, E. F., 1995, PhD thesis, University of California, Berkeley
Canavezes, A. and Springel, V. and Oliver, S. J. and Rowan-Robinson, M. and
Keeble, O. and White, S. D. M. and Saunders, W. and Efstathiou, G. and
Frenk, C. S. and McMahon, R. G. and Maddox, S. and Sutherland, W. and
Tadros, H., 1998, MNRAS, 297, 777
Cole, S., Weinberg, D. H., Frenk, C. S., and Ratra, B.,1997, MNRAS, 289, 37
Colless, M. and Dalton, G. and Maddox, S. and Sutherland, W. and zes Norberg,
P. and Cole, S. and Bland-Hawthorn, J. and Bridges, T. and Cannon, R. and
Collins, C. and Couch, W. and Cross, N. and Deeley, K. and De Propris, R.
and Driver, S. P. and Efstathiou, G. and Ellis, R. S. and Frenk, C. S. and
Glazebrook, K. and Jackson, C. and Lahav, O. and Lewis, I. and Lumsden,
S. and Madgwick, D. and Peacock, J. A. and Peterson, B. A. and Price, I. and
Seaborne, M. and Taylor, K., 2001, MNRAS, 328, 1039
da Costa, L. N. and Bernardi, M. and Alonso, M. V. and Wegner, G. and
Willmer, C. N. A. and Pellegrini, P. S. and Rité, C. and Maia, M. A. G.,
2000, ApJ, 120, 95
Davis, M. and Nusser, A and Willick, J., 1996, ApJ, 473, 22
Dekel, A. and Bertschinger, E. and Yahil, A. and Strauss, M. A. and Davis, M.
and Huchra, J. P., 1993, ApJ, 412, 1
Dekel, A. and Eldar, A. and Kolatt, T. and Yahil, A. and Willick, J. A. and
Faber, S. M. and Courteau, S. and Burstein, D., 1999, ApJ, 522, 1
Eke, V., Cole, S., and Frenk, C.: 1996, MNRAS, 282, 263
Fisher, K. B. and Huchra, J. P. and Strauss, M. A. and Davis, M. and Yahil, A.
and Schlegel, D., 1995, ApJS, 100, 69
Fisher, K. B. and Lahav, O. and Hoffman, Y. and Lynden-Bel, D. and Zaroubi,
S., 1995, MNRAS, 272, 219
Haynes, M. P. and Giovanelli, R. and Salzer, J. J. and Wegner, G. and Freudling,
W. and da Costa, L. N. and Herter, T. and Vogt, N. P., 1999, ApJ, 117, 1668
Hernquist, L. and Katz, N., 1989, ApJS, 70, 419
Kogut, A., Linewater, C., Smoot, G., Bandey, C. B. A., N.W.Boggess, Cheng,
E., Amici, G. D., Fixsen, D., Hinshaw, G., Jackson, P., Janssen, M., Keegstra,
P., Loewenstein, K., Lubin, P., Mahter, J., Tenorio, L., Weiss, R., Wilkinson,
D., and Wright, E., 1993, ApJ, 419, 1
Lahav, O. and Fisher, K. B. and Hoffman, Y. and Scharf, C. A. and Zaroubi, S.,
Jackson, J., 1999, Classical Electrodynamics, New York: Wiley
Juszkiewicz, R. and Ferreira, P. G. and Feldman, H. A. and Jaffe, A. H. and
Davis, M., 2000, Science, 287, 109
Nusser, A. and Davis, M. , 1994, ApJ, 421, L1
Nusser, A. and da Costa, L. N. and Branchini, E. and Bernardi, M. and Alonso,
M. V. and Wegner, G. and Willmer, C. N. A. and Pellegrini, P. S., 2000,
MNRAS, 320, L21
Rowan-Robinson, M. and Lawrence, A. and Saunders, W. and Crawford, J. and
Ellis, R. and Frenk, C. S. and Parry, I. and Xiaoyang, X. and Allington-Smith,
J. and Efstathiou, G. and Kaiser, N., 1990, MNRAS, 247, 1
Saunders, W. and Sutherland, W. J. and Maddox, S. J. and Keeble, O. and
Oliver, S. J. and Rowan-Robinson, M. and McMahon, R. G. and Efstathiou,
G. P. and Tadros, H. and White, S. D. M. and Frenk, C. S. and Carramiñana,
A. and Hawkins, M. R. S., 2000, MNRAS, 317, 55
Sigad, Y. and Eldar, A. and Dekel, A. and Strauss, M. A. and Yahil, A., 1998,
ApJ, 495, 516
Springel, V.,1996, MSc thesis, Eberhard-Karls-Univerität Tübingen
Strauss, M. A.and Willick, J. A., 1995, PhR, 261, 271
Zaroubi, S. and Branchini, E. and Hoffman, Y. and da Costa, L. N., 2002,
MNRAS, 336, 1234
Tadros, H., Ballinger, W. E., Taylor, A. N., Heavens, A. F., Efstathiou, G.,
Saunders, W., Frenk, C. S., Keeble, O., MCMahon, R., Maddox, S. J., Oliver,
S., Rowan-Robinson, M., Sutherland, W. J., and White, S. D. M.: 1999,
MNRAS, 305, 527
Teodoro, L., 1999, PhD thesis, University of Durham
Willick, J. and Strauss, M. and Dekel, A. and Kolatt, T., 1997, ApJ, 486,629
Willick, J. and Courteau, S. and Faber, S. and Burstein, D. and Dekel, A. and
Strauss, M. A., 1997, ApJS, 109, 333
Willick, J. and Strauss, M. A., 1998, ApJ, 507, 64
Yahil, A. and Strauss, M. A. and Davis, M. and Huchra, J., 1991, ApJ, 372, 380
York, D. G. et al., 2000, ApJ, 120, 1579
Zehavi, I. and Riess, A. G. and Kirshner, R. P. and Dekel, A., 1998, ApJ, 503,
483
|
0801.2867 | 3 | 0801 | 2008-08-01T17:21:52 | The Helium and Heavy Elements Enrichment of the Galactic Disk | [
"astro-ph"
] | We present chemical evolution models for the Galactic disk. We also present a new determination of X, Y, and Z for M17 a Galactic metal-rich Hii region. We compare our models for the Galactic disk with the Galactic Hii regions abundances. The $\Delta Y/\Delta O$ ratio predicted from the galactic chemical evolution model is in very good agreement with the $\Delta Y/\Delta O$ value derived from M17 and the primordial helium abundance, Yp, taking into account the presence of temperature variations in this Hii region. From the M17 observations we obtain that $\Delta Y/\Delta Z = 1.97 \pm 0.41$, in excellent agreement with two $\Delta Y/\Delta Z$ determinations derived from K dwarf stars of the solar vicinity that amount to $2.1 \pm 0.4$ and $2.1 \pm 0.9$ respectively. We also compare our models with the solar abundances. The solar and Orion nebula O/H values are in good agreement with our chemical evolution model. | astro-ph | astro-ph |
Manuscript for Revista Mexicana de Astronom´ıa y Astrof´ısica (2007)
THE HELIUM AND HEAVY ELEMENTS
ENRICHMENT OF THE GALACTIC DISK
L. Carigi1,2 and M. Peimbert1
Draft version: November 13, 2018
RESUMEN
Presentamos modelos de evoluci´on qu´ımica para el disco de nuestra galaxia.
Tambi´en presentamos una nueva determinaci´on de X, Y , y Z para M17, una
regi´on H ii de nuestra galaxia rica en elementos pesados. Comparamos nue-
stros modelos del disco gal´actico con las abundancias de las regiones H ii. El
valor predicho por nuestro modelo para ∆Y /∆O es muy similar al valor que
obtenemos por medio de las observaciones de M17 y la abundancia primordial
de helio, Yp. A partir de M17 y Yp obtenemos que ∆Y /∆Z = 1.97 ± 0.41
resultado que concuerda con dos determinaciones de ∆Y /∆Z, obtenidas a
partir de observaciones de estrellas enanas K de la vecindad solar, que corre-
sponden a 2.1 ± 0.4 y 2.1 ± 0.9 respectivamente. Nuestros modelos ajustan
razonablemente bien el valor de O/H con el que se form´o el Sol.
ABSTRACT
We present chemical evolution models for the Galactic disk. We also present a
new determination of X, Y , and Z for M17 a Galactic metal-rich H ii region.
We compare our models for the Galactic disk with the Galactic H ii regions
abundances. The ∆Y /∆O ratio predicted from the galactic chemical evolution
model is in very good agreement with the ∆Y /∆O value derived from M17
and the primordial helium abundance, Yp, taking into account the presence
of temperature variations in this H ii region. From the M17 observations we
obtain that ∆Y /∆Z = 1.97 ± 0.41, in excellent agreement with two ∆Y /∆Z
determinations derived from K dwarf stars of the solar vicinity that amount
to 2.1 ± 0.4 and 2.1 ± 0.9 respectively. We also compare our models with
the solar abundances. The solar and Orion nebula O/H values are in good
agreement with our chemical evolution model.
Key Words: galaxies: abundances -- galaxies: evolution -- H II regions --
ISM: abundances -- ISM: individual: M17, Orion nebula -- Sun: abundances
1. INTRODUCTION
The main purpose of this work is to study the evolution of the helium
abundance with respect to the heavy elements as a function of time and posi-
tion in the Galactic disk. For this purpose we will use the Galactic chemical
evolution model by Carigi et al. (2005) that has been successful in explaining:
the observed O/H and C/H abundance gradients in the interstellar medium,
1Instituto de Astronom´ıa, UNAM, M´exico.
2Centre for Astrophysics, UCLan, UK.
1
2
CARIGI & PEIMBERT
ISM, the present gaseous distribution in the Galactic disk, the current star
formation rate, the stellar mass as a function of the Galactic radius, and most
of chemical properties of the solar vicinity.
To have useful observational constraints we need accurate X, Y , and Z
determinations or at least ∆Y /∆Z determinations to compare the models
with observations. In this paper we recompute the X, Y , and Z values for
the H ii region M17, the best Galactic H ii region for which it is possible
to compute an accurate enough Y value, for this purpose we make use of
the best observations available and the new He I atomic data needed for the
helium abundance determination. We also compare the chemical evolution
model with the ∆Y /∆Z determination derived from K dwarf stars of the
solar vicinity with metallicities similar or higher than solar by Jim´enez et al.
(2003) and Casagrande et al. (2007).
We also compare our model with the initial solar O/H value and with the
Orion O/H value. The initial solar O/H value is representative of the ISM,
4.5 Gyr ago when the Sun was formed, and the Orion nebula O/H value is
representative of the present day ISM.
The model together with the primordial helium determination is also used
to provide an equation between the Y enrichment and the O enrichment of
the ISM. This equation can be used to provide the initial Y values for those
stars for which we can derive their initial oxygen abundances. These initial Y
values provide meaningful initial abundances for a set of stellar evolutionary
models with different heavy elements content.
We adopt the usual notation X, Y , and Z to represent the hydrogen,
helium and heavy elements abundances by mass, respectively. Based on our
models we study the increase of helium, ∆Y , as a function of the increase of
C, O, F e, and Z by mass.
In Section 2 we discuss the general properties of the chemical evolution
models, we discuss inflow models for the Galaxy with two sets of stellar yields
and present the chemical abundances for the disk at 5 galactocentric distances.
For the two models discussed we present the increase of helium by mass ∆Y ,
relative to the increase of carbon, oxygen, iron and heavy elements by mass,
∆C, ∆O, ∆F e, and ∆Z. We also discuss the evolution of the Y and O
abundances for our models, and present an equation that predicts for the
Galaxy the Y enrichment as a function of the O enrichment of the ISM. In
Section 3 we present a new determination of the helium abundance for the
metal rich H ii region M17; this abundance is compared with our Galactic
chemical evolution models.
In Section 4 we compare abundances for the Orion nebula with those
of B stars of the Orion association.
In Section 5 we discuss the absolute
calibration of the O/H ratio in the local ISM, this ratio is one of the most
important observational constraints for Galactic chemical evolution models,
the absolute calibration of the O/H ratio is obtained based on the Orion nebula
and the solar abundances. In Section 6 we compare the solar initial helium
abundance, inferred from the standard solar models by Bahcall et al. (2006),
HE ENRICHMENT OF THE GALACTIC ISM
3
with our Galactic chemical evolution models considering the time since the
Sun was formed and the presence of gravitational settling and diffusion. The
conclusions are presented in Section 7.
Throughout this paper we will use the primordial helium abundance by
mass, Yp, derived by Peimbert et al. (2007) based on direct helium abun-
dance determinations of metal poor extragalactic H ii regions that amounts
to 0.2477 ± 0.0029. This result is in excellent agreement with the Yp deter-
mination by Dunkley et al. (2008) that amounts to 0.2484 ± 0.0003. This
determination is based on the Ωbh2 value derived from WMAP observations,
the assumption of standard big-bang nucleosynthesis, and the neutron life-
time, τn, of 885.7 ± 0.8 sec obtained by Arzumanov et al. (2000). Following
Mathews et al. (2005) the value of Yp derived from WMAP is revised down-
wards to 0.2468 ± 0.0003 by adopting the τn = 878.5 ± 0.8 sec derived by
Serebrov et al. (2005), and to 0.2475 ± 0.0006 by adopting for τn the new
world average that amounts to 881.9 ± 1.6 sec, average that includes the re-
sults by Arzumanov et al. (2000) and Serebrov et al. (2005).
2. CHEMICAL EVOLUTION MODELS
2.1. Model Parameters
We present chemical evolution models for the Galactic disk using the
CHEMO code (Carigi 1994) that considers the lifetime of each star until
it leaves the main sequence. The models have been built to reproduce the
present gas mass distribution and the present-day O/H values for each galac-
tocentric distance, the values were obtained from observations of H ii regions
in the galaxy (Esteban et al. 2005). Specifically, the characteristics of the
models are:
i) An inside-outside scenario with primordial infalls but without any type
of outflows. The infall rate as a function of time and galactocentric distance
r is given by IN F ALL(r, t) = A(r)e−t/τhalo + B(r)e−(t−1Gyr)/τdisk, where the
formation timescales are τhalo = 0.5 Gyr and τdisk = 6 + (r/r⊙ − 1)8 Gyr. We
assume the location of the solar vicinity is r⊙ = 8 kpc. The constants A(r)
and B(r) are chosen to match, first, the present-day mass density of the halo
and disk components in the solar vicinity, 10 and 40 M⊙pc−2, respectively, and
second, to reproduce the radial profile total mass in the Galaxy, M tot(r) =
50e−(r−8)/3.5 (Fenner et al. 2003).
ii) 13 Gyr as the age of the models, the time elapsed since the beginning
of the formation of the Galaxy.
iii) The Initial Mass Function (IMF) proposed by Kroupa, Tout & Gilmore
(1993), in the mass interval given by 0.01 < m/M⊙ < Mup, with Mup = 80
and 60 M⊙. This IMF is a three power-law approximation, given by IMF
∝ m−α with α = −1.3 for 0.01 - 0.5 M⊙, α = −2.2 for 0.5 - 1.0 M⊙, and
α = −2.7 for 0.5 - Mup.
Note that, our models were computed assuming an IMF with Mlow = 0.01
(1993) truncated their IMF at 0.08 M⊙because they
M⊙. Kroupa et al.
4
CARIGI & PEIMBERT
Fig. 1. Newly formed mass of a given element by massive stars, in M⊙, ejected to the
ISM. The initial heavy elements of the stars amount to Z = 0.02. Continuous lines:
high wind yields by Maeder (1992), dashed lines: low wind yields by Hirschi et al.
(2005).
considered only stars, but in our work we are assuming a non negligible amount
of substellar objects (0.01 < m/M⊙ < 0.08). In models with Mup = 60 M⊙,
the mass of objects with m < 0.08M⊙ is 12 % of the total Mstars, that
includes stars and remnants; this percentage is practically independent of
Mup. Even at present, the fraction of mass in substellar objects is unknown
and we consider that our predicted percentage might be realistic.
Due to the uncertainties in the current Mgas(r), Mstars(r), and SFR(r)
values we cannot discriminate between chemical evolution models assuming
Mlow = 0.01 M⊙and Mlow = 0.08 M⊙. The first ones predict smaller fractions
of massive stars and LIMS per single stellar generation. Therefore Mlow =
0.01 M⊙models with identical galaxy formation scenario, galactic age, and
stellar yields to those of Mlow = 0.08 M⊙models, require a higher SFR to
match the present-day O/H(r) values. The Mlow = 0.01 M⊙model with a
HE ENRICHMENT OF THE GALACTIC ISM
5
more efficient SFR predicts a lower Mgas and similar chemical abundances.
Such model with higher SFR and lower Mgas is also able to reproduce the
observational constraints (Carigi 1996).
iv) A star formation rate that depends on time and galactocentric distance,
that varies from almost constant and low (at large r values) to bursting and
high (at short r values), this SFR has been represented by the following re-
lation SF R(r, t) = νM 1.4
gas(r, t) (Mgas + Mstars)0.4(r, t), in order to reproduce
the current O/H gradient and the gas mass distribution of the Galactic disk
(Carigi 1996), where ν is a constant in time and space that is chosen in order
to reproduce the present-day radial distribution of the gas surface mass den-
sity. A ν value of 0.016 is required when the high-wind yields and Mup = 80
M⊙are adopted, the best model of Carigi et al. (2005), while ν values of 0.015
and 0.010 are required when the low-wind yields with Mup = 60 and Mup = 80
M⊙are adopted, respectively.
v) Two sets of stellar yields. Since the main difference between these
sets is the assumed mass-loss rate due to stellar winds by massive stars with
Z = 0.02, we will call them high-wind yields (HWY) and low-wind yields
(LWY), see Figure 1.
The HWY set is the one considered in the best model (model 1) of Carigi et al.
(2005). The HWY set includes: A) For massive stars (MS), those with
8 < m/M⊙ < 80, the following yields by: a) Chieffi & Limongi (2002) for
Z = 0.00; b) Meynet & Maeder (2002) for Z = 10−5 and Z = 0.004; c)
Maeder (1992) for Z = 0.02 (high mass-loss rate yields presented in his Table
6); d) Woosley & Weaver (1995) only for the Fe yields (Models B, for 12 to
30 M⊙; Models C, for 35 to 40 M⊙; while for m > 40 M⊙, we extrapolated
the m = 40 M⊙Fe yields). B) For low and intermediate mass stars (LIMS),
those with 0.8 ≤ m/M⊙ ≤ 8, we have used the yields by Marigo et al. (1996,
1998) and Portinari et al. (1998) from Z = 0.004 to Z = 0.02. C) For Type Ia
SNe we have used the yields by Thielemann et al. (1993). We have assumed
also that 5 % of the stars with initial masses between 3 and 16 M⊙ are binary
systems which explode as SNIa.
In the LWY set we have updated the yields of massive stars only for Z ∼ 0
and Z = 0.02 assuming the yields by Hirschi (2007) and Hirschi et al. (2005)
respectively. The rest of the stellar yields are those included in the high wind
set.
The main differences between the LWY set and the HWY set are due to the
contribution of massive stars at Z = 0.02. Therefore in Figure 1 we compare
the He, C, and O yields for Z = 0.02. The main difference between the HWY
and the LWY models is due to the stellar yields assumed for massive stars at
high Z. The HWY assume a relatively high mass-loss rate for massive stars
with Z = 0.02 (yields by Maeder 1992), while the LWY assume a relatively
low mass-loss rate for massive stars with Z = 0.02 (yields by Hirschi et al.
2005). These difference between a high and a low mass-loss rate produces
opposite differences in the C and O yields (see Figure 1), the reasons are the
following: a) a high mass-loss rate produces a high loss of C and consequently
6
CARIGI & PEIMBERT
Fig. 2.
Evolution of some common properties for all models of the Galactic
disk: gas mass surface density (M⊙pc−2), star formation rate and infall rates
(M⊙pc−2Gyr−1), at four galactocentric distances, 16, 12, 8, and 4 kpc (continu-
ous, long-dashed, short-dashed, and dotted lines, respectively).
a high C yield, and b) since C is needed to produce O, the high loss of C
reduces the O yield.
Since the solar vicinity and the Galactic disk contain stars and H ii regions
of a broad range of metallicities, our galaxy is a proper laboratory to study
the ∆Y /∆Xi behavior at high Z values and to try to observationally test the
predictions of the HWY models and the LWY models.
2.2. Results
The models presented in this paper reproduce the present stellar and gas
mass distributions in the Galactic disk, the current star formation rate as
a function of the Galactic radius, the O/H gradient evolution inferred from
PNe, the SN rates, the distribution of G-dwarf stars as a function of [Fe/H],
HE ENRICHMENT OF THE GALACTIC ISM
7
Fig. 3. Chemical evolution models for the Galactic disk and the solar vicinity (r = 8
kpc). The left panel shows the C/O evolution in the ISM of the solar vicinity with
O/H. The right panels show the present-day ISM abundance ratios as a function of
galactocentric distance. Continuous lines: high wind yields with Mup = 80 M⊙by
Carigi et al. (2005) , dashed and dotted lines: low wind yields with Mup = 60 and 80
M⊙, respectively (this paper). Filled circles: H ii regions, gas plus dust values; the
gaseous values from Esteban et al. (2005) have been corrected for the dust fraction.
Filled squares: Dwarf stars from Akerman et al. (2004). Open circle: Solar values
from Asplund et al. (2005).
the infall rate, and the evolution of [Xi/Fe] vs [Fe/H]. See Allen et al. (1998),
Carigi (1994), Carigi (1996), Carigi (2000).
In Fig. 2 we present some of those model properties, like gas mass surface
density, star formation rate, and infall rate for four galactocentric distances:
16, 12, 8, and 4 kpc. Infall rate follows the inside-outside scenario, the inner
parts of the Galaxy are formed faster than the outer parts. Since gas mass
comes from infall, mainly, M gas reflects the inside-outside scenario and the
SFR shows similar behaviour as the gas mass, due to the SFR is proportional
to the gas mass.
8
CARIGI & PEIMBERT
TABLE 1
PRESENT DAY VALUES FROM THE GALACTIC DISK MODELS
Galactocentric distance O(tf inal)(10−3) ∆Y /∆C ∆Y /∆O ∆Y /∆F e ∆Y /∆Z
4
8
12
16
4
8
12
16
High wind yields and Mup = 80 M⊙
5.91
8.89
6.38
6.68
4.26
7.01
8.89
4.01
3.29
1.95
3.23
Low wind yields and Mup = 60 M⊙
4.06
10.05
12.81
7.13
6.78
3.28
1.34
7.73
9.44
10.18
3.79
3.99
4.39
14.11
18.36
21.60
25.28
14.13
18.36
21.52
26.11
1.85
1.67
1.62
1.65
1.67
1.67
1.78
1.90
In Figure 3 we show O and C gradients in the Galactic disk and the
evolution of the C/O-O/H relation in the solar vicinity predicted by three
models that combine different yields and IMF Mup values. Models that assume
LWY fail to reproduce the C/O gradient and the C/O values for halo stars or
disk stars. The LWY model with Mup = 60 M⊙reproduces poorly the C/O
gradient and the C/O values in disk stars, and predicts C/O values for halo
stars higher than observed. The LWY model with Mup = 80 M⊙does not
reproduce at all the C/O gradient, matches partially the C/O values of disk
stars, and explains the observed C/O values for halo stars. The HWY model
with Mup = 80 M⊙reproduces successfully the C/O Galactic gradient and the
C/O values in the solar vicinity.
Based on Figure 3 we conclude that the LWY model with Mup = 60
M⊙reproduces the main behavior of C/O vs O/H in the solar vicinity but
cannot reproduce the C/O Galactic gradient. On the other hand, the HWY
model reproduces very well the C/O vs O/H relation in the solar vicinity and
the C/O Galactic gradient. Since the LWY model with Mup = 80 M⊙produces
the poorest fit to the C/O and C/H observed values it will not be considered
further.
Gibson et al (2006) considering the O yields by Arnett (1991), that are
lower than those by Maeder (1992) for MS with m < 30 M⊙, reproduce the
[O/Mg] values present in the Galactic Bulge. With the same yields Gibson
(1997) explains the [O/Fe] values in the intracluster medium and predicts a
small increase in the C/O evolution at late times for a massive elliptical galaxy.
By adopting the Arnett (1991) yields in our model we may obtain flatter C/O
gradients and might not be able to reach the C/O values observed in the H ii
regions and dwarf stars of the solar vicinity. The yields by Arnett (1991) do
not consider stellar winds. Chiappini et al. (2005) using LWY studied the C
and O evolution in the solar vicinity and the Galactic disk, they reproduce
HE ENRICHMENT OF THE GALACTIC ISM
9
also the C/O vs O/H behavior in the solar vicinity, but they predict flatter
C/O gradients than the observed ones and they cannot match the high C/O
values shown by the metal rich star in the solar neighborhood.
Recently, McWilliam et al. (2007) suggested that the strong metallicity-
dependent yields for massive stars by Maeder (1992), can explain the O/Mg
vs O-Mg/H and the O/Fe vs Fe/H relations in the Galactic bulge and in
the solar vicinity, in agreement with our results that favor the HWY model
over the LWY model. The HWY model includes low O yields at high Z
and therefore explains: a) the small O increase in the solar vicinity from the
time the Sun was formed until the present, and b) the flattening of the O
gradient in the direction of the Galactic center. Massive stars with high Z
values have strong winds, lose a considerable amount of C and produce high
C yields. With this C lost, the stars keep a small amount of C needed to
produce O and consequently their O yields are low. Therefore the C yields
for massive stars are important at high metallicities, while their O yields
are more important at low metallicities, as has been shown previously by
Akerman et al. (2004) and Carigi et al. (2005). By adding our previous results
(Akerman et al. 2004; Carigi et al. 2005) to those of McWilliam et al. (2007)
and of this paper we insist that the stellar winds with a high mass loss rate
are essential to reproduce the high C/O values observed in the disk stars of
the solar vicinity.
In the upper panel of Figure 4 we present the evolution of the model that
assumes HWY and Mup = 80 M⊙at different galactocentric radii (4, 8, 12,
and 16 kpc) that correspond to different final metallicities (Z = 0.028, 0.016,
0.009, and 0.004, respectively). The ∆Y /∆O increase at O > 4 × 10−3 present
in Figure 4 is due to the lower O yields for massive stars with Z = 0.02. In
the upper half of Table 1 we show the present-day O values and the ∆Y /∆Xi
values for each galactocentric radius. We note that ∆Y /∆Z increases slightly
with Z for large r values or low O values, while ∆Y /∆Z increases significantly
with Z for short r values or high O values.
In the lower panel of Figure 4 we show the results for the model with
LWY and Mup = 60 M⊙. This model does not predict an increasing ∆Y /∆O
value with increasing O for high O values. In the lower half of Table 1 we
show the present-day O values and the ∆Y /∆Xi values for each galactocentric
radius. With the LWY model we find higher final O values for lower r values
because the O yields are higher than than those of the HWY model at high Z.
Even if the HWY and LWY yields are identical for low Z, we get lower final O
values for higher r values for the LWY model because it does not include stars
with m > 60 M⊙. The increase or decrease of O is reflected on the ∆Y /∆O
and ∆Y /∆Z values because the final Y values are nearly independent of the
models.
The helium to oxygen mass ratio, ∆Y /∆O, is an important constraint in
the study of the chemical evolution of galaxies. We have studied the variation
of ∆Y /∆O as O increases in the HWY and LWY evolution models (see Figure
4). For the HWY model, the model that fits the C/O gradient, we have found
10
CARIGI & PEIMBERT
Fig. 4. Evolution of Helium vs Oxygen for the Galactic disk at four galactocentric
distances, 16, 12, 8, and 4 kpc (continuous, long-dashed, short-dashed, and dotted
lines, respectively). Upper panel: The chemical evolution model assumes Mup =
80 M⊙and high wind yields. Lower panel: The chemical evolution model assumes
Mup = 60 M⊙and low wind yields. See Table 1. Low metallicity H ii regions from
Peimbert et al. (2007) (filled triangles) and adopting Yp = 0.2477.
the following relations between Y and O:
Y = Yp + ∆Y = Yp + (3.3 ± 0.7)O,
(1)
for O < 4.3 × 10−3, and
Y = Yp + (3.3 ± 0.7)O + (0.016 ± 0.003)(O/4.3 × 10−3 − 1)2,
(2)
for 4.3 × 10−3 < O < 9 × 10−3.
For the LWY model with Mup = 60 M⊙we have found the following relation
between Y and O:
Y = Yp + (4.0 ± 0.7)O,
(3)
HE ENRICHMENT OF THE GALACTIC ISM
11
for 0 < O < 11 × 10−3.
Jim´enez et al. (2003) from a set of isochrones and observations of nearby
K dwarf stars found that ∆Y /∆Z = 2.1 ± 0.4. Casagrande et al. (2007)
found also that ∆Y /∆Z = 2.1 ± 0.9 from the newly computed set of Padova
isochrones and observations of nearby K dwarf stars. These observational
results are in very good agreement with the models presented in Table 1.
Assuming yields with low mass-loss rate due to stellar winds (similar to
that considered by Hirschi 2007), and yields without mass loss due to stellar
winds by Woosley & Weaver (1995), Chiappini et al. (2003) find for the solar
vicinity ∆Y /∆Z ∼ 2.4 and 1.5, respectively.
Since stellar winds change significantly the C and O yields, but not the
Y yields, we are interested to quantify the evolution of the Y contribution
due to massive stars and due to low and intermediate mass stars at different
metallicities. For that reason we show in Figure 5 the accumulative percentage
of Y for four galactocentric distances due to MS and LIMS obtained from the
HWY model, our successful model for Galactic disk.
The fraction of helium in the ISM due to MS and and to LIMS depends
strongly on time, but not on galactocentric radius or Z. At present about half
of the ∆Y in the ISM has been produced by MS and half by LIMS.
The strong dependency on time of the ∆Y contribution is due to the
lifetime of the stars. The ∆Y contribution of LIMS decreases less than 5
% from 4 to 16 kpc due mainly to the star formation history. In the inside-
outside scenario the SFR at 4 kpc is more intense and it is also higher at earlier
times than at larger distances (see the middle panel in Figure 2), therefore
the big number of LIMS formed in the first Gyrs at 4 kpc enrich the gas at
later times. This fact produces the O dilution that can be seen in the right
hand side panels of Figure 5, just after the model at 4 kpc reaches the value
of O = 3 × 10−3 for the first time.
3. THE HELIUM AND OXYGEN ABUNDANCES OF M17, A HIGH
METALLICITY GALACTIC H II REGION
We will compare the predictions of the HWY and the LWY models with
observations of Y and O in the Galactic disk. At present the best H ii region
in the Galaxy to derive the Y and O abundances is M17. The reason is that
the correction for the presence of neutral helium in the abundance determina-
tion is the smallest for the well observed Galactic H ii regions. This is due to
the high ionization degree of M17 (Peimbert et al. 1992; Esteban et al. 1999;
Garc´ıa-Rojas et al. 2007). Due to the large amount of neutral helium present
in the other well observed Galactic H ii regions, the error in the Y determi-
nation is at least two times larger than the error for M 17, therefore the Y
determinations for the other Galactic H ii regions will not be considered in
this paper.
To determine very accurate He/H values of a given H ii region we need to
consider its ionization structure. For objects of low degree of ionization it is
necessary to consider the presence of He0 inside the H+ zone, while for objects
12
CARIGI & PEIMBERT
Fig. 5. Cumulative percentage of He as a function of time and oxygen in the ISM,
produced and ejected by massive stars (MS) and low and intermediate mass stars
(LIMS) at four galactocentric distances on the Galactic disk (lines as Figure 2).
The model assumes Mup = 80 M⊙and high mass loss due to stellar winds, the HWY
model.
of high degree of ionization it is necessary to consider the possible presence of a
He++ zone inside the H+ zone. Peimbert et al. (1992), hereafter PTR, found
for M17 an upper limit of N (He++)/N (H+) of 8 ×10−5 a negligible amount;
alternatively they found differences with position of the N (He+)/N H(+) ratio
correlated with the sulphur ionization structure, result that implies that M17
is ionization bounded and the presence of a small but non negligible amount
of He0 inside the H+ zone. Therefore for this object the helium abundance is
given by
N (He0) + N (He+)
N (H+)
N (He)
N (H)
=
.
(4)
To minimize the effect of the correction for neutral helium we took into
account only regions M17-1, M17-2, and M17-3, from now on M17-123, that
show the highest degree of ionization of the observed regions by PTR, Esteban et al.
HE ENRICHMENT OF THE GALACTIC ISM
13
(1999), and Garc´ıa-Rojas et al. (2007), as well as the highest accuracy in the
line intensity determinations by PTR. Following PTR we will assume that He
is neutral in the regions where S is once ionized, that is
N (He0)
N (He)
=
N (S+)
N (S)
,
therefore
N (He)
N (H)
= ICF(He) ×
N (He+)
N (H+)
= (cid:20)1 +
N (S+)
N (S) − N (S+)(cid:21) ×
N (He+)
N (H+)
.
(5)
(6)
To estimate the ICF(He) value we recomputed the N (S+)/N (S) ratios derived
by PTR taking into account that the [S ii] λλ 4069 + 4076 lines are blended
with the O ii λλ 4069 and 4076 lines, the correction diminishes the [S ii] elec-
tron temperatures from about 12 000 K to about 7700 K (Garc´ıa-Rojas et al.
2007), the lower temperatures increase the N (S+)/N (S) ratios, and the av-
erage ICF (He) for M17-123 amounts to 1.035.
To obtain the N (He+)/N H(+) value for M17-123 we decided to recom-
pute the determinations by PTR based on their line intensities and the new
helium recombination coefficients by Porter et al. (2005), with the interpo-
lation formula provided by Porter et al. (2007).
In addition we used the
hydrogen recombination coefficients by Storey & Hummer (1995), and the
collisional contribution to the He i lines by Sawey & Berrington (1993) and
Kingdon & Ferland (1995). The optical depth effects in the triplet lines were
estimated from the computations by Benjamin et al. (2002). At the tem-
peratures present in M17 the collisional excitation of the hydrogen lines is
negligible and was not taken into account.
To determine the N (He+)/N (H+) value we took into account the following
He i lines λλ 3889, 4026, 4471, 4922, 5876, 6678, and 7065. We corrected the
4922 line intensity by considering that it was blended with the [Fe iii] 4924
line and that the contribution of the Fe line amounted to 5% of the total line
intensity (Esteban et al. 1999; Garc´ıa-Rojas et al. 2007). The M17-123 line
intensities adopted are presented in Table 2.
We did not correct the H and He line intensities for underlying absorp-
tion, the reasons are the following: a) the average observed equivalent width in
emission of H(β), EWem(Hβ), amounts to 668 A , b) the predicted EWem(Hβ)
for Te = 7000 K amounts to about 2000 A (Aller 1984), therefore about 1/3 of
the continuum is due to the nebular contribution and 2/3 to the dust scattered
light from OB stars, consequently the underlying stellar absorption only affects
two thirds of the observed continuum, d) considering the nebular contribu-
tion to the observations and based on the models by Gonz´alez Delgado et al.
(1999, 2005) for a model with an age of 2 Myrs as well as the observations by
Leone & Lanzafame (1998) for λ 7065, (since λ 7065 was not included in the
models by Gonz´alez Delgado et al. 1999, 2005), we estimated that the EWab
of the λλ 3889, 4026, 4471, 4922, 5876, 6678, and 7065 lines amount to 0.4,
0.4, 0.4, 0.1, 0.1, 0.2 A respectively, an almost negligible amount considering
14
CARIGI & PEIMBERT
TABLE 2
HE I LINE INTENSITIES RELATIVE TO Hβ FOR M17-123
He i Line
I
3889
4026
4471
4922
5876
6678
7065
7281
0.1738 ± 0.0086
0.0240 ± 0.0012
0.0508 ± 0.0013
0.0127 ± 0.0010
0.1549 ± 0.0038
0.0395 ± 0.0010
0.0499 ± 0.0025
0.0064 ± 0.0007
TABLE 3
PHYSICAL CONDITIONS AND CHEMICAL ABUNDANCES IN M17
Parameter
t2 = 0.000
t2 = 0.036 ± 0.013
T [O ii + O iii]
8300 ± 200
n
τ3889
N (He+)/N (H+)
ICF (He)
691 ± 246
9.5 ± 0.8
8300 ± 200
744 ± 247
11.0 ± 0.9
0.1014 ± 0.0014
0.0982 ± 0.0019
1.035 ± 0.010
1.035 ± 0.010
N (He)/N (H)
0.1049 ± 0.0017
0.1016 ± 0.0022
Y
∆Y
O
∆Y /∆O
Z
∆Y /∆Z
0.2926 ± 0.0034
0.0403 ± 0.0044a
0.00446 ± 0.00045
0.2837 ± 0.0044
0.0360 ± 0.0053b
0.00811 ± 0.00081
9.04 ± 1.35a
4.44 ± 0.79 b
0.0101 ± 0.0015
0.0183 ± 0.0027
4.00 ± 0.75a
1.97 ± 0.41 b
aWhere we adopted Yp = 0.2523 ± 0.0027
for t2 = 0.000 from Peimbert et al. (2007).
bWhere we adopted Yp = 0.2477 ± 0.0029
for t2 6= 0.000 from Peimbert et al. (2007).
the large EWem observed values (see Table 4 in Peimbert et al. 1992), e) the
weighted increase in the helium line intensities amounts to about 0.7%, again
an almost negligible amount, f) the Balmer lines also show underlying absorp-
tion and the average correction to the line intensities amounts to about 0.5%,
again a negligible amount that cancels to a first approximation the underlying
correction effect on the He/H line ratios.
HE ENRICHMENT OF THE GALACTIC ISM
15
TABLE 4
PRESENT DAY VALUES FROM THE GALACTIC DISK MODELS FOR
R = 6.75 KPC
Model
O(tf inal)(10−3) ∆Y /∆C ∆Y /∆O ∆Y /∆F e ∆Y /∆Z
HWY Mup = 80 M⊙
LWY Mup = 60 M⊙
7.16
8.28
6.54
7.26
4.47
3.73
17.04
17.04
1.70
1.62
To determine the helium physical conditions of the nebula simultaneously
with the N (He+)/N (H+) value we used the maximum likelihood implemen-
tation presented by Peimbert et al. (2002). This implementation requires as
inputs: a) the oxygen temperatures, T [O iii] and T [O ii], and the oxygen
ionization degree that provide us with the following restriction
T [O II + O III] =
N (O+)T [O II] + N (O++)T [O III]
N (O+) + N (O++)
,
(7)
and b) a large set of helium to hydrogen line intensity ratios. In addition an
estimate of the electron density, n, in the region where the He lines originate
is not required but it is useful. This implementation can determine the con-
ditions of the H ii regions either with the restriction of uniform temperature,
or relaxing this restriction.
From PTR we adopted T [O iii] = 8200 ± 200 K. From the I(3727/7325)
ratios for M17-123 by PTR and for M17-3 by Esteban et al. (1999), after
correcting the λ 7325 A lines for the recombination contribution (Liu et al.
2000), we obtained 8100 ± 1300 K and 9900 ± 1300 K respectively, therefore
we adopted for T [O ii] a value of 9000 ± 1000 K. From the T [O iii] and T [O ii]
values and the observations by PTR we find that N (O+)/N (H+) = 0.12 and
N (O++)/N (H+) = 0.88. Finally from the previous results and equation (7)
we obtained that T [O ii + O iii] = 8300 ± 200 K.
To estimate the electron density we used three determinations the n[S ii]
for M17-123 by PTR that amounts to 720 ± 250 cm−3, the n[O ii] for M17-3
by Esteban et al. (1999) that amounts to 790 ± 250 cm−3, and the n[Cl iii] for
M17-123 that we estimated from the observed I(5518)/I(5538) ratios by PTR
and the atomic physics parameters by Keenan et al. (2000) that amounts to
650 ± 450 cm−3. From the average of these three determinations we adopted
a value of n = 740 ± 250 cm−3 for M17-123.
By using as inputs for the maximum likelihood method T [O ii + O iii]
= 8300 ± 200 K, n = 740 ± 250 cm−3, and the helium line intensities pre-
sented in Table 2 we obtain for M17-123 the n, τ3889, and N (He+)/N (H+)
values presented in Table 3. The results for M17-123 are presented in Ta-
ble 3 for t2 = 0.000 (constant temperature over the observed volume) and
for t2 6= 0.000 (the temperature variations method). Without the restriction
of uniform temperature the maximum likelihood of the temperature fluctu-
ation parameter amounts to t2 = 0.036 ± 0.013. This t2 value is in good
16
CARIGI & PEIMBERT
Fig. 6. Predicted evolution of He vs Oxygen for r = 6.75 kpc. Continuous line:
model assumes Mup = 80 M⊙and high wind yield. Dotted line: model assumes
Mup = 60 M⊙and low wind yield. M17 H ii region at r = 6.75 kpc (filled circle,
t2 6= 0.000) (open circle, t2 = 0.00)
agreement with those for M17 derived by PTR, Esteban et al. (1999), and
Garc´ıa-Rojas et al. (2007) that are in the 0.033 to 0.045 range; these values
were determined with two different methods: a) combining the temperature
derived from the ratio of the Balmer continuum to the Balmer line intensi-
ties with the temperature derived from I(4363)/I(5007) [O iii] ratio, and b)
combining the O ii recombination line intensities with the λ 5007 [O iii] line
intensities.
From the mean values of N (He+)/N (H+) given in Table 3 and the ICF(He)
given by equation (6) we obtain N (He)/N (H) ratios for M17-123 of 0.1049
and 0.1016 for t2 = 0.000 and t2 = 0.036 respectively. These values are similar
to but more precise than those derived by PTR for M17-123, that amount to
0.106 and 0.100 for t2 = 0.000 and t2 = 0.040 respectively.
We obtained the ∆Y /∆O and the ∆Y /∆Z values presented in Table 3
HE ENRICHMENT OF THE GALACTIC ISM
17
based on the Yp determinations by Peimbert et al. (2007). The O abundance
presented in Table 3 includes both the gaseous and the dust contribution
and corresponds to the average of the values derived by PTR, Esteban et al.
(1999), and Garc´ıa-Rojas et al. (2007), these three values are in excellent
agreement.
M17 is located at a galactocentric distance of 6.75 kpc, under the assump-
tion that the Sun is located at a galactocentric distance of 8 kpc (Dias et al.
2002). In Figure 6 we have plotted the ∆Y /∆O value for t2 = 0.036, from
this figure it can be noted that this value is in good agreement, at about the
one σ level, with the Galactic chemical evolution model based on the HWY
set, and that the O value corresponds to the prediction by the models for a
galactocentric distance of 6.75 kpc (see also Table 4); alternatively the M17
∆Y /∆O value for t2 = 0.000 is considerably higher, by more than 3σ, than
the value predicted by the HWY model. Similarly in Figure 6 we compare
the M17 results with the LWY model, again the values for t2 = 0.036 are
in good agreement with the Galactic chemical evolution model while the val-
ues for t2 = 0.000 are not. Furthermore from Table 4 it is also found that
the t2 = 0.036 value for ∆Y /∆Z is within one σ of the models predictions,
while the t2 = 0.000 value for ∆Y /∆Z is about 3σ away form the models
predictions.
With the present accuracy of the ∆Y /∆O and ∆Y /∆Z determinations in
Galactic H ii regions it is not possible to distinguish between the HWY and
the LWY models.
4. COMPARISON OF STELLAR AND NEBULAR ABUNDANCES IN
ORION
To test the nebular abundances derived with different t2 values for the
Orion nebula we decided to compare them with those derived for B star abun-
dances of the Orion association. Cunha et al. (2006) obtained 12 + log O/H
= 8.70±0.09 from 11 B stars, while Lanz et al. (2008) obtained 12 + log Ar/H
= 6.66 ± 0.06 from 10 B stars. These results are in excellent agreement with
the nebular abundances derived by Esteban et al. (2004) for t2 6= 0.00 that
amount to 8.73 ± 0.03 and 6.62 ± 0.05 for O and Ar respectively. Alternatively
the values derived for t2 = 0.00 amount to 8.59 ± 0.03 and 6.50 ± 0.05 for
O and Ar respectively, values that are about 1σ and 3σ smaller than those
derived from B stars.
5. ABSOLUTE CALIBRATION OF THE O ABUNDANCE IN THE
SOLAR VICINITY
The predicted abundances by chemical evolution models are often com-
pared with stars and with H ii regions abundances. The most popular com-
parisons are made with the solar and the Orion nebula abundances. The
comparisons among the models, the Sun, and the Orion nebula are based on
the absolute abundances, therefore it is necessary to estimate not only the
18
CARIGI & PEIMBERT
statistical errors but also the systematic ones in the observational determina-
tions.
We will start by considering the solar O/H abundance. What we want
from the Sun is the O/H value when it was formed, the so called initial value,
and to keep in mind that it is representative of the ISM 4.5 Gyr ago when
the Sun was formed. We have also to consider that the photospheric and the
interior solar abundances might be different due to diffusion and gravitational
settling.
In Table 5 we present the most popular O/H photospheric determina-
tions of the last fifty years, the quoted values and the errors are the orig-
inal ones published by the authors. By looking at the differences among
the different determinations it is clear that for many determinations prob-
ably the errors represent only the statistical errors and that systematic er-
rors have not been taken into account, in short that the total errors have
been underestimated. The determinations, by Asplund et al. (2005) and by
Allende Prieto (2007), are qualitatively different to the previous five because
they are based on 3D models while the other five are based on 1D models.
The last two determinations included in Table 5, those by Caffau et al. (2008)
and Centeno & Socas-Navarro (2008), indicate that the possibility of a further
revision of the solar abundance is still open.
The abundances inferred from interior models of the Sun, that are based on
stereosismological data, are in disagreement with the 3D photospheric models
and predict heavy element abundances about 0.2 dex higher than the 3D
photospheric ones (e. g. Basu & Antia 2007, and references therein).
To compare with our models we will use as the low O/H value the 3D
photospheric determination by Asplund et al. (2005) and as the high O/H
value the 1D photospheric determination by Grevesse & Sauval (1998), that
is in good agreement with the helioseismic determination. The next step is
to have reliable stellar interior models to determine the initial O/H value,
for this purpose we will use the models by Bahcall et al. (2006) presented in
Table 6.
The models by Bahcall et al. (2006) indicate that the photospheric values
of Z and Y do not represent the initial values due to diffusion and gravitational
settling. By assuming that the O/Z ratio is not affected by these processes
the initial O/H values correspond to 12 + log O/H = 8.70 and to 8.89 for the
AGS05 and the GS98 surface abundances respectively. To obtain the present
day ISM value we have to consider that the Sun was formed 4.5 Gyr ago and
according to the Galactic chemical evolution model by Carigi et al. (2005) the
O/H ratio in the ISM has increased by 0.13 dex since the Sun was formed.
Therefore our determinations of the present day ISM 12 + O/H values based
on the AGS05 and the GS98 abundances amount to 8.83 and 9.02 respectively.
In Table 5 we also present the most popular O/H determinations for the
Orion nebula including the errors presented in the original papers. The pre-
dictions from the solar abundances and the chemical evolution models have to
be compared with the total abundances in the nebula that have to include gas
HE ENRICHMENT OF THE GALACTIC ISM
19
TABLE 5
12 + LOG(O/H)
Solar photosphere a
Year
t2 6= 0.00
t2 = 0.00
Year
Orion nebula b
8.96
8.77 ± 0.05
8.84 ± 0.07
8.93 ± 0.035
8.83 ± 0.06
8.736± 0.078
8.66 ± 0.05
8.65 ± 0.03
8.86 ± 0.07
8.76 ± 0.07
1960(1)
1968(2)
1976(3)
1989(4)
1998(5)
2001(6)
2005(7)
2007(8)
2008(9)
2008(10)
8.79 ± 0.12
...
8.75 ± 0.10
8.52 ± 0.10
...
8.49 ± 0.08
8.72 ± 0.07
8.55 ± 0.07
...
...
8.51 ± 0.08
8.49 ± 0.06
8.73 ± 0.03
8.59 ± 0.03
1969(11)
1977(12)
1992(13)
1998(14)
2000(15)
2003(16)
2004(17)
a
1 Goldberg et al.
(1960).
(1998).
(2001).
8 Allende Prieto (2007).
2
(1968).
3 Ross & Aller
4 Anders & Grevesse (1989).
6
7 Asplund et al.
9
10
Lambert
(1976).
5 Grevesse & Sauval
Holweger
(2005).
Centeno & Socas-Navarro (2008).
Caffau et al. (2008).
b 11 Peimbert & Costero (1969).
12
(1977).
Peimbert & Torres-Peimbert
14
13 Osterbrock et al.
Esteban et al.
in-
(1998),
cludes the fraction of O tied up in dust
15
grains that amounts to 0.08 dex.
Deharveng et al. (2000).
16 Pilyugin
(2003).
17 Esteban et al. (2004), this
value includes the fraction of O tied up in
dust grains that amounts to 0.08 dex.
(1992).
this value
and dust. With the exception of the determinations by Esteban et al. (1998,
2004) that take into account the dust fraction all the other determinations only
include the gaseous content. The other difference is that there are two possible
sets of values: a) those that assume constant temperature over the observed
value given by the 4363/5007 ratio of [O iii], the t2 = 0.00 case, where t2 is
the mean square temperature variation (Peimbert 1967), or b) those based on
the O ii recombination lines, that are in agreement with those derived from
the 5007/Hβ ratio taking into account the presence of temperature variations
over the observed volume, and consequently that t2 6= 0.00.
20
CARIGI & PEIMBERT
TABLE 6
STANDARD SOLAR MODELS A
Values
initial X
initial Y
initial Z
initial O
initial O/H
initial ∆Y /∆Z
initial ∆Y /∆O
surface X
surface Y b
surface Z
surface O/H b
GS98
0.70866
0.27250
0.01884
0.00879
8.89
1.32
2.82
0.7410
AGS05
0.72594
0.26001
0.01405
0.00582
8.70
0.88
2.11
0.7586
0.2420 ± 0.0072
0.2285 ± 0.0067
0.0170
0.0122
8.83 ± 0.17
8.66 ± 0.17
heavy
element
The GS98
aStandard solar models by Bahcall et al.
(2006).
and AGS05
correspond to models with
columns
de-
the
rived from photospheric
observations
(1998) and by
by Grevesse & Sauval
Asplund et al. (2005) respectively.
bThe errors are the conservative ones
adopted by Bahcall et al. (2006).
abundances
Esteban et al. (2004) obtain for the Orion nebula that 12 + O/H = 8.73
for t2 6= 0.00, value that includes the dust correction. By taking into account
the presence of the O/H Galactic abundance gradient that amounts to −0.044
dex kpc−1 (Esteban et al. 2005) we obtain a value of 12 + O/H = 8.75 for
the local ISM. This value is smaller than the values estimated for the local
ISM based on: the solar photospheric abundances by AGS05 and the GS98,
the standard solar models by Bahcall et al. (2006), and the chemical evolution
of the Galaxy that amount to 8.83 and 9.02 respectively. Similarly from the
results by Esteban et al. (2004) for the Orion nebula for t2 = 0.00 and the
observed Galactic gradient we obtain that 12 + O/H = 8.51 for the local
ISM. From the previous discussion it follows that the best agreement for the
derived O/H ISM value is given by the t2 6= 0.00 result from Orion and the
AGS0 result from the Sun. From these two determinations we recommend for
the present day local ISM the value 12 + log O/H = 8.79 ± 0.08.
From the previous discussion it follows that the best agreement between
the solar and the Orion nebula abundances is obtained for the high nebular
abundances, that are derived from the t2 6= 0.00 values and that include the
fraction of atoms tied up in dust grains, and the AGS05 solar value.
HE ENRICHMENT OF THE GALACTIC ISM
21
6. THE SOLAR HELIUM ABUNDANCE
We want to compare also our Galactic chemical evolution model with the
helium abundance when the Sun was formed, the initial Y value. Basu & Antia
(2004) based on seismic data have derived the Y value in the solar convective
envelope and amounts to 0.2485±0.0034. To derive the initial value we need a
model of the solar interior that takes into account helium and heavy elements
diffusion and that agrees with the helium abundance of the envelope.
Again we have at our disposal the solar interior models by Bahcall et al.
(2006) presented in Table 6. The GS98 model agrees with the Y value in the
envelope derived by Basu & Antia (2004), but not with the O/H value in
the envelope derived by Asplund et al. (2005). On the other hand the AGS05
model agrees with the O/H value in the envelope derived by Asplund et al.
(2005) but not with the Y value derived by Basu & Antia (2004). Based
on the discussion of the previous section we conclude that the Orion nebula
O/H value agrees with the O/H value predicted by the Galactic chemical
evolution model and the photospheric value by Asplund et al. (2005) and not
with the photospheric value by Grevesse & Sauval (1998). The discrepancy
between the photospheric abundances by Asplund et al. (2005) and the Y
value derived from helioseimological data is a very important open problem,
an excellent review discussing this issue has been presented by Basu & Antia
(2007).
Bahcall et al. (2006) present the initial Y , Z, and O solar values for
their standard solar models, see Table 6. By adopting the Yp value by
Peimbert et al. (2007) for t2 6= 0.00 it is also possible to obtain the ∆Y /∆Z
and the ∆Y /∆O values for the GS98 and the AGS05 standard solar models.
The values so derived are in fair agreement with the predictions of the HWY
and LWY Galactic chemical evolution models. To try to make a more rigorous
comparison between the solar interior and Galactic chemical evolution models
it is necessary to estimate the errors in the initial solar values by Bahcall et al.
(2006), a task which is beyond the scope of this paper.
7. CONCLUSIONS
Based on the HWY model we find the following equation to estimate the
initial helium abundance with which stars form in the Galactic disk
for O < 4.3 × 10−3, and
Y = Yp + (3.3 ± 0.7)O,
Y = Yp + (3.3 ± 0.7)O + (0.016 ± 0.003)(O/4.3 × 10−3 − 1)2,
for 4.3 × 10−3 < O < 9 × 10−3.
The increase of ∆F e/∆Z has to be taken into account in order to deter-
mine the ∆Y /∆Z value based on the [Fe/H] abundances of stars in the solar
vicinity.
22
CARIGI & PEIMBERT
High mass loss rates due to stellar winds should be adopted in the evolu-
tionary stellar models for massive stars of high metallicity, because only the
Galactic chemical evolution models with HWY can reproduce simultaneously
the O/H and C/O Galactic gradients, the C/O versus O relation in the solar
vicinity, and the ∆Y /∆O value in the inner Galactic disk.
Based on the O/H value of the Orion nebula and the solar photospheric
value together with a chemical evolution model of the Galaxy we recommend
for the present day local ISM a value of 12 + log O/H = 8.79 ± 0.08, where
both the gaseous and the dust components of O are taken into account.
By comparing the O/H value of the Orion nebula with the solar value
we find that the nebular ratio derived using O recombination lines, that is
equivalent to the use of the forbidden O lines under the adoption of a t2 6= 0.00,
is in considerably better agreement with the initial solar value than the Orion
nebula value derived adopting t2 = 0.00.
The stellar O/H and Ar/H abundance ratios derived by Cunha et al. (2006)
and Lanz et al. (2008) for B stars of the Orion association are in excellent
agreement with the nebular abundance ratios derived from the t2 6= 0.00 val-
ues for the Orion nebula.
The ∆Y /∆Z = 1.97 ± 0.41 value derived from observations of M17 for
t2 = 0.036 is in very good agreement with the 2.1 ± 0.4 and the 2.1 ± 0.9 values
derived by Jim´enez et al. (2003) and Casagrande et al. (2007) from K dwarf
stars of the solar vicinity. On the other hand the value ∆Y /∆Z = 4.00 ± 0.75
derived from observations of M17 for t2 = 0.000 is not.
Both Galactic chemical evolution models with the HWY set and the LWY
set are in agreement with the observed ∆Y /∆Z for t2 = 0.036 but not with
the ∆Y /∆Z for t2 = 0.000. Higher accuracy determinations of ∆Y /∆Z for
high metallicity objects are needed to discriminate between the HWY model
and the LWY model predictions.
We are grateful to Brad Gibson and Antonio Peimbert for several fruitful
discussions. We are also grateful to the anonymous referee for some excellent
suggestions. This work was partly supported by the CONACyT grants 46904
and 60354.
REFERENCES
Akerman, C. J., Carigi, L., Nissen, P. E., Pettini, M., & Asplund, M. 2004, A&A,
414, 931
Allen, C., Carigi, L., & Peimbert, M. 1998, ApJ, 494, 247
Allende Prieto, C. 2007, 14th Cambridge Workshop on Cool Stars, Stellar Systems
and the Sun, ed. G. van Belle, ASP Conference Series, in press, astro-ph/0702429
Aller, L. H. 1984, Physics of Thermal Gaseous Nebulae (Dordretch: Reidel)
Anders, E., & Grevesse, N. 1989, Geochimica et Cosmochimica Acta, 53, 197
Arnett, D. 1991, in: Frontiers of Stellar Evolution, ed. D. L. Lambert, ASP Confer-
ence Series, 20, 389
HE ENRICHMENT OF THE GALACTIC ISM
23
Arzumanov, S., et al. 2000, Physics Letters B, 483, 15
Asplund, M., Grevesse, N., & Sauval, A. J. 2005, in: Cosmic Abundances as Records
of Stellar Evolution and Nucleosynthesis, ed. F. N. Bash & T. G. Barnes, ASP
Conference Series, 336, 25
Bahcall, J. N., Serenelli, A. M., & Basu, S. 2006 ApJS, 165, 400
Basu, S., & Antia, H. M. 2004, ApJ, 606, L85
Basu, S., & Antia, H. M. 2007, Physics Reports, in press, arXiv:0711.4590
Benjamin, R. A., Skillman, E. D., & Smits, D. P. 2002, ApJ, 569, 288
Caffau, E., Ludwig, H.-G., Steffen, M., Ayres, T. R., Bonifacio, P., Cayrel, R., Frey-
tag, B., Plez, B. 2008, A&A, submitted, arXiv0805.4398
Carigi, L. 1994, ApJ, 424, 18
Carigi, L. 1996, RevMexAA, 32, 179
Carigi, L. 2000, RevMexAA, 36, 171
Carigi, L., Peimbert, M., Esteban, C., & Garc´ıa-Rojas, J. 2005, ApJ, 623, 213
Casagrande, L., Flynn, C., Portinari, L., Girardi, L. & Jim´enez, R. 2007, MNRAS,
382, 1516
Centeno, R., & Socas-Navarro, H. 2008, ApJ, 682, L61
Chiappini, C., Matteucci, F., & Ballero, S. K. 2005, A&A, 437, 429
Chiappini, C., Matteucci, F., & Meynet, G. 2003, A&A, 410, 257
Chieffi A., & Limongi M. 2002, ApJ, 577, 281
Chiosi, C., & Matteucci, F. 1982, A&A, 105, 140
Cuhna, K., Hubeny, I., & Lanz, T. 2006, ApJ, 647, L143
Deharveng, L., Pena, M., Caplan, J., & Costero, R. 2000, MNRAS, 311, 329
Dias, W. S., Alessi, B. S., Moitinho, A., & L´epine, J. R. D. 2002, A&A, 389, 871
Dunkley, J., et al. 2008, ApJ, submitted, arXiv0803.0586
Esteban, C., Garc´ıa-Rojas, J., Peimbert, M., Peimbert, A., Ruiz, M. T., Rodr´ıguez,
M., & Carigi, L. 2005, ApJ, 618, L95
Esteban, C., Peimbert, M., Garc´ıa-Rojas, J., Ruiz, M. T., Peimbert, A., &
Rodr´ıguez, M. 2004, MNRAS, 355, 229
Esteban, C., Peimbert, M., Torres-Peimbert, S., & Escalante, V. 1998, MNRAS, 295,
401
Esteban, C., Peimbert, M., Torres-Peimbert, S., & Garc´ıa-Rojas J. 1999,
RevMexAA, 35, 65
Fenner, Y., & Gibson, B. K. 2003 PASA, 20, 189
Garc´ıa-Rojas, J., Esteban, C., Peimbert, A., Rodr´ıguez, M., Peimbert, M., & Ruiz,
M. T. 2007, RevMexAA, 43, 3
Gibson, B. K. 1997, MNRAS, 290, 471
Gibson, B. K. , MacDonald, A. J., Sanchez-Blazquez, P. Carigi, L. 2006, in The
Metal-Rich Universe, eds. G. Israelian & G. Meynet, (Cambridge Univ. Press),
in press (arXiv:astro-ph/0611879)
Goldberg, L., Muller, E. A., & Aller, L. H. 1960, ApJS, 5, 1
Gonz´alez Delgado, R. M., Cervino, M., Martins, L. P., Leitherer, C., & Hauschildt,
P. H. 2005, MNRAS, 357, 945
Gonz´alez Delgado, R. M., Leitherer, C., & Heckman, T. M. 1999, ApJS, 125, 489
Grevesse, N., & Sauval, A.J. 1998, Space Sci. Rev. 85, 161
Hirschi, R. 2007, A&A, 461, 571
Hirschi, R., Meynet, G., & Maeder, A. 2005, A&A, 433, 1013
Holweger, H. 2001,
in Solar and Galactic Composition, ed. R. F. Wimmer-
Schweingruber, AIP Conference Proceedings, 598, 23
24
CARIGI & PEIMBERT
Jim´enez, R., Flynn, C., MacDonald, J., & Gibson, B. K. 2003, Science, 299, 1552
Keenan, F. P., Aller, L. H., Ramsbottom, C. A., Bell, K. L., Crawford, F. L., &
Hyung, S. 2000, P. Natl. Acad. Sci. USA, 97, 4551
Kingdon, J., & Ferland, G. 1995, ApJ, 442, 714
Kroupa, P., Tout, C. A., & Gilmore, G. 1993, MNRAS, 262, 545
Lambert, D. L. 1968, MNRAS, 138, 143
Lanz, T., Cunha, K., Holzman, J., Hubeny, I. 2008, ApJ, 678, 1342
Leone, F., & Lanzafame, A. C. 1998, A&A, 330, 306
Liu, X.-W., Storey, P. J., Barlow, M. J., Danziger, I. J., Cohen, M., & Bryce, M.
2000, MNRAS, 312, 585
Maeder, A. 1992, A&A, 264, 105
Marigo, P., Bressan, A., & Chiosi, C. 1996, A&A, 313, 545
Marigo, P., Bressan, A., & Chiosi, C. 1998, A&A, 331, 580
Mathews, G. J., Kajino, T., & Shima, T. 2005, Physical Review D, 71, 021302
McWilliam, A., Matteucci, F., Ballero, S., Rich, R. M., Fulbright, J. P., & Cescutti,
G. 2007, AJ, submitted (arXiv:0708.4026)
Meynet G., & Maeder A. 2002, A&A, 390, 561
Osterbrock D. E., Tran H. D., & Veilleux S. 1992, ApJ, 389, 196
Peimbert, A., Peimbert, M., & Luridiana, V. 2002 ApJ, 565, 668
Peimbert, M. 1967, ApJ, 150, 825
Peimbert, M., & Costero, R. 1969, Bol. Obs. Tonantzintla y Tacubaya, 5, 3
Peimbert, M., Luridiana, V., & Peimbert, A. 2007, ApJ, 666, 636
Peimbert, M., & Torres-Peimbert, S. 1977, MNRAS, 179, 217
Peimbert, M., Torres-Peimbert, S., & Ruiz, M. T. 1992, RevMexAA, 24, 155 (PTR)
Pilyugin, L. S. 2003, A&A, 401, 557
Porter, R. L., Bauman, R. P., Ferland, G. J., & MacAdam, K. B. 2005, ApJ, 622L,
73
Porter, R. L., Ferland, G. J., & MacAdam, K. B. 2007, ApJ, 657, 327
Portinari, L., Chiosi, C., & Bressan, A. 1998, A&A, 334, 505
Ross, J. E., & Aller, L. H. 1976, Science, 191, 1223
Sawey, P. M. J., & Berrington, K. A., 1993, Atomic Data and Nuclear Data Tables,
55, 81
Serebrov, A., et al. 2005, Physics Letters B, 605, 72
Storey, P. J., & Hummer, D. G. 1995, MNRAS, 272, 41
Thielemann, F. K., Nomoto, K., & Hashimoto, M. 1993, in Origin and Evolution of
the Elements eds. N. Prantzos et al., Cambridge University Press, p. 297
Woosley, S. E., & Weaver, T. A. 1995, ApJS, 101, 181
Both authors: Instituto de Astronom´ıa, Universidad Nacional Aut´onoma de
M´exico, Apdo. Postal 70-264, M´exico 04510 D.F., M´exico (carigi, peim-
[email protected]).
Leticia Carigi: Centre for Astrophysics, University of Central Lancashire, Pre-
ston, Lancashire, PR1 2HE, United Kingdom ([email protected]).
|
astro-ph/9804177 | 2 | 9804 | 1998-06-22T10:00:19 | Assisted inflation | [
"astro-ph",
"gr-qc",
"hep-ph"
] | In inflationary scenarios with more than one scalar field, inflation may proceed even if each of the individual fields has a potential too steep for that field to sustain inflation on its own. We show that scalar fields with exponential potentials evolve so as to act cooperatively to assist inflation, by finding solutions in which the energy densities of the different scalar fields evolve in fixed proportion. Such scaling solutions exist for an arbitrary number of scalar fields, with different slopes for the exponential potentials, and we show that these solutions are the unique late-time attractors for the evolution. We determine the density perturbation spectrum produced by such a period of inflation, and show that with multiple scalar fields the spectrum is closer to the scale-invariant than the spectrum that any of the fields would generate individually. | astro-ph | astro-ph | Assisted inflation
Andrew R. Liddle, Anupam Mazumdar and Franz E. Schunck
Astronomy Centre, University of Sussex, Falmer, Brighton BN1 9QJ, U. K.
(August 4, 2018)
In inflationary scenarios with more than one scalar field, inflation may proceed even if each of the
individual fields has a potential too steep for that field to sustain inflation on its own. We show
that scalar fields with exponential potentials evolve so as to act cooperatively to assist inflation, by
finding solutions in which the energy densities of the different scalar fields evolve in fixed proportion.
Such scaling solutions exist for an arbitrary number of scalar fields, with different slopes for the
exponential potentials, and we show that these solutions are the unique late-time attractors for the
evolution. We determine the density perturbation spectrum produced by such a period of inflation,
and show that with multiple scalar fields the spectrum is closer to the scale-invariant than the
spectrum that any of the fields would generate individually.
PACS numbers: 98.80.Cq
Sussex preprint SUSSEX-AST 98/4-3, astro-ph/9804177
8
9
9
1
n
u
J
2
2
2
v
7
7
1
4
0
8
9
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
I. INTRODUCTION
The idea of cosmological inflation [1,2] is an attrac-
tive one, solving a range of otherwise troubling prob-
lems. Inflation is normally achieved by a period of the
Universe's evolution during which the energy density is
dominated by the potential energy of a scalar field. Al-
though quite probably the early Universe contained sev-
eral scalar fields, it is normally assumed that only one of
these fields remained dynamically significant for a long
time, with the others rapidly finding their way into the
minima of their respective potential energies.
In this paper we consider scalar fields with exponential
potentials. These are already known to have interesting
properties; for example, if one has a universe containing
a perfect fluid and such a scalar field, then for a wide
range of parameters the scalar field 'mimics' the perfect
fluid, adopting its equation of state [3,4]. These scaling
solutions are attractors [5] at late times. The behaviour
of such a field during an inflationary epoch has also been
considered [5].
What was not considered in Ref. [5] is the effect of in-
troducing a scalar field with an exponential potential on
the other scalar field. The simplest example would be if
the other field also possessed an exponential potential.
Then the behaviour of both fields will be modified, since
they feel only their own potential gradient, but experi-
ence, via the expansion, the frictional effect of all scalar
fields present.
where mPl is the Planck mass. Note that there is no
direct coupling of the fields, which influence each other
only via their effect on the expansion. The equations of
motion are
H 2 =
8π
3m2
Pl
φi = −3H φi −
m
Xi=1(cid:20)V (φi) +
dV (φi)
.
dφi
1
2
φ2
i(cid:21) ;
(2)
(3)
Our fields are combined additively; this is different from
soft inflation [6], where an exponential potential multi-
plies the potential of another scalar field.
If there is only a single scalar field, this leads to the
well-known power-law solution [7]
a(t) ∝ tp .
(4)
This is inflationary only if p > 1, i.e. for sufficiently shal-
low exponentials. The power-law solution also applies for
any p in the range 1/3 to 1, where it is non-inflationary.
For p < 1/3, the asymptotic solution is that of a free
scalar field, with a ∝ t1/3 regardless of the value of p in
the range (0, 1/3).
We first find a particular solution where all the scalar
fields are equal: φ1 = φ2 = · · · = φm. We shall later show
it is the unique late-time attractor. With this ansatz the
equations become
H 2 =
8π
3m2
Pl
m (cid:20)V (φ1) +
dV (φ1)
1
2
φ2
1(cid:21) ;
.
(5)
(6)
dφ1
II. DYNAMICS
φ1 = −3H φ1 −
For simplicity, we begin by considering m scalar fields,
φi, which each have an identical potential
V (φi) = V0 exp(cid:18)−r 16π
p
φi
mPl(cid:19) ,
(1)
1
These can be mapped to the equations of a model with
a single scalar field φ by the redefinitions
φ2
1 = m φ2
1
;
V = m V
;
p = mp ,
(7)
so the expansion rate is a ∝ t p, provided that p > 1/3.
The expansion becomes quicker the more scalar fields
there are. And in particular, potentials with p < 1, which
for a single field are unable to support inflation, can do so
as long as there are enough scalar fields to make mp > 1.
Note also that this solution does not require p to exceed
1/3, only the product mp. If mp is less than one third
then the solution will instead be that of a free scalar field.
Although the solution with all scalar fields equal is a
particular one, it is in fact the generic late-time attrac-
tor. To see this, keep φ1 but replace the rest with the
redefined fields
ψi = φi − φ1 ,
i = 2, . . . , m .
These fields obey the equation
ψi + 3H ψi =
p
V0
mPlr 16π
×(cid:20)exp(cid:18)−r 16π
exp(cid:18)−r 16π
mPl(cid:19) − 1(cid:21) ,
ψi
p
p
φ1
mPl(cid:19)
(8)
(9)
i = 2, . . . , m .
This is the equation of a scalar field in an effective po-
tential
ψi + 3H ψi = −
∂Veff(φ1, ψi)
∂ψi
,
(10)
with
p
Veff = V0r 16π
×(cid:20)r p
16π
p
exp(cid:18)−r 16π
exp(cid:18)−r 16π
φ1
mPl(cid:19)
mPl(cid:19) +
ψi
p
ψi
mPl(cid:21) .
(11)
The minimum in the ψi direction is always at ψi = 0,
regardless of the behaviour of φ1, so the late-time solution
has all the φi equal. The length of time to reach this
attractor will depend on the initial separation (the value
of ψi) and the extent to which friction, coming from the
expansion rate H, is important.
III. DENSITY PERTURBATIONS
It is now well known how to calculate the density per-
turbation produced in multi-scalar field models. Sasaki
and Stewart [8] (see also [9]) quote the result
PR =(cid:18) H
2π(cid:19)2 ∂N
∂φi
∂N
∂φj
δij ,
(12)
where PR is the spectrum of the curvature perturbation
R in the usual units [2], N is the number of e-foldings of
inflationary expansion remaining, and there is a summa-
tion over i and j. Since N = −R H dt, we have
φi = −H ,
∂N
∂φi
Xi
(13)
2
where in our case each term in the sum is the same,
yielding
PR =(cid:18) H
2π(cid:19)2 1
m
H 2
φ2
1
.
(14)
Note that this last expression only contains one of the
scalar fields, chosen arbitrarily to be φ1. This expression
looks as if it is m times smaller than the usual formula for
a single scalar field (see e.g. Ref. [2]); however, remember
that the presence of multiple fields has modified both H
and φ1.
Of particular interest is the spectral index n. This is
given by [8]
n − 1 = 2
H
H 2
− 2
∂N
∂φi (cid:16) 8π
m2
Pl
φi φj
H 2 −
∂N
δij
∂φi
Pl
m2
8π
∂N
∂φj
V,i,j
V (cid:17) ∂N
∂φj
, (15)
where there is a summation over repeated indices and
the commas indicate derivatives with respect to the cor-
responding field component. Under our assumptions, the
complicated second term on the right-hand side of the
above equation cancels out, and Eq. (15) reduces to the
simple form
1 − n = −2
H
H 2 =
m2
Pl
8π " ∂V (φ1)
V (φ1)#
∂φ1
2
=
2
mp
.
(16)
This result shows that the spectral index also matches
that produced by a single scalar field with p = mp. The
more scalar fields there are, the closer to scale-invariance
is the spectrum that they produce. Note however that if
the fields have such steep potentials as to be individually
non-inflationary, p < 1, then many fields are needed be-
fore the spectrum is flat enough (say n > 0.7) to have the
possibility of explaining the observed structures. Large
numbers of scalar fields are predicted by some theories,
for example the 70 scalar fields, with unknown poten-
tials, of the low-energy compactified superstring effective
action [10].
IV. POTENTIALS WITH DIFFERENT SLOPES
We now generalize the above discussion, by considering
each potential to have a different slope pi
Vi(φi) = V0 exp(cid:18)−r 16π
pi
φi
mPl(cid:19) .
(17)
Notice that we keep the same V0 for each field; since
changing V0 is equivalent to shifting the scalar field defi-
nition by a constant, this serves to fix the zero values of
the fields.
We conjecture that scaling solutions exist, where the
energy densities of the different fields attain fixed ratios
at late times, relative to an arbitrarily chosen field φ1:
φ2
i
φ2
1
=
Vi(φi)
V1(φ1)
= Ci .
which generalizes Eq. (9) to
(18)
To guess the appropriate form of Ci, we note that the
slow-roll approximation gives
which suggests
φ2
i
≃
2
3pi
,
V 2
i (φi)
Pi Vi(φi)
Ci =
pi
p1
.
(19)
(20)
We stress though that the slow-roll approximation is not
needed in what follows.
Integrating the kinetic part of Eq. (18) then gives
ψi + 3H ψi =
pi
V0
mPlr 16π
p1 (cid:20)exp(cid:18)−r 16π
exp(cid:18)−r 16π
mPl(cid:19) − 1(cid:21) ,
ψi
p1
pi
pi
×
φ1
mPl(cid:19)
(27)
i = 2, . . . , m .
As before, the effective potentials for the ψi fields have
a unique minimum at ψi = 0 for all i = 2, . . . , m. Our
scaling solution is therefore the unique late-time attrac-
tor.∗
The calculation of the spectral index follows the same
lines as before, yielding
1 − n =
2
p
.
(28)
φi =r pi
p1
φ1 + αi ,
(21)
This reduces to Eq. (16) when the slopes of the potentials
are same.
where αi are the integration constants. Ensuring the
potentials also scale as in Eq. (18) requires the constants
to have values
αi = −r pi
16π
mPl ln
pi
p1
.
(22)
To see that this solution will solve the full dynamical
equations, we generalize the scaling argument of Section
II. Eq. (21) reduces us to a single degree of freedom φ1 in
a manner consistent with the equations of motion, which
become
H 2 =
8π
3m2
Pl Pm
i=1 pi
p1
(cid:20)V1(φ1) +
φ1 = −3H φ1 −
dV1(φ1)
.
dφ1
1
2
φ2
1(cid:21) ;
(23)
(24)
Using the scaling of the potential from Eq. (18), the other
scalar wave equations all match the latter of these, con-
firming consistency of the ansatz. Note in particular that
Eq. (21) brings all the exponentials into the same form.
These can then be turned into the equations of a model
with a single scalar field via the redefinitions
φ2
1 = P pi
p1
φ2
1
;
V1 = P pi
p1
V1
;
pi , (25)
p =Xi
of which Eq. (7) is a special case. This result is exact,
not requiring a slow-roll approximation, and once more
shows that the presence of multiple scalar fields increases
the expansion rate. The expansion law is a ∝ t p, and is
valid provided that p > 1/3.
The scaling construction shows that this solution ex-
actly solves the multi-scalar field equations. One can
show that this solution is an attractor by generalizing
the argument of Section II, via the ansatz
V. CONCLUSION
Although the early Universe is likely to contain many
scalar fields, a common assumption when analyzing in-
flation is that all but one of these fields has become dy-
namically irrelevant. However, for scalar fields with ex-
ponential potentials the late-time behaviour is for the
energy densities of the different fields to scale with each
other, as had already been noted for the case of a scalar
field with an exponential potential plus a barotropic fluid
[3 -- 5], even if the fields have no direct coupling to each
other and if their potentials have different slopes.
Such multiple scalar fields can act cooperatively to
drive a period of inflation, even if the individual fields
have potentials which are too steep in their own right;
the expansion law in the scaling solution is t p, where
p =P pi with the pi being the power-law expansion rates
that the individual fields would drive in isolation. The
reason for this behaviour is that while each field experi-
ences the 'downhill' force from its own potential, it feels
the friction from all the scalar fields via their contribution
to the expansion rate.
We have also studied the density perturbation spec-
trum produced, which has a spectral index n matching
that of power-law inflation driven by a single field at rate
p. The spectrum is therefore brought closer to scale-
invariance the more fields participate in the inflationary
expansion. A perturbation spectrum close to scale in-
variance is preferred by current observations, and this
phenomena may offer assistance to supergravity-based
inflation models which often predict spectra which are
not all that close to scale invariance [11].
ψi = φi −r pi
p1
φ1 − αi ,
i = 2, . . . , m .
(26)
∗We have also confirmed these solutions as late-time attrac-
tors numerically for a wide range of values of pi.
3
ACKNOWLEDGMENTS
A.R.L. is supported by the Royal Society, A.M. by the
Inlaks foundation and the ORS, and F.E.S. by the Euro-
pean Union TMR Marie Curie programme. We thank Ed
Copeland, Jim Lidsey and David Wands for discussions,
and acknowledge use of the Starlink computer system at
the University of Sussex.
[1] A. Guth, Phys. Rev. D 23, 347 (1981); E. W. Kolb and
M. S. Turner, The Early Universe, Addison -- Wesley, Red-
wood City (1990).
[2] A. R. Liddle and D. H. Lyth, Phys. Rep. 231, 1 (1993).
[3] C. Wetterich, Nucl. Phys. B302, 668 (1988).
[4] D. Wands, E. J. Copeland and A. R. Liddle, Ann. N. Y.
Acad. Sci. 688, 647 (1993).
[5] E. J. Copeland, A. R. Liddle and D. Wands, Phys. Rev.
D 57, 4686 (1998).
[6] A. L. Berkin, K. Maeda and J. Yokoyama, Phys. Rev.
Lett. 65, 141 (1990); G. Abolghasem, A. Burd, A. Coley
and R. van den Hooden, Phys. Rev. D 48, 557 (1993).
[7] F. Lucchin and S. Matarrese, Phys. Rev. D 32, 1316
(1985).
[8] M. Sasaki and E. D. Stewart, Prog. Theor. Phys. 95, 71
(1996).
[9] A. A. Starobinsky and J. Yokoyama, Report. No. gr-
qc/9502002 (1995); J. Garc`ıa-Bellido and D. Wands,
Phys. Rev. D 53, 5437 (1996).
[10] E. Cremmer and B. Julia, Phys. Lett. B80, 48 (1978);
Nucl. Phys. B156, 141 (1979).
[11] D. H. Lyth, Lancaster preprint hep-ph/9609431.
4
|
astro-ph/9907197 | 1 | 9907 | 1999-07-14T23:30:14 | Warm Gas and Ionizing Photons in the Local Group (ApJL version) | [
"astro-ph"
] | Several lines of argument suggest that a large fraction of the baryons in the universe may be in the form of warm (T\sim 10^5-10^7 K) gas. In particular, loose groups of galaxies may contain substantial reservoirs of such gas. Observations of the cosmic microwave background by COBE place only weak constraints on such an intragroup medium within the Local Group. The idea of a Local Group corona dates back at least forty years (Kahn & Woltjer 1959). Here we show that gas at T\sim 2-3 X 10^6 K (the approximate virial temperature of the Local Group) -- extremely difficult to observe directly -- can in principle radiate a large enough flux of ionizing photons to produce detectable H alpha emission from embedded neutral clouds. However, additional constraints on the corona -- the most stringent being pulsar dispersion measures towards the Magellanic Clouds, and the timing mass -- rule out an intragroup medium whose ionizing flux dominates over the cosmic background or the major Local Group galaxies. A cosmologically significant coronal gas mass could remain invisible to H alpha observations. More massive galaxy groups could contain extensive coronae which are important for the baryon mass and produce a strong, local ionizing flux. | astro-ph | astro-ph |
Warm Gas and Ionizing Photons in the Local Group
Philip R. Maloney1 and J. Bland-Hawthorn2
ABSTRACT
Several lines of argument suggest that a large fraction of the baryons in the universe
may be in the form of warm (T ∼ 105 − 107 K) gas. In particular, loose groups of
galaxies may contain substantial reservoirs of such gas. Observations of the cosmic
microwave background by COBE place only weak constraints on such an intragroup
medium within the Local Group. The idea of a Local Group corona dates back at
least forty years (Kahn & Woltjer 1959). Here we show that gas at T ∼ 2 − 3 × 106
K (the approximate virial temperature of the Local Group) -- extremely difficult to
observe directly -- can in principle radiate a large enough flux of ionizing photons to
produce detectable Hα emission from embedded neutral clouds. However, additional
constraints on the corona -- the most stringent being pulsar dispersion measures
towards the Magellanic Clouds, and the timing mass -- rule out an intragroup medium
whose ionizing flux dominates over the cosmic background or the major Local Group
galaxies. A cosmologically significant coronal gas mass could remain invisible to Hα
observations. More massive galaxy groups could contain extensive coronae which are
important for the baryon mass and produce a strong, local ionizing flux.
Subject headings: Local Group -- cosmic microwave background -- intergalactic medium
-- diffuse radiation
1.
Introduction
The standard Big Bang cosmological model makes remarkably precise predictions for
the abundance of baryons in the universe:
in terms of the critical density parameter Ω, the
prediction is Ωb ≃ (0.068 ± 0.012)h−2, where we take the present-day Hubble constant to be
H0 = 100h km s−1 Mpc−1 (e.g., Olive et al. 1991; Schramm & Turner 1998). An inventory of
baryons observed at high redshift (z ∼ 2 − 3), chiefly in the form of the low-column density
Lyα-forest clouds, gives an estimate of Ωb which, although subject to substantial, systematic
uncertainties, is in reasonable agreement with the standard prediction of Big Bang nucleosynthesis
1Center
for Astrophysics and Space Astronomy, University of Colorado, Boulder, CO 80309-0389;
[email protected]
2Anglo-Australian Observatory, P.O. Box 296, Epping, NSW 2121, Australia; [email protected]
-- 2 --
(see the summary in Fukugita, Hogan, & Peebles 1998). However, as has been noted repeatedly
(e.g., Persic & Salucci 1992; Fukugita et al. 1998), at z ≈ 0 only a small fraction of the expected
number of baryons has been observed, suggesting that there is a substantial, even dominant
reservoir of baryons which has not yet been characterized.
A plausible suggestion for one reservoir of baryons is that loose groups of galaxies contain
substantial masses of warm (T < 107 K) ionized gas, an idea which appears to have originated
with Kahn & Woltjer (1959; see also Oort 1970; Hunt & Sciama 1972). X-ray observations of
poor groups of galaxies frequently detect intragroup gas at T ∼ 1 keV (e.g., Pildis, Bregman, &
Evrard 1995; Mulchaey et al. 1996). In general, only groups dominated by ellipticals are detected;
spiral-rich groups tend to show only emission from individual galaxies. Although this may be due
to the absence of gas in such groups, it is also plausible that the gas has not been seen because
its temperature is too low: the velocity dispersions characterizing groups dominated by spiral
galaxies are significantly smaller than those of compact, elliptical-dominated groups, and imply
temperatures T ∼ 0.2 − 0.3 keV (Mulchaey et al. 1996), making detection even at relatively soft
X-ray wavelengths very difficult.
Most recently, Blitz et al. (1998) have suggested that the majority of high-velocity clouds
(HVCs; for a review, see Wakker & van Woerden 1997) are not associated with the Galactic ISM,
but represent remnants of the formation of the Local Group (LG), as material continues to fall
into the LG potential. In this scenario, some fraction of these infalling clouds will collide in the
vicinity of the LG barycenter and shock up to the virial temperature, T ∼ 2 × 106 K, producing a
warm intragroup medium.
In this paper, we explore the possibility that the Local Group contains such a reservoir of
warm ionized gas. In particular, we examine whether significant constraints can be placed on the
amount of gas through the detection of recombination lines from neutral gas within the Local
Group. In the next section we briefly recapitulate the existing constraints on such an intragroup
medium, and in §3 we estimate the flux of ionizing photons. §4 discusses the implications and
additional constraints which can be imposed, in particular, mass flux due to cooling and the
timing mass of the Local Group.
2. COBE and X-Ray Constraints on Local Group Gas
Suto et al. (1996) suggested that a gaseous LG halo could significantly influence the CMB
quadrupole moment observed by COBE. Assume the Local Group contains an isothermal plasma
at temperature Te whose electron number density is (for core density no and core radius ro)
ne(r) = no
r2
o
r2 + r2
o
cm−3;
(1)
i.e., the nonsingular isothermal sphere. Since we allow ro as well as no to vary, the parameterization
of equation (1) includes density distributions ranging from ne ≈ constant to ne ∝ r−2. As in Suto
-- 3 --
et al. (1996), we calculate the resulting Sunyaev-Zeldovich temperature decrement as a function
of angle, expand in spherical harmonics and average over the sky to obtain the monopole and
quadrupole anisotropies. The COBE FIRAS data (Fixsen et al. 1996) imply that the Compton
y-parameter y = T0,sz/2 < 1.5 × 10−5 (95% CL), which imposes the constraint
noroTkeV < 7.4 × 1021θ−1
o
R
ro (cid:18)
y
1.5 × 10−5(cid:19) cm−2,
where θo ≡ tan−1(R/ro). Similarly, the COBE quadrupole moment requires
noroTkeV < 1.6 × 1020QµK
R
ro "θo − 3(cid:18) ro
R(cid:19) + 3θo(cid:18) ro
R(cid:19)2#−1
(2)
(3)
cm−2
where the rms quadrupole amplitude QRMS = 10−6QµK K; the observed value QµK ≈ 6 (e.g.,
Bennett et al. 1996). Suto et al. (1996) argued that a LG corona which satisfied equation (2) could
significantly affect the quadrupole term, as equation (3) is more restrictive than (2). However,
Banday & Gorski (1996) showed there is no evidence for a LG corona in the COBE DMR skymaps.
In addition, Pildis & McGaugh (1996) pointed out that the typical values of noroTkeV observed
in poor groups of galaxies, resembling the Local Group, are well below the limit (2), generally
no more than a few ×1020 cm−2. Furthermore, spiral-rich groups usually reveal no evidence for
intragroup gas at X-ray energies; Pildis & McGaugh give upper limits of a few ×1019 cm−2 for an
assumed temperature Tkev ∼ 1.
Thus, although the COBE constraints on a LG corona are in fact quite weak3, analogy
with similar poor groups suggests that the LG is unlikely to have a significant gaseous X-ray
corona. However, as noted in §1, the lower temperature expected for the gas in spiral-rich
groups significantly relaxes the X-ray constraints on warm gas in groups similar to the LG. If the
product noroTkeV in a LG corona is typical of that seen in more compact groups, merely at lower
temperature, the mass in baryons can still be very substantial: for the density distribution (1),
scaling noroTkeV to 1020 cm−2, the mass inside radius r is approximately (assuming r/ro
>∼ a few)
M (r) ≈ 7 × 1011(cid:18) ro
100 kpc(cid:19)2(cid:18) r
ro(cid:19)(cid:18) noroTkeV
1020 cm−2(cid:19)(cid:18) TkeV
0.2 (cid:19)−1
M⊙ ;
(4)
this could be a substantial fraction of the mass of the Local Group (see §4).
Direct detection of emission from gas at such temperatures is exceedingly difficult. Using
deep ROSAT observations, Wang & McCray (1993) (WM) find evidence for a diffuse thermal
component with TkeV ∼ 0.2 and ne ∼ 1 × 10−2 x−0.5
kpc cm−3 (assuming primordial gas) where xkpc
is the line-of-sight depth within the emitting gas in kiloparsecs. In the next section we consider
an indirect method of detection: the recombination radiation from neutral gas embedded in the
corona, due to the ionizing photon flux generated by the corona gas.
3There is some confusion in the literature regarding the interpretation of the COBE limits. In evaluating eq. (9)
of Suto et al. (1996), there is no numerical fudge factor suggested by Pildis & McGaugh (1996). Moreover, in Fig. 4
of Suto et al. , the 6µK curve is displaced downwards by a factor of 3.
-- 4 --
3.
Ionizing Photon Flux from a Local Group Corona
We assume the density distribution (1). Approximating the surface of a cloud as a
plane-parallel slab, the normally incident flux on the inner (facing r = 0) cloud face is
φi(r) ≈
πn2
oro
(1 + r2/r2
o)1.5 ξih0.8 + 1.3(r/ro)1.35i phot cm−2 s−1
(5)
where ξi is the frequency-integrated ionizing photon emissivity and the term in brackets is
accurate to 10% for 10−3 ≤ r/ro ≤ 12. (For r/ro
>∼ 2, the flux on the outer face of the cloud is
insignificant.) To calculate ξi, we have used the photoionization/shock code MAPPINGS (kindly
provided by Ralph Sutherland). Models have been calculated for metal abundances Z = 0.01, 0.1,
and 0.3 times solar, and for equilibrium and nonequilibrium ionization. For 104 < T < 107 K,
3 × 10−15 ≤ ξi ≤ 3 × 10−14 phot cm−3 s−1 sr−1. Scaling to physical values,
φi(r) ≈ 104n2
−3r100(cid:18) ξi
10−14(cid:19) (cid:2)0.8 + 1.3(r/ro)1.35(cid:3)
(1 + r2/r2
phot cm−2 s−1
(6)
o)1.5
where the central density no = 10−3n−3 cm−3 and the core radius ro = 100r100 kpc. Poor groups
show a very broad range of core radii, from tens to hundreds of kpc (Mulchaey et al. 1996), and
typical central densities no ∼ a few ×10−3 cm−3 (Pildis & McGaugh 1996).
In Fig. 1, we plot φi as a function of core radius ro for densities no = (1, 3, 10) × 10−3 cm−3,
for a metallicity Z = 0.1 times solar; results differ by <∼ 20% for the other values of Z. The
value of φi is evaluated at r = 350 kpc, the assumed distance rMW of the Galaxy from the center
of the LG (solid lines), and at r = 0 (dashed lines). The fluxes can be very large, exceeding
106 phot cm−2 s−1. However, for ro ≪ rMW, equation (6) shows that the incident flux at rMW is
greatly diminished compared to the peak value of φi.
At distances r ∼ 2ro or less, the ionizing flux produced by a LG corona could be large enough
for detection in Hα: the emission measure is related to the normally incident photon flux by
Em = 1.25 × 10−2(φi/104) cm−6 pc. However, to produce a significant flux, no must be so large
<∼ 109 years. Even though the LG may be,
that the cooling time tc within r ∼ ro is short, tc
dynamically, considerably younger than a Hubble time, such a short cooling timescale makes it
necessary to consider explicitly the fate of cooling gas.
To estimate the mass cooling flux M , we assume that the flow is steady, spherical, and
subsonic, and that any gradients in the potential are small compared to the square of the sound
M = 4πρvr2; v is
speed. In this case the pressure is constant, and mass conservation requires that
the inflow velocity. The cooling radius rc is set by the condition tc ∼ tLG, the Local Group age.
c / M , where ρc is the gas density at rc. We assume
The flow time from rc is tf ∼ rc/v ∼ 4πρcr3
tf ∼ tc, so that the gas has time to cool before reaching r = 0. This sets v ∼ rc/tLG at rc. If
the cooling function Λ does not vary rapidly with T , the density and temperature within rc scale
nearly as ρ ∝ rc/r, T ∝ r/rc (Fabian & Nulsen 1977). We have used these scalings to calculate M
and φi, including the variation of ξi and Λ with radius. Fig. 2 shows several models. The ionizing
-- 5 --
M if ro is large and no is low, but in many cases
M is prohibitively
flux can be large for small
large, ruling out any such coronae. However, there are several important caveats. Unless the LG is
very old, it is unlikely that a steady-state flow has been established (e.g., Tabor & Binney 1993),
especially as infall of gas into the LG is likely to be ongoing. (If a steady-state flow existed with
M , one would expect the line luminosity -- e.g., Hα -- from the cooled gas to be high:
substantial
M is sensitive to the assumed density distribution. For a
see Donahue & Voit 1991.) Furthermore,
given metallicity (and therefore Λ(T )) and LG age, there is a unique value of no at which tc = tLG
and M → 0. As no is raised above this value M increases rapidly, since rc increases and M ∝ r2
c .
M also depends on Z, since the reduced Λ for low Z means
The value of φi at a given value of
that no is larger for a fixed tc. Given these uncertainties, it is not clear that the estimated values
of
M should be regarded as serious constraints.
4. Discussion
The results of the previous section show that a warm Local Group corona could in principle
generate a large enough ionizing photon flux to produce detectable Hα emission from neutral
hydrogen clouds embedded within it. This would offer an indirect probe of gas which is extremely
difficult to observe in emission. Whether the flux seen by clouds at distances comparable to the
offset of the Galaxy from the center of the Local Group is high enough for detection depends to a
large extent on the core radius characterizing the gas distribution, due to the dropoff in flux for r
substantially greater than ro. As shown in Fig. 1, for sufficiently large values of ro and no, φi can
be detectably large even at a few hundred kpc from the LG barycenter.
These large−no, large−ro models run into insurmountable difficulties, however, when we
examine the additional constraints which can presently be imposed on a LG corona. In Fig. 3 we
show, shaded in gray, the range in (ro, no) for which the resulting ionizing photon flux is between
φi = 104 and φi = 105 phot cm−2 s−1, for radial offsets r = 0 (lower region) and r = rMW = 350
kpc (upper region). The cosmic background is probably φi,cos ∼ 104 phot cm−2 s−1 (Maloney &
Bland-Hawthorn 1999: MBH). We also plot the following constraints:
(1) The assumption that any LG intragroup medium is "typical" (Mulchaey et al. 1996; Pildis
<∼ 1.5 × 1021 cm−2, assuming Tkev ∼ 0.2. This is
& McGaugh 1996) constrains the product noro
plotted as the short-dashed line in Fig. 3. Any corona which is not unusually rich must lie to the
left of this line. This restriction alone rules out any significant contribution to φi at rMW.
(2) Assuming that the relative velocity of approach of the Galaxy and M31 is due to their mutual
gravitational attraction, one can estimate the mass MT of the Local Group (Kahn & Woltjer
1959; q.v., Zaritsky 1994). This 'timing mass' depends somewhat on the choice of cosmology; we
take MT = 5 × 1012 M⊙ within r = 1 Mpc of the LG center. The timing mass constraint (using
equation [10]) is shown as the solid line in Fig. 3. As plotted, it is barely more restrictive than the
COBE quadrupole constraint (the long-dashed line), and is only more stringent than restriction
-- 6 --
(1) for large core radii. However, realistically the MT constraint is much more severe, as the Milky
Way and M31 undoubtedly dominate the mass of the Local Group, and so the timing mass curve
in Fig. 3 should be moved downward in density by a factor of at least ∼ 5 − 10.
(3) We possess some information on (more precisely, upper limits to) the actual electron densities
at r ∼ rMW. Constraints on ne(rMW) come from two sources. Observations of dispersion measures
Dm toward pulsars in the LMC and the globular cluster NGC 5024 (Taylor, Manchester & Lyne
1993) require a mean n−3 ∼ 1; this is a slightly weaker constraint than provided by MT . However,
most of this column must be contributed by the Reynolds layer, and some fraction of the Dm
toward the LMC pulsars presumably arises within the LMC, so probably <∼ 10% can be due
to a LG corona. Second, a mean density of no more than n−3 ∼ 0.1 is allowed by models of
the Magellanic Stream; otherwise, the Stream clouds would be plunging nearly radially into the
Galaxy (Moore & Davis 1994). This limits the central density to n−3 ≈ 0.1 + (rMW/ro)2. The
hatched region in Fig. 3 indicates the portion of (ro, no) space in which ne(rMW) ≤ 10−4 cm−3.
(4) As noted earlier (§2), WM found evidence for a thermal soft X-ray component at Tkev ≈ 0.2.
If this emission arises in a LG corona, then the corresponding electron density as derived from
the emission measure Em is ne ∼ 3 × 10−4 x−0.5
Mpc cm−3, where x is the extent of emitting region
along the line of sight; the density would be ∼ 3 times smaller for gas of solar rather than zero
metallicity. This density constraint is comparable to the Dm constraint plotted in Fig. 3.
Some of these constraints can be avoided if the corona gas is clumped. The estimates of mass
(equation[4]) and φi assume a smooth density distribution. However, if the actual densities are
a factor C higher than the mean (smoothed) density at a given radius, φi can be kept constant
while reducing both the gas mass and Dm by 1/C. This is ad hoc, but if the LG halo is being
fueled by ongoing infall, it would not be at all surprising for the gas distribution to be nonuniform.
However, the WM X-ray determination is unaffected by clumping, as it is derived from Em.
The constraints on a LG corona shown in Fig. 3 rule out a significant contribution to the
ionizing flux at r ∼ rMW. If the core density no is high, the core radius ro must be small; conversely,
for large ro, no must be low. LG coronae within the allowed region of parameter space can produce
fluxes φi ≫ φi,cos, but only on scales of a few tens of kpc, at best. Thus the maximum volume
in which a corona ionizing flux exceeds φi,cos is only of order 1% of the LG volume, comparable
to the volume which can be ionized by galaxies (MBH). This has important implications for the
model of Blitz et al. (1998), in which most HVCs are remnants of the formation of the Local
Group. If HVCs are at megaparsec distances, φi will be dominated by the cosmic background. The
resulting emission measures will be small: barring unusually favorable geometries, the expected
Hα surface brightnesses ( <∼ 10 mR) are at the limit of detectability. Any HVCs which are truly
extragalactic and detectable in Hα would need to lie close to the dominant spiral galaxies (within
their "ionization cones": Bland-Hawthorn & Maloney 1999a,b) or the LG barycenter.
In summary, a warm LG corona which significantly dominates the UV emission within the
Local Group is ruled out, although such a corona could contain a cosmologically significant
-- 7 --
quantity of baryons. More massive galaxy groups could well contain coronae that are both
cosmologically important and dominate over the ionizing background. Such coronae could have
major impact on the group galaxies through ionization and ram pressure stripping4. Finally, we
note that, four decades later, the observational limits on a LG corona have yet to improve on the
values suggested by Kahn & Woltjer (1959).
PRM is supported by the Astrophysical Theory Program under NASA grant NAG5-4061.
Bandy, A.J., & G´orski, K.M. 1996, MNRAS, 283, L21
REFERENCES
Bennett C.L., Banday A.J., G´orski K.M., Hinshaw, G., Jackson, P., Keegstra, P., Kogut, A.,
Smoot, G.F., Wilkinson, D.T., & Wright, E.L. 1996, ApJ, 464, L1
Bland-Hawthorn, J., & Maloney, P.R. 1999a, ApJ, 510, L33
Bland-Hawthorn, J., & Maloney, P.R. 1999b, in The Stromlo Workshop on High Velocity Clouds,
ed. B.K. Gibson & M.E. Putnam (San Francisco: ASP), 212.
Blitz, L., Spergel, D.N., Teuben, P.J., Hartmann, D., & Burton, W.B. 1998, ApJ, 514, 818
Donahue, M., & Voit, G.M. 1991, ApJ, 381, 361
Fabian, A.C., & Nulsen, P.E.J. 1977, MNRAS, 180, 479
Fixsen, D.J., Cheng, E.S., Gales, J.M., Mather, J.C., Shafer, R.A., & Wright, E.L. 1996, ApJ, 473,
576
Fukugita, M., Hogan, C.J., & Peebles, P.J.E. 1998, ApJ, 503, 518
Hunt, R., & Sciama, D.W. 1972, MNRAS, 157, 335
Kahn, F.D., & Woltjer, L. 1959, ApJ, 130, 705
Maloney, P.R., & Bland-Hawthorn, J. 1999, in The Stromlo Workshop on High Velocity Clouds,
ed. B.K. Gibson & M.E. Putnam (San Francisco: ASP), 199.
Moore, B., & Davis, M. 1994, MNRAS, 270, 209
Mulchaey, J.S., Davis, D.S., Mushotzky, R.F., & Burstein, D. 1996, ApJ, 456, 80
4We note that, in principle, observations of the O VI doublet at 1032 and 1038 A are extremely sensitive to the
presence of such a corona: for the maximum allowed coronae of Fig. 3, the expected line fluxes could be as large as
F ∼ a few × 10−10 erg cm−2 s−1. However, the observational difficulties (absorption and scattering of the photons
within the ISM of the Galaxy and the very large spatial extent of the source for a LG corona) are severe.
-- 8 --
Olive, K.A.; Steigman, G., & Walker, T.P. 1991, ApJ, 380, L1
Oort, J.H. 1970, A& A, 7, 381
Persic, M., Salucci, P. 1992, MNRAS, 258, 14P
Pildis, R.A., & McGaugh, S.S. 1996, ApJ, 470, L77
Pildis, R.A., Bregman, J.N., & Evrard, A.E. 1995, ApJ, 443, 514
Schramm, D. N., & Turner, M. S. 1998, Rev. Mod. Phys., 70, 303
Suto, Y., Makishima, K., Ishisaki, Y., & Ogasaka, Y. 1996, ApJ, 461, L33
Tabor, G., & Binney, J. 1993, MNRAS, 263, 323
Taylor, J.H., Manchester, R.N., & Lyne, A.G. 1993, ApJS, 88, 529
Wakker, B.P., & van Woerden, H. 1997, ARA&A, 35, 217
Wang, Q.D., & McCray, R.M. 1993, ApJ, 409, L37
Zaritsky, D. 1994, in The Local Group: Comparative and Global Properties, 3rd CTIO/ESO
Workshop, ed. A. Layden, R.C. Smith, & J. Storm (Garching: ESO), p. 187
This preprint was prepared with the AAS LATEX macros v4.0.
-- 9 --
Fig. 1. -- Normally incident ionizing photon fluxes φi from a Local Group corona, for core densities
n0 = (1, 3, 10) × 10−3 cm−3 (bottom to top), as a function of core radius ro. The solid lines are
for r = 350 kpc from the center of the corona, and the dashed lines are for r = 0. A metallicity
Z = 0.1Z⊙ has been assumed.
Fig. 2. -- Ionizing photon fluxes φi versus the mass cooling rate M , for a steady-state cooling flow.
A core radius ro = 50 kpc has been assumed. From top to bottom, the curves are for assumed
corona ages tLG = 1, 2, 4, and 8 Gy.
Fig. 3. -- Constraints on a Local Group corona in the (ro, no) plane. Coronae within the gray-
shaded regions produce ionizing photon fluxes between φi = 105 and 104 phot cm−2 s−1 (upper and
lower edges) at radii r = 0 (lower region) and r = 350 kpc (upper region) with respect to the LG
center. The long-dashed line is the COBE quadrupole constraint, the short-dashed line assumes
the LG medium is "typical", the solid line is the timing mass constraint, and the hatched region
satisfies ne ≤ 10−4 cm−3 at r = 350 kpc. See §4 for discussion.
-- 10 --
Fig. 1. --
-- 11 --
Fig. 2. --
-- 12 --
Fig. 3. --
|
astro-ph/0701608 | 2 | 0701 | 2007-07-12T13:11:17 | Could we identify hot Ocean-Planets with CoRoT, Kepler and Doppler velocimetry? | [
"astro-ph"
] | Planets less massive than about 10 MEarth are expected to have no massive H-He atmosphere and a cometary composition (50% rocks, 50% water, by mass) provided they formed beyond the snowline of protoplanetary disks. Due to inward migration, such planets could be found at any distance between their formation site and the star. If migration stops within the habitable zone, this will produce a new kind of planets, called Ocean-Planets. Ocean-planets typically consist in a silicate core, surrounded by a thick ice mantle, itself covered by a 100 km deep ocean. The existence of ocean-planets raises important astrobiological questions: Can life originate on such body, in the absence of continent and ocean-silicate interfaces? What would be the nature of the atmosphere and the geochemical cycles ?
In this work, we address the fate of Hot Ocean-Planets produced when migration ends at a closer distance. In this case the liquid/gas interface can disappear, and the hot H2O envelope is made of a supercritical fluid. Although we do not expect these bodies to harbor life, their detection and identification as water-rich planets would give us insight as to the abundance of hot and, by extrapolation, cool Ocean-Planets. | astro-ph | astro-ph | I09772: rev ised vers ion, accepted by Icarus
Could we identify hot Ocean-Planets with CoRoT, Kepler
and Doppler velocimetry?
--------
version 21 11 Jul. 07,
F. Selsis (1), B. Chazelas (2), P. Bordé(3,*), M. Ollivier (2), F. Brachet(2), M. Decaudin(2), F.
Bouchy(4), D. Ehrenreich(4), J.-M. Griessmeier(5), H. Lammer(6), C. Sotin (7), O. Grasset(7), C.
Moutou(8), P. Barge(8), M. Deleuil(8), D. Mawet(9), D. Despois (10), J. F. Kasting(11), A. Léger(2)
(1) Ecole Normale Supér ieure de Lyon, Cen tre de Recherche As tronomique de Lyon , 46 allée d'Italie,
F-69364 Lyon Cedex 07 , France ; CNRS, UMR 5574 ; Université de Lyon 1, Lyon , France.,
franck.selsis@ens-lyon .fr
(2) Institu t d ’As trophysique. Spa tia le, bat 121 , Université Paris-Sud and CNRS (UMR 8617); Un iv.
Paris-Sud, F-91405 Orsay; Fr, bruno [email protected]; Marc [email protected] .fr;
Frank.Brache t@ias .u-psud .fr ; miche l.decaudin@ias .u-psud.fr ; Alain [email protected]
(3) Harvard-Smithson ian Center for Astrophysics, 60 Garden Stree t, Cambridge , MA 02138 , USA,
[email protected] .edu
(*) Michelson Postdoctoral Fellow. Curren t address : Michelson Science Cen ter, Ca lifornia Institu te of
Technology, 770 S W ilson Avenue, MS 100-22, Pasadena, CA 91125 , USA
(4) Ins titut d’Astrophysique de Paris ; CNRS (UMR 7095) ; Université Pierre & Marie Cur ie ; 98, bis
boulevard Arago, F-75014 Paris, France ; bouchy@iap .fr; ehrenrei@iap .fr ;
(5) LESIA, CNRS-Observatoire de Paris , 92195 Meudon, France ,
jean-math [email protected]
(6) Space Research Institu te , Aus trian Academy of Sciences, Schmiedlstr . 6, A-8042, Graz , Austria ,
helmu t.lammer@oeaw .ac .at
(7) Géophysique , Un iversité de Nantes, F-44321 Nantes cedex 3 , Fr, sotin@chimie .univ-nantes .fr ;
Olivier .Grasset@chimie .univ-nantes .fr
(8) Labora toire d’Astrophysique de Marseille (LAM/OAMP), CNRS, – BP 8 - Traverse du Siphon ,
13376 Marseille Cedex 12 , C laire .Moutou@oamp .fr ; p [email protected]; magali.deleuil@oamp .fr
(9) université de Liège , 17 allée du 6 Août, 4000 Sart-Tilman , Belguim, dimitri.mawet@ulg .ac.be
(10) Observatoire de Bordeaux ( INSU/CNRS) , B.P. 89 , F-33270 Flo irac, Fr, D idier [email protected]
bordeaux1.fr
(11) Dept. of Geosciences , The Pennsylvania State University, University Park , Pennsylvania 16802,
USA, [email protected] .edu
ABSTRACT
Planets less massive than abou t 10 MEarth are expec ted to have no massive H-He atmosphere and a
cometary composition (~ 50% rocks, 50% wa ter, by mass) provided they formed beyond the
snowline o f protoplanetary disks. Due to inward migration , such planets cou ld be found a t any
distance between their forma tion site and the star . If migra tion s tops within the habitable zone, this
may produce a new kind of p lane ts, ca lled Ocean-Planets . Ocean-p lane ts typica lly consist in a
silicate core , surrounded by a thick ice mantle, itse lf covered by a 100 km – deep ocean . The
possible exis tence of ocean-plane ts raises impor tan t astrobiological questions: Can life origina te on
such body, in the absence of continen t and ocean-silicate inter faces? Wha t would be the nature of
the atmosphere and the geochemica l cycles ?
In th is work, we address the fa te of Ho t Ocean-Plane ts produced when migra tion ends at a closer
distance . In this case the liqu id/gas inter face can disappear, and the hot H2O envelope is made o f a
supercritical fluid . Although we do no t expect these bodies to harbor life, the ir de tection and
identifica tion as water-rich planets wou ld give us insight as to the abundance o f hot and, by
extrapola tion, cool Ocean-Planets .
The wa ter reservoir of these planets seems to be weakly affected by gravita tiona l escape , provided
that they are loca ted beyond some minimum dis tance, e .g. 0 .04 AU for a 5-Earth-mass plane t
around a Sun-like star . The swelling o f their wa ter atmospheres by the h igh stellar flux is expected
not to sign ificantly increase the planets’ rad ii. We have studied the possibility o f detec ting and
characterizing these Hot Ocean-Planets by measur ing their mean densities using transit missions in
space – CoRoT (CNES) and Kepler (NASA) – in combina tion with Dopp ler velocime try from the
ground – HARPS (ESO) and possible future ins truments. We have de termined the domain in the
[stellar magnitude , orbital dis tance ] plane where discr imination be tween Ocean-Plane ts and rocky
plane ts is possible with these instrumen ts.
The brigh test s tars of the mission target lists and the plane ts closest to their stars are the most
favorable cases. Full advan tage o f h igh precision photometry by CoRoT, and par ticularly Kepler,
can be ob tained on ly if a new genera tion of Doppler ins truments is built.
JID:YICARAID:8261/FLA[m5+;v1.73;Prn:12/06/2007;8:22]P.2(1-16)2F.Selsisetal./Icarus•••(••••)•••–•••1.Introduction1.1.Water-richvssilicate-richplanetsOcean-planets(OPs)representanewkindofplanets,re-centlyproposedbyLégeretal.(2004,2003)andKuchner(2003),whichnaturecanbedescribedasfollows:(1)Theirformationtakesplacebeyondthesnowline,giv-ingthemacometary-likebulkcomposition:silicatesandwater,inroughlyequalamountsbymass.Planetsthataccretedplanetesimalsformedatlessthanabout3AUfromasolartypestarhaveamuchlowerwatercontent(w≡H2O/silicate,bymass):wis5×10−4fortheEarthandremainsbelow10%foraplanetentirelymadeofcarbonaceous-chondriticmaterial.Raymondetal.(2006b,2007)simulatedtheformationofhabitableplanetsintheabsenceofmigrationandfoundw<5%attheendoftheaccretion.Bysimulatingplanetaryformationafterthemi-grationofanoutergiantplanettoaclose-inorbit,Raymondetal.(2006a)predictedthatthemixingofmaterialfromtheinnerandouterregionscanproduceplanetsinthehabit-ablezoneconsistingofupto∼30%water.ThismightbeanotherwaytoformOPs.(2)UnlikeUranusandNeptune,OPsdonotreachasufficientmasstoaccreteagaseousenvelopedirectlyfromthepro-toplanetarydisk.Theyarethuslessmassivethanabout10MEarthandassumedtohavenoH2–Heenvelope.Thisas-sumptionisrobustforplanetsof6MEarthandbelow,andhastobediscussedfurtherforplanetsintherange6–10MEarthforwhichsignificantgasaccretionmaystillhaveoccurred(Alibertetal.,2006;Rafikov,2006).(3)TobecomeanOP,therocky–icyplanethastomigratetotheinnerpartoftheplanetarysystemthroughtype-Imigration(Papaloizouetal.,2006).Ifmigrationstopsinthehabit-ablezone(HZ)ofitsstar,theplanetbecomeswhatLégeretal.(2004,2003)callanocean-planet(OP).Whenmigra-tionstopsatshorterorbitaldistances,itcangivebirthtoplanetswithathickandhotH2Oenvelopewithnoliquid–atmosphereinterface.(4)PlanetsofafewMEarthareexpectedtobemainlymadeofsilicateandwater,exceptwhenfoundaroundstarswithaC/Oratiosignificantlyhigherthansolar,wherecarbon-planetsmayform(KuchnerandSeager,2005).Althoughoneshouldkeepinmindthispossibilityforsomeplane-tarysystems,weconsiderinthispaperthatplanetscanbecharacterizedwithasinglenumber:thewatercontentw.Inthiscase,internalstructuremodelsprovidedifferentplan-etaryradii,foragivenmass,asafunctionofthiswaterratio.Fora6-Earth-massplanet,Légeretal.(2004,2003)foundRpl=2.00REarthforanOPandRpl=1.63REarthforarockyone.Sotinetal.(2007)generalizedthesecalcula-tionsformassesinthe(0.01–10MEarth)rangeandvariablewatercontent(Fig.1).Independently,Valenciaetal.(2006)alsoderivedmass–radiuscurvesforplanetswithvariouswatercontents.Thesetwoworksareinverygoodagree-Fig.1.Relationbetweenthemassandtheradiusofocean-planets(OPs)(m50%rocks,50%water)uppercurve,androckyplanets(m100%rocks)lowercurve,asobtainedfrommodelsoftheirinternalstructurebySotinetal.(2007)whenplanetsarelocatedinthehabitablezoneoftheirstar,orfurther.Foragivenmass,themeandensitiesandtheradiioftheplanetsaresignificantlydifferentaccordingtotheirnature.Planetarymassandradiusareaccessibletoobserva-tionalmeasurementsbyDopplervelocimetryandplanetarytransitphotometry,respectively.Thesemeasurementshavethepotentialcapabilityofdiscrimi-natingbetweenthetwotypesofplanets.ItshouldbenotedthatsomeoftheGalileansatellitesofJupiterareexamplesofOPs,butfortheirlowermasses.ment.Bothfoundthat,inallcases,acoupleofvalues(Mpl,Rpl)areindicativeofaplanetarycomposition.1.2.Onthevergeofdetectingocean-planets?Thankstoremarkableprogress,radialvelocity(RV)mea-surementsarenowunveilingapopulationofplanetswithamassbetween10and20MEarththatarefoundatshortorbitaldistancesandasfarasthehabitablezoneoftheirstar.ThisiswellillustratedbythetriplesystemHD69830a,b,c(Lovisetal.,2006).Modelsreproducingthissysteminvoketheformationoficy–rockycoresbeyondthesnowline,followedbyinwardmi-gration(Alibertetal.,2006;TerquemandPapaloizou,2006).Becauseoftheirhighermass,thesehotNeptunesdifferfromOPsbytheirH–HeenvelopeofseveralMEarth.Coresabove10MEarthareindeedexpectedtoaccreteasignificantfractionoftheirmassashydrogen-richgas.Fortworeasons,thesediscoveriesareextremelypromisingforthesearchofOPs:First,thesamemechanismthatcom-monlyproduceshotNeptunesshouldalsogenerateOPsforlessmassivemigratingrocky–icycores.Second,theincreasedaccu-racyinRVmeasurementsshouldsoonallowustodetectplanetsbelow10MEarth,intherangeofmassesofOPs.TheCoRoTmission(http://CoRoT.oamp.fr;Rouanetal.,1999)wassuccessfullylaunchedbytheendof2006andKeplerisscheduledforlaunchin2008(http://kepler.nasa.gov;Kochetal.,2006).Theirtransitsearchprogramswilldeterminetheradiiofdiscoveredplanets.TheDopplerfollow-up,whenever JID:YICARAID:8261/FLA[m5+;v1.73;Prn:12/06/2007;8:22]P.3(1-16)Identificationofhotocean-planets3Fig.2.ProbabilityofobservinganexistingplanetaroundatargetstarbyCoRoT,asafunctionofitsorbitalperiod.Assumingarandomorientationofplanetaryorbits,thegeometricalprobabilitythatatransitisobservedfromtheEarthispg=Rst/a,theratioofthestellarradiustothatoftheplanetdistancetoitsstar,inthelimitRpl%Rst.UsingtheKeplerianrelationbetweenaandtheplanetaryperiodTorb(foragivenstellarmass,T2orb/a3=const)andthetimeprobabilitythat!3eventsoccurwithina150dayduration,p3,oneob-tainstheprobabilitythatagivenplanetaroundaGVstarisdetected.Thisisarapidlydecreasingfunctionoftheplanetaryperiod,whichpointsoutthatplanetdetectionisbiasedtowardsshortperiodobjects.possible,willdeterminetheirmasses.Foreachdetectedplanet,theseobservationswilldetermineapointanditsboxerrorinthe(Mpl,Rpl)plane.ModelsoftheirinternalstructurepredictRpl(Mpl)curvesinthatplane,whichvarywiththeplanetarycomposition(Fig.1).Assumingthatthemodelsarecorrect,therelativepositionsoftheerrorboxandthecurveswilllead,ornot,tothecharacterizationofthenatureoftheplanet.ThisapproachwassuccessfullyappliedtotransitinghotJupiters,amongwhichwecandistinguishacore-dominatedplanet(HD149026b;Satoetal.,2005)fromtheothergas-dominatedplan-ets(forinstanceTrEs-1;Alonsoetal.,2004).Inreturn,theobservationshaveputseriousconstraintsonthemodelsandthestructureofhotJupiters,orPegasides.Forinstance,thelargeradiusofHD209458bisnotyetwellexplained.1.3.PlanetorbitalperiodsaccessibletoCoRoTandKeplerCoRoTwasinitiatedasaCNES“smallmission.”Itislo-catedinalow-altitudeEarthorbit(∼900km).Asitmustnotpointtoofarfromtheanti-solardirection,itcanobserveagivenstellarfieldcontinuouslyonlyduring5months(150days).Itisconsideredthatthedetectionofaplanetarycandidatebythetransitmethodrequirestheobservationof3transitsormore.Thus,onlyplanetswithperiodP<75dayscanbedetected(Fig.2).Forasolar-typestar,thiscorrespondstoadistancetoitsstar,a,lessthan0.35AU(circularorbit)andablackbodytemperature(albedo=0,nogreenhouseeffect)Tbb>460K.PlanetswithP<75dayscanbehabitableiftheirhost-starislessmassivethanaK5star(M∼0.7MSun).However,theselow-massstarsrepresentonly1.5%ofthestarsaccessiblewithCoRoT.Withlessthan200oftheminitsfieldandatransitprobabilityof∼1%,CoRoTisunlikelytodetectthetransitofahabitableplanet.Statistically,theprobabilitythat3transitsormoreofagivenplanetaredetectedwithin150daysisarapidlydecreasingfunc-tionofitsorbitalperiod(Fig.2).Ifplanetswereuniformlydistributedindistancesaroundtheirstars,andifCoRoThadthesensitivitytodetectallofthem,thehistogramofdetectionswouldbeproportionaltothisprobability,stressingthestrongbiastowardsshortperiods.Inaddition,thelargerthenumberoftransitsthelargerthedetectionS/Nforsmall-sizeplanets.Asaconsequence,itisimportanttoaddressthefateofOPswhentheyareclosetotheirstars.ThesituationfortheKeplermissionismorefavorable(http://kepler.nasa.gov;Kochetal.,2006).Itisalargermis-sionthatwilloperateonanEarth-trailingheliocentricorbit,continuouslyfor4years.Itcandetectplanetswithperiod"1.33yraroundsolar-typestars,correspondingtosemimajoraxisa"1.21AUandTbb!250K,whichincludesEarth-likeplanets,themaingoalofthemission.However,evenforKepler,thedetectionofinnerplanetswillbeeasierandmoreaccurate.AccordingtoBordéetal.(2003),CoRoTcoulddetectsev-eraltensofplanetswithRpl∼2REarthanda×(L/LSun)1/2<0.35AU,whereListhestellarluminosity,ifeachstarhasoneplanetwithintheserangesofsizeandlocation.Thishypothesisisarbitrary,butitindicatesthattheactualnumberofdetectionscouldtellustheabundanceoftheseplanetsifthemissionca-pacitiesareasexpected.ForKepler,theprospectsarehigherandstartfromplanetswitha"1.21AUforSun-likestars.1.4.OutlineofthisstudyDuetotheobservationmethodsdescribedabove,exoplanetswiththeshortestperiodsaretheeasiesttocharacterizebytheirmassandradius.Ontheotherhand,andasalreadydiscussedbyKuchner(2003),thestrongstellarirradiationcanaffecttheevo-lutionofthewaterreservoirofclose-inplanets(byatmosphericescape)andoftheirradius.Insummary,OPsthathavekeptenoughwatertobedistinguishedfromsilicate-dominatedplan-etsmaybefoundonlyatorbitalperiodthatwillnotbeavailablesoontoaccuratemassandradiusmeasurements.Thisquestionprovidesthegeneraloutlineofourpaper.InSection2,weaddressthefateofwater-richplanetsthathavemigratedveryclosetotheirstar,theirswellingduetothestrongexternalheating,and,especially,thelossoftheirwa-terthroughthermalandnon-thermalevaporationinducedbystellarXandEUVradiation,windandcoronalmassejections(CMEs).InSection3,weestimatetheaccuracyoftheradiusandmassdetermination,asafunctionoftheorbitalperiodandstellarproperties,thatcanbeobtainedwithCoRoT,Kepler,andRVfollowup.Then,onthebasisofthisaccuracyandofthetheo-reticalRpl(Mpl,w)curves,weevaluatethecapabilitytodistin-guishawater-richfromasilicate-richpopulationofplanet.InSection4,wediscusswhattherealisticdiversityof1–10MEarthshort-periodplanetscouldbe.Inparticularweaddressthepossibledegeneracyof(Rpl,Mpl)couplesofvaluesduetotheexistenceofanaccretedatmosphereofH2–He,andtothe JID:YICARAID:8261/FLA[m5+;v1.73;Prn:12/06/2007;8:22]P.4(1-16)4F.Selsisetal./Icarus•••(••••)•••–•••existenceofplanetswithadifferentbulkcompositionsuchascarbon-planets(KuchnerandSeager,2005).OurconclusionsarepresentedinSection5.2.Structureandfateofashortperiodocean-planetOPsaredefinedasplanetsthatformfartherawaythanthesnowlineandmigrateinwards.Whentheorbitaldistancesta-bilizeswithinthehabitablezone,andassumingthattheplanethascooledtothethermalequilibriumwiththestellarirradi-ation,aglobalandthickoceanwouldcondenseaboveanicymantle(Légeretal.,2004).Forshorterorbitaldistancesnogas/liquidinterfacecanbesustained.Assumingthatwateristhemainconstituentoftheatmosphere,thishappenswhenthereisnomorephasetransitionbetweenthevaporphaseandtheliquidone,i.e.whenthetemperaturedistributionwithintheplanetT(P)runsabovethecriticalpointofwater(Tc=647K,Pc=22MPa∼220bars)asillustratedinFig.3a.Whenanoceanispresent,theT(P)curvecrossesthegas–liquidtransi-tionline,whichdeterminestheoceansurfacelocation(Fig.3b).Intheabsenceofliquid–gasinterface,thewholefluidlayercanbecalledenvelopeasingiantplanets.Thewholewaterenve-lopeissupercritical.Thesehotplanetscouldbenamed“planetswithsupercriticalwaterenvelope”or“sauna-planets.”Forthesakeofsimplicity,weshallkeepthenameofhotocean-planets.ThepossibilitythattransitmissionsdetectandcharacterizesomeOPsdependsontheirexistence/abundanceandthein-strumentcapabilitiestodetectsmallplanets,saywithRpl∼2REarth,atsmalldistancesfromtheirstars,asdiscussedinSec-tion3.Inthepresentsection,wediscusswhethertheRpl(Mpl)relationestablishedbySotinetal.(2007)(Fig.1)intheHZismodifiedbyaswellingoftheplanetduetostrongerirradiation(Section2.1)andhowclosetotheirstarOPscanbewithoutlosingmostoftheirwater(Section2.2).2.1.Estimatedswellingofanocean-planetclosetoitsstarSotinetal.(2007)andValenciaetal.(2006)modeledtheinternalstructureofOPsandestablishedtheRpl(Mpl)relationforOPsintheHZ.Whentheseplanetsareclosertotheirstars,thehigherstellarfluxinducesahighertemperatureattheouterboundary,whichmayresultinanenhancedradius.Inordertotestthesensitivityoftheradiustothisouterboundarytem-perature,Sotinetal.(2007)havecalculatedthat,inthecaseofa5MEarthOP,anincreaseof1000Kresultsina0.9%in-creaseoftheradius.However,thiscomputationwasdoneonlyforthecondensed(solidandliquid)partoftheplanet.Amoresignificantswellingisexpectedfromtheexpansionofthehotwatervaporatmosphere.InsidetheHZ,theatmosphericwatervaporcontentisafunctionoftheorbitaldistance:aroundthepresentSun,intheabsenceofothergreenhousegasesandas-sumingalargeoceanatthesurface,thepartialpressureofH2Owouldbe1barat0.93AUandwouldreachthecriticalpressure(220bar)at0.84AU(Kasting,1988).Atcloserdistances,theouterpartoftheplanetisathicksupercriticalwaterenvelope,or“steam-ocean,”boundedbythehigh-pressureicemantle,orbythesilicatemantleifthetemperatureimposesthatallthewa-terisfluid.Thedepthofanhot“steam-ocean”wouldbelargerthanthedepthofliquid–solidwatermantleofthesamemass,whilethesubcriticalatmosphereaboveP=220barexpandsquasi-linearlywithitstemperatureandmightthusresultinaswelling.Letusfirstconsidertheswellingduetotheexpansionoftheopaqueatmosphere.ThepressureintheatmospheredependsFig.3.WaterphasediagramandqualitativeT(P)profilesfortheouterlayersofawater-richplanet.Planet(a)isinthehabitablezone(HZ)ofitsstarandhasanocean–atmosphereinterface.Planet(b)isatanorbitaldistanceslightlyshorterthantheinneredgeoftheHZandplanet(c)hasaveryhotequilibriumtemperature(>500K).Planets(b)and(c)havenoocean–atmosphereinterfacebutathickfluidenvelopeofsupercriticalH2O. JID:YICARAID:8261/FLA[m5+;v1.73;Prn:12/06/2007;8:22]P.5(1-16)Identificationofhotocean-planets5uponthealtitude,z,as(1)P=P0e−z/H,H=CkT/µg,whereHistheatmosphericheightscale,P0thecriticalpres-sureatthelowerboundary,ktheBoltzmannconstant,µthemeanmolecularweight,gthelocalgravity,andCisthecom-pressibilityfactor,whichdependsonthepressureandtempera-ture,C=1foraperfectgasandC<1foradensegas.Theradiusdeterminedbyatransitobservationisthedis-tancez,totheplanetarycenter,ofalightraygrazingtheplanetaryatmospherewithanopticalthicknessofunityattherelevantwavelength.Wavelength-dependentradiioftransitingEarth-sizeplanetshavebeencalculatedbyEhrenreichetal.(2006).Theypresenttransmittedspectraoftransitingocean-planetswithorwithoutclouds.Forahotocean-planet,watercloudsarenotexpectedbecausethegasisatatemperature!1000K(seebelow),whichisabovethecriticalpointofwaterandpreventsthecondensationofliquiddroplets.Assumingthattherearenootheraerosolsorgrains(Feorsilicates),intheup-peratmosphere,Rayleighscatteringisexpectedtobethemaincontributortothetotalextinction.Anapproximationofthecol-umndensityNacrosstheatmosphericlimbalongthelineofsightisN=n(2πRplH)1/2,wheren=P/(kT)isthevolumicdensity(Fortney,2005).ThecorrespondingopticaldepthhasbeencalculatedbyEhrenreichetal.(2006)andcanexpressedas(2)τ(λ)=√2πP(z)kTR1/2plH1/2σRayl(l),whereσRaylistheRayleighcross-sectionofthegasmolecules(herewater).Requiringthatτ=1at0.6µm,acentralwave-lengthofthephotometricmeasurements,impliesavalueforPandthenzfromEq.(1).Normalizingtoareferencecase,usingσRayl(H2O,0.6µm)=2.32×10−27cm2,onereads:(3)P=0.38¯T1/2¯M1/2¯R−3/2bar,with¯T=T/103K,¯M=Mpl/6MEarth,¯R=Rpl/6REarth.Forthereferencecase,onefindsz−z0=13kmwherezandz0arethealtitudeswhereτ=1at0.6µm(P=0.38bar),andP0=1bar,respectively.Theeffectivelengthleffofthelighttravelintheatmosphereisthen1000km,correspondingtoasustainingangle2θ=4.3◦.Forcomparison,foratransitingEarth,usingσRayl(air,0.6µm)=3.16×10−27cm2andne-glectingtheabsorptionsduetoO3andtheaerosols,onegets:P=0.24barandz−z0=11.4km,leff=640km,2θ=5.8◦.Thesevalueshavetobecomparedwiththemuchhigherattenu-ationoftheSunatsunsetwhenthevalueofPis1barandonlyhalfofthegrazingtravelisperformedbythelight.InthefollowingestimatesoftheswellingofahotOP,weconsiderthattheplanetaryradiusdeterminedbyatransitobser-vationcorrespondstoP∼0.4bar,andwedeterminetheheightoftheatmosphericlayerboundedbythe0.4and220barlevels.AttheorbitaldistanceofahotJupiter(0.05AU),thetem-peratureofthe0.4barlevelofanOPwouldbeclosetoitsequilibriumtemperature,thatis1100K.InahotJupiterofwithasimilarequilibriumtemperature,thetransitionbetweentheouterradiativelayerandtheconvectiveregionoccursintheH2–Heenvelopeatpressuresandtemperaturesoftheorderof500–1000bar,2000–3000K(Barmanetal.,2005).Althoughwewouldneedaself-consistentmodeltocomputetheT–Pprofileofawaterenvelope,asafunctionoftheorbitaldistanceandtakingintoaccountthepreciseopacitiesofH2O,wecanassumethattheatmospherictemperatureincreasesfromabout1100Kintheupperandoptically-thinlayers,totypically2000–2500Katthetransitionbetweentheradiativeandconvectiveregion,thatwearbitrarilyfixat200barshere.Byestimatingtheheightofa1000and2500Kisothermalatmosphereathy-drostaticequilibrium,wecanbrackettheheightoftheradiativelayerthatisopaquetothetransitobserver.Fora6MEarthplanetandthesetwoatmospherictemperatures,the0.4barlevelisreached,respectively,at260and700kmabovethe220barlevel.Theseestimatestakeintoaccountthenon-idealproper-tiesofthehighpressurewatervapor(NISTsteamtablesforthecompressibilityfactor),butonecannotethatusingtheperfectgasapproximationyieldssimilarvalues(210and700km).Theintrinsicswellingoftheconvectiveenvelopeunderneaththeradiativeatmosphereislowerbutitaffectsalargefractionoftheradius.Theproblemtoestimatethedepthofthislayerasafunctionofthetemperatureoftheouterlayersisdouble.Firstwearerapidlylimitedbythelackofavailabledatafortheequa-tionofstateofwaterathightemperatureandpressure.Then,asthecoolingoftheplanetsisdeterminedbythegradientoftem-peratureintheouterlayers,ahighequilibriumtemperatureoftheplanetcanleadtoasignificantlydecreasedheatfluxandveryhotinterior.Addressingthisissuewouldrequireaconsis-tentformationmodelprovidinginitialthermalconditionsandanevolutioncode.Verypreliminaryestimateusingextrapola-tionsofavailabledatashowthatthedepthoftheconvectiveenvelopeofanhot6MEarthcanincreaseby∼300–500km.Comparetothe2REarth(12,800km)radiusitwouldhaveintheHZ,ahot6MEarthOP(at0.05UAfromitsstar)wouldexhibitaradiusincreasedby5–10%:+200–700kmfortheatmosphere,+300–500kmforthefluidwaterenvelope,+<100kmforthesilicate+metalinterior.Forsmallerplan-etsandplanetsstillclosertotheirstars,theswellingcanbesomewhatlarger.For1MEarth,at0.05AU,theswellingshouldbe>10%.Rockyplanetscouldalsoexperiencesuchaswellingoftheiratmosphere,inprinciplemakingthemdifficulttodistin-guishfromOPs.However,attheorbitaldistanceswheretheswellingcouldbesignificant,silicate-dominatedplanetswouldloserapidlytheirvolatilecontentsthroughthermalandnon-thermalatmosphericlossprocesses,whileOPshaveanalmostinexhaustiblereservoirofwater(seebelow).Hence,whenanOPisclosetoitsstar,itmayundergosomeswellingthatwouldhelptodiscriminateitfromarockyplanet.However,astheseestimatesfortheswellingareveryprelimi-naryandconcernonlyvery-short-periodplanetswewillassumeconservativelythat,ifitsurvives,anOPwillhavearadiussim-ilartowhatitwouldhaveifitwerelocatedwithintheHZ.Duringtheplanetformationprocess,somehydrogenmaybeaccreted,andwaterislikelytobeincontactwithreducingFe2+ JID:YICARAID:8261/FLA[m5+;v1.73;Prn:12/06/2007;8:22]P.6(1-16)6F.Selsisetal./Icarus•••(••••)•••–•••ions,possiblyevenmetalliciron;hence,onecanexpectachem-icalreactionwithahydrogenoutput.ThepresenceofH2wouldimplyalargerswellingoftheplanet.However,theatmosphericerosionprocessesthatareconsideredinthenextsectionsaremuchmoreefficientforhydrogenthanfortheothergasesandwillrapidlytakehydrogenawayfromtheatmosphere,espe-ciallyattheorbitaldistancesrelevantforCoRoT.Thereafter,weconsiderthatOP’satmospheresaremainlyfreeofhydrogen,buttheactualhydrogencontentofOPsmeritsfurtherinvestiga-tions.2.2.Evaporationofshortperiodocean-planetsOcean-planetsclosetotheirstarsmaylosetheirwatertospaceandbecomerockyplanets.Theaimofthepresentsec-tionistoestimatehowclosetotheirstarswecanexpectOPstokeeptheirwaterandpossiblybeidentifiedbyCoRoTandKepler.TheprecisemodelingoftheerosionofanOPiscom-plex,andthecorrespondingworkisstilltobedone.Wecan,however,estimatetheupperlimitonatmosphericlossesbasedonmaximumenergydepositionintheupperatmosphere.Usingenergy-limitedescape,whichislikelytobesignificantlylargerthanactuallossrates,wecandeterminetheminimumperiod(ororbitaldistance)atwhichmostofthewaterreservoirshouldsurviveduringagiventime,forinstance5Gyrs.DeterminingthisminimumorbitaldistanceisimportantinthecontextofCoRotsearchesthatwillfocusonclose-inplanets.Note,how-ever,thatplanetsfoundatshorterperiodscouldstillhavekeptmostoftheirwater,dependingonwhetheractuallossesarelim-itedbymechanismsotherthanenergydeposition.Themainprocessesforatmosphericescapethatcanleadeventuallytotheexhaustoftheplanetarywaterreservoirare:(1)thermalescapedrivenbyexosphereheatingbyX-raysandExtremeUV(orXUV:0.1–100nm),(2)non-thermalescapedrivenbytheactionofparticlesfromthestar.2.2.1.WatererosionbythermalescapeTheheatingoftheupperatmospherebyXUVphotonscanresultintheescapeofgasestospace.EscapeaffectsmainlylightspecieslikeHandHe,butheavierspecieslikeOcanalsobecarriedawaybythehydrodynamicflowwhenitishighenough(Chassefiére,1996):suchaneffecthasbeenobservedinthecaseofthehotJupiterHD209458b(at0.05AUfromitsstar)forwhichnotonlyH,butalsoCandO,havebeende-tectedintheescapingupperatmosphere(Vidal-Madjaretal.,2003,2004).Anupperlimittothemasslossisgivenbytheenergy-limitedescaperate(Lammeretal.,2003),whichwouldbereachedifalltheenergyabsorbedatλ<100nmwouldbeconvertedintogravitationalenergy,asexpressedinthefollow-ingequation(4)εFXUVπR2pl(a/1AU)2=GMplmH2ORpl.ThelefttermisthefractionoftheXUVfluxthatisinterceptedbytheplanetandavailablefordrivingthethermalescape,andtherighttermisthevariationofgravitationalenergyin-ducedbythemass-lossofwater.FXUVistheXUVenergyflux[Wattm−2]receivedat1AU,M,Randaarethemass,radiusandorbitaldistanceoftheplanet,respectively;thefactorεcor-respondstotheefficiencyoftheconversionofincidentXUVenergyintoeffectiveescapeofgas,andcontainsallthephys-icalcomplexityoftheescapeprocess.Themassofwaterlostoveratimetis(5)mH2O=επR3plGMpl(a/1AU)2t!0FXUV(t()dt(.Wecanconsiderεastheproductofaheatingefficiencyεh,whichgivesthefractionoftheincidentenergythatisnotre-radiatedtospace,andanescapeefficiencyεesc,whichisthefractionofthedepositedenergythatiseventuallylostthroughescapinggas.Theheatingefficiency,εh,istypicallylowerthan0.2,especiallyathighXUVirradiation.Yelle(2004)usedadetailedphotochemicalmodelandfoundavalueof∼0.1forHD209458b.Here,weuseavalueof0.2inordertoobtainanupperlimitonthelossrate.Intheprecedingrelations,RplissometimesreplacedbytheradiusatwhichtheXUVradiationisabsorbedRXUV(e.g.Lammeretal.,2003).ThiscomesfromWatsonetal.(1981)hydrodynamicmodelingoftheescapeinwhichalltheincom-ingXUVradiationisassumedtobeabsorbedinaninfinitelythinlayeratRXUV.Inthisapproach,RXUVcanbeaslargeasseveraltimesRpl,forterrestrialplanets.Thisexpansionoftheupperlayercanimplyextremelyhighescaperates.Recenthydrodynamicmodels(steadystate:Tianetal.,2005;timede-pendent:Penzetal.,2006)includingarealisticdepositionoftheXUValongtheescapingflow,showthatthethinabsorb-inglayerapproachoverestimatestheescaperatebyordersofmagnitude,especiallyforhighincomingXUVflux.OneofthereasonsforthisoverestimationistheXUVself-shielding:someXUVphotonsthatareabsorbedwithintheescapingoutflowcontributetoheatingtheoutflowbutnottoenhancingthelossrate.WhentheXUVdepositionisjusthighenoughtodrivehy-drodynamicescape,theescapingoutflowisnotdenseenoughtoabsorbasignificantfractionoftheincomingXUVenergy,whichismostlydepositeddeeperwherepartofitiseventuallyconvertedintogravitationalenergy.Inthisphase,theescaperateincreasesnearlylinearlywiththeXUVflux,butwithin-creasingirradiation,escapereachesapointatwhichthecolumndensityofescapingatomscannolongerbeconsideredtranspar-enttoXUV:theoutflowbecomeshotterbuttheefficiencyεescdecreases.TheescapeefficiencywasestimatedbyPenzetal.(2006)bymodelingthehydrodynamicflow.ForHD209458btheyfoundanescapeefficiencyofabout0.5at0.045AUfromthepresentSun(conditionsassumedforHD209458b)thatde-creasesrapidlyforincreasingXUVfluxes(εesc=0.15iftheXUVfluxismultipliedby2).Therefore,thetotalefficiencyε=εhεescvariesfrom0.02to0.1intherangeoforbitaldis-tancesthatareobservablebyCoRoT.Fig.4givesthetimerequiredtolosethewholewaterreser-voirofanOPthroughthisenergy-limitedmass-lossasafunc-tionoftheorbitaldistancea(solidlines).ThisestimatetakesintoaccounttheevolutionoftheXUVstellarluminosityac- JID:YICARAID:8261/FLA[m5+;v1.73;Prn:12/06/2007;8:22]P.7(1-16)Identificationofhotocean-planets7Fig.4.Minimumlifetimeofthewatercontentofa10,5and1Earth-massOPasafunctionofitsorbitaldistance.Solidlinesgivethelifetimeofthewaterreservoirerodedbythermalescapeatitsenergy-limitedrate(determinedbythedepositionofstellarXUVradiation).Dashedlinesgivetheenergy-limitedlifetimefornon-thermalescapeinducedbythestellarwind(determinedherebythedepositionofprotons’kineticenergy).SeeSection2.2fortheassumptionsontheefficiencyofthethermalescape,onstellarXUVandwindproperties.Arounda5Gyroldstar,OPswouldbeweaklyaffectedatorbitaldistances!0.03AU(10MEarth),0.04AU(5MEarth)and0.15AU(1MEarth).cordingtoRibasetal.(2005)andthevariationofεwiththeXUVfluxfromPenzetal.(2006).ForOPswithamassbelow6Earthmasses,thelossislimitedbytheXUVenergydeposi-tion.Fora5Earth-massOP,Fig.4showsthatthethermallossofthewaterreservoir(HandO)wouldtakemorethan5Gyrsata!0.04AU.2.2.2.WatererosionbythestellarwindandcoronalmassejectionsNow,thequestioniswhethertheerosionbythestellarwindcansignificantlychangethelowerlimitofthewaterreservoirlifetime.Thisprocessisefficientwhentheupperatmosphereofaplanetisnotprotectedbyanintrinsicmagneticfield,whichisstrongenoughtodeflectthestellarplasmaflowatlargedis-tancesabovetheexobase.OPsclosebytheirstarsareexpectedtohavetheirspinrota-tiontidallylockedontheirorbitalrotation.Fromsimplethe-oreticalmodels,severalanalyticalscalinglawswerederivedwhichallowonetoestimatetheplanetarymagneticdipolemo-mentfromtheplanetarycharacteristics(e.g.density,rotationrate,sizeofthedynamoregion).ThesemodelsaresummarizedandcomparedinGriessmeieretal.(2004,2005).Accordingtothesemodels,theexpectedmagneticmomentsofslowlyrotat-ingexoplanets,e.g.tidallylockedplanets,aresmallerthanthatoffastrotatingplanetliketheEarth.Byapplyingthesescal-ingrelationstoanOPof6.0Earthmassesand2.0Earthradii(Légeretal.,2004),whichistidallylockedat0.2AUaroundastarwith0.2solarmasses,weobtainamaximumintrinsicmag-neticmomentof0.7timesthatofthepresentEarth.Becauseoftherelativelyweakmagneticmoment(consideringthelargeplanetaryradius),theplanetarymagnetospheremaynotbeeffi-cientatprotectingtheplanetaryatmosphere.InsuchaVenus-likeinteraction(e.g.Teradaetal.,2002)theplanetaryneutralgasintheupperatmospherecanbeionizedbyelectronimpact,X-raysandExtremeUVradiations.Af-terchargeexchangeduringcollisionswiththeincidentstellarplasma,itispickedupbythisstellarplasmaflowandescapesfromtheplanet.Inthetwonextsectionsweestimatethenon-thermalat-mosphericlossduetotheinteractionwiththemeanstellarwindandwiththesporadiccoronalmassejections(CMEs),inte-gratedoverthewholehistoryofthesystem.StellarwindAsanupperlimittotheerosionduetothestel-larwind,wecalculatedthefluxofprotonsinterceptedbytheplanetintheabsenceofmagneticdeflectionandthecorrespond-ingdepositionofkineticenergyintotheupperatmosphere.Byanalogywiththeenergy-limitedthermallossratedescribedinSection2.2.1,wecaninferanenergy-limitedescapeinducedbythestellarwindprotons.Themaindifficultywiththisap-proachisthepoorlyknownevolutionofthestellarwind.Usingindirectobservations,Woodetal.(2002,2005)showedthatthemasslossofSun-likestarsiscorrelatedwiththeirX-raylu-minosityforagesabove0.7Gyr.Youngerstarsdonotseemtoexhibitahighermassloss.Inordertomakeanupperesti-mateofthepossiblelosses,weusedconstantwindpropertiescorrespondingthehighestmasslossmeasuredbyWoodetal.(2005).Thisallowsustooverestimatetheenergydeposition.Animprovementofthisassumptionwouldrequiremoredataonstellarwinds,especiallyonthoseofyoungstars.DashedlinesonFig.4showtheresultwefindfortheminimumlifetimeof JID:YICARAID:8261/FLA[m5+;v1.73;Prn:12/06/2007;8:22]P.8(1-16)8F.Selsisetal./Icarus•••(••••)•••–•••thewaterreservoir.Atadistancea!0.04AU,a5Earth-massOPwouldkeepitswaterformorethan5Gyrsifstellarwinderosionisconsideredalone.Thisisanupperlimitforthenon-thermallossthroughmean-stellar-windprotons,buttheeffectofcoronalmassejections(CMEs)fromthestarhasalsotobeaddressed.CoronalmassejectionsByusingtheparametersestimatedbyKhodachenkoetal.(2007)oneobtainsat0.05AU,anaver-ageCMEplasmavelocityof∼500kms−1andplasma(proton)densities"5×104cm−3.Lammeretal.(2006)studiedmaximumratesofO+ionpickupbyCMEformagnetizedandnon-magnetizedEarth-likeexoplanets.Theycalculatedtheplasmaflowaroundthemagne-topause/ionopauseforXUVfluxvalues10–100timeshigherthanthatofthepresentSun,asexpectedforayoungstar(Ribasetal.,2005).ThetotalionpickuplossratewasobtainedbycalculatingtheO+productionratecausedbychargeexchange,electronimpactandphotonsalongthestreamlinesintheplasmaflowaroundtheplanetaryobstacle.TheyfoundthatforstrongCMEsandfortheestimatedmin-imumplanetarymagneticmoment,thesituationwouldresultinaVenus-likeplasmainteraction(e.g.Teradaetal.,2002).PuttinginthetypicalnumbersfortheCMEprotondensityandtheaveragewindspeedmentionedabove,wefindthat,duringthecriticalfirst0.5GyrforanOPat0.05AU,thewaterlossescouldbeupto"200Earthoceans(bymass),whichissignifi-cantlylessthanthelossesbystellarwindsandcanbeneglected.Oneshouldnotethattheseestimatesarealsoupperlim-itsbecausethemoremassiveOPconsideredinourstudyhasahighergravitationalpotentialthantheEarth-massplanetsstudiedbytheprecedingauthors,whichwillholdbackthethermosphere–exosphereenvironmentexpansion,resultinginlowerlossrates.Furthermore,theevaporatinghydrogenwillformacoronaaroundtheplanetthatalsodecreasesthenon-thermallossofoxygenandotherheavyspeciesbyprotectingthem.Thoseupperestimatesofnon-thermallossprocessesfromweaklymagnetizedOPsdonotcomeclosetoexhaustingtheimportantwaterreservoirofanOPthatislocatedfartherthan0.05AU.Forinstance,200Earthoceans(200×1.45×1021kg)representonly1.5%ofthemassofa6Earth-massOP.2.2.3.Howclosetoitsstarcanweexpecttofindanocean-planet?AccordingtoFig.4,non-thermalescapeinducedbythestel-larwindseemstodominatetheatmosphericlossforMpl"5MEarth.Forlargermasses,thesurvivalofthewaterreservoirislongerthan5Gyrs.Intheabsenceofmoredetailedmodelsforthestellarwind-inducedlosses,ourassumptions(100%ef-ficiency,veryintensestellarwind,nointrinsicmagneticfield)certainlyoverestimatesignificantlythesolarwind-inducedloss(Section2.2.2).Withtheseovermaximizingassumptions,find-ingnon-thermallossesmuchhigherthanthethermalescapewouldhavepreventedusfromprovidingafirmconclusiononthesurvivalofOPs.Fortunately,wefindanupperlimitonthesolarwindinducedlossthatislower,oronlyslightlyabove,thethermalloss.Wecanthusderiveanestimatefortheclosestdis-tanceatwhichOPswillnotbeaffectedbywaterlosstospace:asatypicalresult,a6MEarthOPwillkeepmostofitswaterwhenitisfartherthan0.04AUfromitsstar.ThisconclusionisverysimilartotheresultsobtainedbyKuchner(2003),whoalreadystudiedthesurvivalofthevolatilecontentofshortperiodOPs:onFig.4ofhispaperonecanseethatthelifetimeof5MEarthvolatile-richplanetat0.04AUis5Gyr.ButKuchneralsopointedout2potentialproblemsaf-fectinghiscalculationsattheshortestorbitalperiods.ThefirstoneisrelatedtotheRXUV/Rplratio(discussedinSection2.2.1)thatstronglyoverestimatesthelossrateathighXUVirradiationinWatson’snumericalapproachofthehydro-dynamicescape.ToestimatetheefficiencythelossathighXUVirradiation,weusedrecenthydrodynamiccalculations(Penzetal.,2006)andfoundlossesmuchbelowwhatisgivenbyWat-son’sscheme.Itshould,however,benotedthatthevalueofthisefficiencyhasastrongimpactonthedeterminationofthemini-mumdistance,andthathydrodynamicmodelingofatmosphericescapeisafieldinprogress.Differentestimatesofthiseffi-ciencymightbeobtainedinthefuture.AsecondpointraisedbyKuchneristhattheatmosphericlosscaninduceanexpansionofthewholegaseousenvelope.Becauseofpositivefeedbackonthelossrate,thiseffectcouldleadtoarunawayescapeandtothelossofthewholevolatilereservoirwithinamuchshortertimescalethanwhatisfoundherebyassumingnocouplingbetweenthehydrostaticstruc-tureandthemassloss.ThisrunawaylosswasstudiedforgasgiantsbyBaraffeetal.(2004)andwasshowntooccurwhenthetimescaleofthelossm/mbecomesshorterthanthether-mal(Kelvin–Helmholtz)timescalerequiredforthermalread-justmentoftheenvelopetoamodificationofitsmass.Thisconditioncanbewritten:(6)mm<Gm2RplLpl.Here,misthemassofthewaterenvelopeandLplistheplanetluminosity,whichcanbewrittenas4πR2plσTeffbyassumingameanbrightnesstemperatureTeff,or2πR2plσTeffifthereisnoredistributionoftheincomingenergybetweenthedayandnighthemispheres.Therunawayconditioncanbewritten(7)2σT4effεFXUV<1orTeff<"εFXUV2σ#1/4,whereεistheconversionefficiency.Fortheclosestorbitaldis-tancesweconsidered,thisconditionfortherunawayimpliesTeff<240K.Attheseshortorbitaldistances,realisticeffectivetemperaturescannotdropbelow500K,andwecanassumethatthermalreadjustmentareinstantaneouscomparedtothelosstimescale,whichiscompatiblewithourquasi-static/hydrostaticapproach.Weconcludethattherunawaylossofwaterisnotex-pected.However,westressagainthecentralinfluenceoftheefficiencyε.HigherefficienciesthantheoneobtainedbyPenzetal.(2006)couldtriggerarunawaylossandthelossofthewholewaterreservoirwithinshorttimescales,asdescribedinBaraffeetal.(2004). JID:YICARAID:8261/FLA[m5+;v1.73;Prn:12/06/2007;8:22]P.9(1-16)Identificationofhotocean-planets93.CouldRplandMplmeasurementsbeaccurateenoughtodiscriminateocean-planetsfromrockyplanets?Inordertoknowwhetheranexoplanetisanocean-planetorarockyone,itsmassandradiusmustbemeasuredwithhighenoughaccuracysothatthepositionoftheerrorboxinthe(Mpl,Rpl)planeallowsadistinctionbetweenthepredictionsforthosetwotypesofplanets,assumingthatthemodelisadequate(Fig.1).Forsmallmassesandradii(Mpl<10MEarth,Rpl"2REarth),thisisanotatrivialtask.Hereafter,weestimatethedifferentuncertaintiesthatdefinetheerrorboxes.3.1.Whatarethedifferentsourcesofuncertainty?Aroundagiven(Mpl,Rpl)point,thephysicalquantitythattheplanetarymodelspredicttobedifferentisthemeanden-sity.Thisdensity,)ρ*=(3/(4π))MplR−3plcanbeobtainedbyproperobservations.TheplanetarymassresultsfromDopplervelocimetry.Theplanetaryradiusisobtainedfromthetargetstarradiusestimateandtherelativestellarfluxdropduringthetransit,i.e.Rpl=Rst(F/F)1/2,ifthestellarlimbdarkeningcanbeneglected.Theuncertaintiesonthemass,thestellarradiusestimateandfluxmeasurementsbeingindependent,theirvariancesaddinquadrature.Weestimatetheuncertaintyonthe1/3powerofthedensity,becausethisquantityhasarelativeuncertaintythatissmallerthanthatonthedensity,whichismoreappropriateforlinearexpansions(Protanov,2002):(8)"σρ1/3ρ1/3#2="σMpl3Mpl#2+"σRstRst#2+"σF/F2F/F#2.ConsideringthecaseMpl=6MEarththatcorrespondstoabigtelluricplanetbutthatisstillfarfromthelimitwherethehy-drogengasislikelytohavebeenaccreted,theinternalstructuremodelingbyLégeretal.(2004)yieldsradiiofR1=2.0REarthandR2=1.63REarthforocean-planetsandrockyplanets,re-spectively.Therelativedifferenceforthecorrespondingρ1/3is:(9)ρ1/3ρ1/3=RplRpl=2.0−1.631.8=20.6%.AssumingaGaussiandistributionoferrors,a2σcharacteriza-tionofρ1/3/ρ1/3wouldresultina95%confidencediscrimi-nationbetweenthetwokindsofplanets.Therefore,weconsiderthatthestandarddeviationshouldsatisfy:(10)σρ1/3ρ1/3"10.3%.Next,thedifferentcontributionsinrelation(8)areestimated.3.2.UncertaintyontheplanetarymassThesemi-amplitudeoftheradialvelocitywobbleduetothepresenceofaplanetaroundasolartypestaris(11)K=0.09"MplMEarth#"MstMSun#−1/2"a1AU#−1/2ms−1.Thecorrespondinguncertaintyinthemassis(12)σMplMpl=$"σKK#2+"σMst2Mst#2%1/2.Theuncertaintyinthestellarmassesofthetargetsstarsthatareconsidered(mv<13)canbeestimatedfromthatinthestellarradius,σRst/Rst∼5%(Section3.3),andM(R)empir-icalrelation(Cox,2000):log(R)=0.92log(M)+const.Theresultisanambiguityof∼6%.Thisisaconservativevaluebecause,inpractice,themassisdeducedfromobservablequan-titiesusingatmosphericmodels,whereastoobtaintheradiusanadditionaloperationisneededthatimpliesstellarevolutionmodeling.ThisresultisinagreementwithCharbonneauetal.(2006),whoestimatethattheuncertaintyonthestellarmasscanbeashighas5%.Anyhow,thistermhasanegligiblecon-tributiontotheplanetarymassuncertaintybecauseitleadstoσMst/2Mst"3%,whereasσK/K∼20%(seebelow),andthehierarchy(σMst/2Mst)2%(σK/K)2alwaysstands.OneoftheverybestDopplerinstrumentspresentlyavail-ableistheHARPSspectrometerattheESO-LaSilla3.6mtelescope.Forfaintstars(mv>10),itsuncertaintyisclosetothefundamentalshotnoiselimit.Itwillbethemostaccurateinstrumentforthefollow-upofplanetarytransitcandidatesde-tectedbyCoRoT.Inthatprogram,theDopplermeasurementswillbeperformedinfavorableconditionsbecausetheplanetaryephemeriswillbedeterminedbeforehandbytransitphotome-try(transitperiodandepochs).Theonlyunknownswillbethesemi-amplitudeofthesinusoidalradialvelocitycurveKandthecenter-of-massvelocityv0,ifacircularorbitoftheplanetisassumed.TheshotnoiseuncertaintyofHARPSdependsnotonlyuponthestellarmagnitudeandtheintegrationtime,butalsouponthespectraltypeandtherotationalbroadeningvsinioftheobservedstar(Bouchyetal.,2001).Forthemostappro-priatespectraltypes(fromG5toK5)andstarswithlowrotation(vsini<2km/s),theexpectedshotnoiseuncertaintyonindi-vidualradialvelocitymeasurements,σV,isgivenbyMayoretal.(2003)andPepeetal.(2005):mV=14→σV=6m/s(1h),mV=12→σV=2m/s(1h).Forsuchstars,theuncertaintiesduetoshotnoisearelargerthaninstrumentalsystematicerrors(estimatedto0.8m/sonHARPS).Wesupposeherenon-activestarswithsurfaceveloc-itiesatthelevelof2m/srms,whichisthecaseofthemoststablesolar-likestarsmonitoredwithHARPS.FromσVandthenumberofobservations,theerrorσKonKcanbecomputedasfollows.Incaseofacircularorbit,theradialvelocityreads(13)v(t)=v0+Kcos"2πtTorb+ϕ#,wheretheorbitalperiodTorbandthephaseϕareknown,thankstotransitphotometry.Therefore,onlytwoparameters,thesys-temicvelocityv0andthesemi-amplitudeK,shouldbedeter-minedfrommeasuredvelocities.Least-squaresfittingwouldbethestandardprocedureforthatpurpose. JID:YICARAID:8261/FLA[m5+;v1.73;Prn:12/06/2007;8:22]P.10(1-16)10F.Selsisetal./Icarus•••(••••)•••–•••FollowingtheproofinSection15.2ofPressetal.(2002),andassumingthatallmeasureshavethesameerrorσv,onecanshowthatthevarianceofKis(14)σ2K=σ3v&Ni=1cos2’2πtiTorb+ϕ(−1N’&Ni=1cos’2πtiTorb+ϕ((2.Ifthemeasurementsareuniformlydistributedalongtheorbitalperiod,thefirstsumcanbeapproximatedbyNtimesthein-tegralofasquaredcosineoveritsperiod,i.e.,N/2,andthesecondsumcanbeapproximatedbythemeanvalueofacosineoveritsperiod,i.e.,zero.Therefore,(15)σ2K≈σ2vN/2.However,ifwetakefulladvantageoftheknowledgeofthephaseφandperformthemeasurementsinequalnumbersatthemaximaandminimaofthevelocity,thenthefirstsumisequaltoN(becausethesquaredcosineisalwaysequalto1),whilethesecondsumisstillzero.Hence(16)σ2K≈σ2vN.Inpractice,itmightbedifficulttotakemeasurementsonlyattheextremaofthevelocitycurve.ArealcaseiswellillustratedbyFig.2inO’Donovanetal.(2006),whoperformedtheirmea-surementsclosetotheextrema.Apossiblevalueis(17)σ2K≈σ2vN/1.2.Weassumethatforanimportantproject,amaximumintegra-tiontimecanbeabout40h,correspondingtoacumulativetimeofabout5nights.InordertoexplicitlyestimatethedependenceofσVuponthedifferentparameters(mv,t,Φtel),wecompute(σV)0foragivensetofparametersandforareferencecase,andweexpressσVasafunctionofitandtheparameters.Intheshotnoise-limitedregime,theuncertaintyontheradialvelocityreads(18)σv(σv)0=100.2(mv−mv0)"tt0#−1/2"ΦtelΦ0#−1,whereΦtelisthetelescopediameter,tthecumulativeintegra-tiontime,andtheindex“0”correspondstothereferencecase.Thereferencecasethatweconsideris:Mpl,0=6MEarth,a0=0.10AU,Mst=MSun.Relation(11)givesK0=1.7m/s.Anmv=12starobservedwithHARPSatthe3.6mESOtele-scopeduringt0=40h,leadsto(σK)0=0.35m/s(Eq.(17))and(1/3)(σK/K)0=0.068.Usingrelations(18),theuncer-taintyontheplanetarymass(relation(12))reads(19)13σMplMpl=6.9%A"a0.10AU#1/2100.2(m−12),withA="Mpl6MEarth#−1"MstMSun#1/2"t40h#−1/2"Φtel3.6m#−1.Fig.5.RelativeerroronthestellarradiusplottedagainstthemagnitudeoftheCoRoTtargets.ThedotsshowtheerrorestimatedfromtheVandKactualmagnitudesfromrelation(22)inKervellaetal.(2004),whentheuncertaintyonthereddening,ordistance,isadded.Thefulllineshowsthefittothedata.3.3.UncertaintyonthestellarradiusThestellarradiusofCoRoTtargetswillbeestimatedbyin-depthfollow-upobservationsoftheparentstar,includinghigh-resolutionhigh-SNRspectroscopy.Afirstideais,however,obtainedfromcalibratedrelationshipsofcolorindiceswithstel-larangulardiameters(Kervellaetal.,2004).Apparentmagni-tudesintheVandKfiltersaremeasuredwithaprecisionoftypically0.03magnitudesintheCoRoT/exoplanetstellarcata-logue.Apreliminaryspectralclassificationisthenderivedfrombroadbandspectrophotometryinthevisibleandnear-infrared,withanestimationofthereddening(Moutouetal.,inprepa-ration).High-resolutionspectroscopywillthenallowamoreaccurateestimationoftheeffectivetemperature,witherrorbarsoftypically50to100K(2–3%ofthetemperature).Thiswillfurtherconstrainthereddening,attheleveloftypicallyanad-ditional0.05magnitudeuncertainty,whichincludestheerrorondistanceestimation.ThefinalerroronthelinearradiusoftheCoRoTtargetsisthenobtainedbycombininganerrorontheangulardiameter(derivingEq.(22)inKervellaetal.,2004)withanerroronthedistance(fromspectraltype,reddening,ab-soluteluminosity).Fig.5showsthecalculatedfinalerroronthestellarradiusforCoRoTtargetsinthemagnituderange12–14.Notethatothermethodsofradiusestimationgiveresultsthatareconsistentwithin5%(DiBenedetto,2005).Forcomparison,thestellarradiusofstarsinknowntran-sitingsystemsisestimatedwitherrorsoftypically5–7%forOGLEsystems(e.g.Santosetal.,2006).TheexpectationsforCoRoTtargetsisoftheorderofwhatisachievedonOGLEtar-getsofsimilarmagnitudes,asathoroughfollow-upobservationprogramisforeseentoconstrainatbestthestellarparameters.WeconcludethatthestandarddeviationonRstisestimatedas(20)σRst/Rst∼5%forstarsbrighterthanmv∼13,whichimpliesaverycarefulphotometricandspectroscopicanalysisofthetarget,asinter- JID:YICARAID:8261/FLA[m5+;v1.73;Prn:12/06/2007;8:22]P.11(1-16)Identificationofhotocean-planets11stellarextinctionwillbepresentinthelineofsightofCoRoTtargets.3.4.UncertaintyontheamplitudeofthetransitTheamplitudeoftherelativefluxvariationduringatransit,F/F,is(Rpl/Rst)2,or:(21)FF=0.84×10−4"RplREarth#2"RstarRSun#−2.ForCoRoT,theuncertaintyonthetransitdepthmeasurementisexpectedtobe1.2timestheshotnoiseforstarswithmv=11–14(Fridlundetal.,2006).Theshotnoisedependsontheinstrument,thestellarmagnitude,thenumberkoftransitsandtheirindividualdurationtr.Foratotaldurationobservation,trun,anorbitalperiodTorb,andanimpactparameter(theratioofthedistanceoftheprojectedplanettrajectorytothestellarcenterandthestellarradius)of0.5,thecumulativetransitingtime,ttot,is(22)ttot=ktr,withtr=3.56(a/0.10AU)1/2(Rst/RSun)(Mst/MSun)−1/2h,Torb=11.5(a/0.10AU)3/2(Mst/MSun)−1/2days,k=trun/Torb.Fisobtainedbymeasuringthedepthofthetransitinthestellarlightcurve,F=Ftr−Fbg.Thebackgroundterm,Fbg,canbeaccuratelyestimatedbyaveragingoverlongdurations;therefore,weconsiderthattheshotnoiseonFismainlythatonFtr.Itisdeterminedbycalculatingthenumberofphotoelec-tronsatthefluxlevelFtr∼F(notatthelevelF)duringthetotaltransitingtime.TheuncertaintyonF,innumberofpho-toelectrons,is(23)σF=1.2(Fttot)1/2.TherelativeuncertaintyonFbeingmuchweakerthanthatonF,aftersomealgebra,onereads(24)σF/FF/F≈1.2(F/F)(Fttot)1/2.CoRoThasa0.27mentrancepupilandobservesduring150-dayruns.Itproducesa2.6×104ph-els−1fluxfora12thmagnitudestar(http://CoRoT.oamp.fr).Consideringasaref-erencecase,thetransitofaRpl=2REarthplanetinacircularorbitaroundasolartypestar(F/F=3.36×10−4),atdis-tance0.10AUobservedwithCoRoT,theuncertaintyis(25)"12σF/FF/F#0=2.7%.Keplerhasa0.95mentrancepupilandobservesduring4years.Fora6.5hdurationtransitanda12thmagnitudestar,thedifferentialphotometryaccuracyisestimatedat20ppm,andismainlyshotnoiselimitedforfainterstars(D.Koch,per-sonalcommunication).Forourplanet-starreferencecase,forthewholemissionduration,onefinds(26)"12σF/FF/F#Kepler=0.25%.Ingeneral,foramainlyshotnoiselimitedobservation,theun-certaintyis(27)12σF/FF/F=2.7%B"a0.10AU#1/2100.2(mv−12)withB="Rpl2REarth#−2"RstRSun#3/2"trun150days#−1/2"Φtel0.27m#−1.Itisremarkablethatthedependencesofbothuncertaintiesontheplanetarymass(Eq.(19))andtransitamplitude(Eq.(27))upontheorbitalradius,a,andstellarmagnitude,mv,arethesame.3.5.Capabilitiestoidentifythenatureofthetelluric-likeplanetsThedifferenttermsinEq.(8)arenowestimated.Using(19),(20)and(27),theconditionfordiscrimination(Eq.(10))reads(28)(0.05)2+)(0.069A)2+(0.027B)2*"a0.10AU#100.4(mv−12)"(0.103)2.ConsideringthecaseofaplanetwithM=6MEarth,R=2REarthinfrontofaSun-likestar,observedwith:•CoRoTduring150daysandwith40hofHARPSfollow-upona3.6mtelescope,onereadsA=B=1.InthesquarebracketofEq.(28),thetermcontainingA(theuncertaintyonthemass)dominatesthatwithBbyafactor5.AlthoughwehavealreadyconsideredaverylongintegrationtimefortheDopplermeasurement,thebottleneckremainstheun-certaintyontheplanetarymass,evenwhenusingthesmalltelescopeofCoRoT;•Keplerduring4yrsanda40hfollow-upwithanHARPS-likespectrometerona3.6m(8m)telescope,onereadsA=1(0.45),B=0.09.Now,inEq.(28),theuncertaintyonthemassfullydominatesthatontheradiusbyafac-tor∼800(160).Thewaytoimprovethediscriminationbetweenthetwotypesofplanetsisthentoimprovetheac-curacyofthemassbyusinglargertelescopes.Thisisundertheassumptionthat,forfaintstars,theuncer-taintiesonMplandF/Faredominatedbyshotnoise,ratherthanbytheastronomicalstellarnoiseortheinstrumentcapa-bilities.Arealisticdetectionexercise,includingstellarnoisesinCoRoTsimulatedlightcurves,hasshownthatstellarmicro-variabilitylimitsthedetectiononlyinextremecases(Moutouetal.,2005);however,theimpactofstellarmicro-variabilityontransitparameterestimationisstillunderstudy.Fig.6showstheregionsinthe(mv,a)planewherethedis-criminationbetweenanocean-planetandarockyplanetwith6 JID:YICARAID:8261/FLA[m5+;v1.73;Prn:12/06/2007;8:22]P.12(1-16)12F.Selsisetal./Icarus•••(••••)•••–•••Fig.6.Regionsinthe(targetmagnitudemv,planettostardistancea)planewherethediscriminationbetweenanOPandarockyplanetcanbemadeat2σ(95%confidence)whentheirmassesareMpl=6or10MEarth,aroundGVstars.Thecorrespondingradiiofocean-planetsareROP=2.0,or2.3REarth,respectively.TheplanetaryradiusisdeterminedusingeithertheCoRoTinstrument(a)ortheKeplerone(b).ThemassdeterminationismadewithaDopplervelocimeterbasedontheperformancesoftheHARPSspectrometermountedona3.6m(40hofcumulatedobservations)or,speculatively,onan8mtelescope.Itisassumedthatbothtransitamplitudeandstellarreflexvelocityaccuraciesareshotnoise-limited.Theuppershadingindicatesaregionthatisnotaccessibleasaresultofinstrumentalconstraints(toolongorbitalperiods).ThelowershadingsindicatearegionwhereOPsmay,ormaynot,havekepttheirwater,becauseourestimatesofwatererosionareonlyupperlimits.Panels(c)and(d)arethehistogramsofthetargetstarsforCoRoTandKepler,respectively.Iftheswellingofahotocean-planetwasconfirmedtobeof10%(seeSection2.1),itwouldsignificantlyhelptodiscriminatebetweenOPandrockyplanets.Forinstance,theboundaryforthe6MEarthplanetwouldapproximatelymovetothatofthe10MEarthcase.Earth-massora10Earth-massplanetaroundaGVstarcanbemadeat2σ(95%confidence).Twotransitinstrumentsarecon-sideredforthetransitmeasurements,thatoftheCoRoTandKe-plermissions.TheDopplerdeterminationofthemassisbasedontheperformancesoftheHARPSspectrometermountedona3.6-mtelescope(thepresentsituation)oran8-mtelescope(aspeculativesituation).Itappearsthatthetargetstarsoftran-sitmissionsarefaintobjects,whichisaserioushandicapfortheDopplermeasurements.Asaresult,thenatureofsmallplan-etscanbeonlybedeterminedforthebrighteststars(mv<13)andtheshortest-periodplanets,atleastforCoRoT.ThispointsouttheimportanceofdetermininghowclosetotheirstarsOPsareexpectedtokeeptheirwater(Section5).Ourestimatesofwatererosionareonlyupperlimits.Planetsthatareinsidetheboundarieswehavefoundmay,ormaynot,havelosttheirwater.Theirexactnaturewillbeconstrainedbyfurthermodelinganddeterminedbytheobservations.Atfixedmagnitude,starswithsmallerradii,e.g.Kstars,aremorefavorable,buttherearefewerofthemthanGstarsinamagnitude-limitedsample.IntheCoRoTtargetlist,thefractionofstarswithmagnitude11,12,13and14areexpectedtobe1,2.3,7and15%ofthetotalnumberoftargets,respectively(Bordéetal.,2003).Thesestarsarenotthemajorityofthetargetobjects,buttheyarenotaminutefractionofthemeither.However,starswithmv"12willsaturatethebrightestpixelsofthepowerSpreadFunction(PSF)intheCCD.Presently,thesestarsarenotconsideredforphotometrybecausetheyaretechnicallymoredifficulttomoni-tor,butthissituationmayimprovesometimeafterthebeginningofthemission.Tobeoptimistic,wewillkeepthesestarsinourfurtherestimates,especiallyastheyarethosewiththehighestS/N.Withinthearbitraryhypothesisthateachstar,approximatedasaGstar(themostprobablestellartypeinthemagnitude-limitedCoRoTfield),hasa6ME[10ME]planetat0.05AU,thenumberofdetectionsbyCoRoTwithsufficientaccuracyfordeterminingthenatureofthisplanetwouldbethetotalnumberofstarswhicharebrightenough(mv"13,accord-ingtoFig.6a)timestheprobabilityoftransit(Rst/a=9%).Forthetotalmission(5stellarfields)andMpl=6ME,this JID:YICARAID:8261/FLA[m5+;v1.73;Prn:12/06/2007;8:22]P.13(1-16)Identificationofhotocean-planets13numberwouldbethenumberofstarswithmv"13,or5×12,000×(1%+2.3%+7%)times0.09,or∼550detections.ForMpl=10ME,thelimitismv"14,andthecorrespondingnumberwouldbe∼1400detections.Inallcases,along(80h)Dopplerfollow-upofthetransitsonatleasta3.6mtelescopeisnecessary.However,thereadershouldnotmisinterpretthesenumbers.WearenotclaimingthatCoRoTisgoingtodetecthundredsof10Earth-massplanets,forthemerereasonthatitwouldbeimpossibletodedicate40hofHARPStimeperstarforsuchanumberofobjects(40h×550objects=22,000h∼3000nights∼8yearsfulltimeassuminga100%efficiency!).Theestimateismainlyinterestingforthepurposeofdetermin-ingatwhatleveltheabsenceofanydetectionwouldlimittheabundanceofsuchplanets.Ifthenumberofplanetswithmassesintherange6–10MEarthanddistancetotheirstarlessthan0.1AUwassimilartothenumberofhotJupiters,i.e.1%,thenumberofexpecteddetectionswouldbebetween5and10.Conversely,anon-detectionofsuchplanets,ifouraccuracyestimatesarecorrect,wouldindicatethattheyarenotpresentatthelevelof∼0.5%.IfamoreaccurateDopplerfollow-upwerepossible,thosediscriminatingdetectionswouldbemorenumerousbecausethemassmeasurementaccuracyispresentlythebottleneck(Eq.(28)).ForKepler,accordingtoFig.6b,ifthefollow-upwasmadewithperformedwithaDopplervelocimeteranalogoustoHARPSona8mtelescope,forMpl=6MEarth,onewouldex-pectthatplanetsata=0.10AUcanbecharacterizedaroundstarswithmv"14.5.Thetotalnumberofmonitoredstarsis100,000.IntheKeplerfieldthere∼600,000starswithmv"16.5,15%ofwhichhavemv"14.5(D.Koch,personalcom-munication).Withtheconservativeassumptionthatthedistrib-utioninmagnitudeofthemonitoredstarsisthesameasfortheotherstarsofthefield,andtheassumptionofa1%abundanceofhotearthswithmasses6–10MEarth,onewouldexpecttobeabletocharacterize5to10ofthem.ThedifferencewiththeCoRoTcaseseemsnottobehuge,althoughitmustbepointedoutthattheprospectofcharacterizingterrestrialplanetswithKeplerisonasignificantlyfirmerbasisbecausesomeofthesemeasurementsareexpectedtohaveafairS/N,whereasthisisnotsofrequentwithCoRoT.WepointoutthekeyinterestofbuildingaDopplerspectrom-eterwithperformancesimilartothatofHARPS,onan8mclasstelescopeforthestudyofterrestrialexoplanets.Thisistrueforthefollow-upofCoRoTanditisespeciallyobviousforthatofKepler.Thisisundertheoptimistichypothesisthatforfaintstars(mv=11–14)bothtransitamplitudeandstellarreflexveloc-ityarebasicallylimitedbyshotnoiseandnotbythestellarvariabilityorinstrumentnoise.Ifthestellarnoisedominates,usinglargetelescopeswillnotbeuseful;theonlywayto(slowly)improvetheS/Nwillbetoincreasetheintegrationtime.4.Discussion:Mass–radiusdegeneracybetweenOPs,H2-richplanetsandcarbonplanets?Aspointedoutintheintroduction,theoreticalworks(Alibertetal.,2006;Rafikov,2006)predicttheexistenceofshortperiodplanetswithacoremass>6MEarthandawidediversityofH2–Heenvelopes:fromMenv%1MEarth(OPsandsuper-telluricplanets),toafewMEarth(likethehotNeptunesinthesystemHD69830),andfinallyMenv-1MEarth(giantplanets).Inplanetaryformationmodels,thecoreisinitiallymadeoficy–rockymaterialfrombeyondthesnowline,butitaccretesmorerockyplanetesimalswhenmigratingbetweenthesnowlineanditsfinalshortperiodorbit.Thecollisionsoccurringdur-ingtheplanetformationcanleadtoafinalcompositionsignif-icantlydepletedinvolatilescomparedwithatypicalcometarycomposition.Also,theplanetcanofcourseconsistofrefrac-torymaterialsonly,ifmigrationstartedclosertothesnowlineandifenoughrockyplanetesimalsareavailable.Therefore,agivencouple[R,M]maycorrespondtoeitheranOPofmassMandanegligibleH2–HeatmosphereortoaplanetwithM=Mcore+Menv.Forthesamemass,ahy-drogenenvelopewouldbesignificantlymoreexpandedthananH2Oenvelope.AttheorbitalperiodsrelevantforCoRoT,thehightemperatureduetothestrongirradiationwouldpro-duceanevenmoreimportantswellingoftheH2–Heenvelope.AnH2–Heenvelopemoremassivethanabout0.1MEarthwouldproduceaplanetaryradiusofmorethan3REarth,whichislargerthanan10MEarthOP.Togiveanexample(kindlypro-videdbyI.Baraffe),aplanetat0.1AU,witharockycoreof[Mcore=7.5MEarth,Rcore=1.76REarth]andanH2–Heen-velopeof2.5MEarth,wouldhaveatotalradiusof5.6REarth.Therefore,tomimicanOPwithnoH2–He,themassofH2–HesurroundingarockycorehastobesignificantlylessmassivethanthewaterreservoirofthisOP.ObservingaplanetwiththispreciseamountofH2–He,obtainedbyaccretionofthegaseousenvelopeanditssubsequentpartiallossbyatmosphericescape,appearsqualitativelyratherunlikely.Althoughthediversityofmass–radiusrelationshipsshouldclearlybeaddressedquantitativelyusingamodelofplanetformation(includingmigration),therebyprovidingarealisticdistributionofM(MRocks,MIce,MH2)versusorbitaldistance,weconcludeherethattheidentificationofmassiveOPscannotrigorouslybedonewithasinglecase.Toshowupamongtran-sitingplanetshavingawiderangeofH2–Heenvelopemasses,OPswithnegligibleH2–Heatmosphereshavetoberelativelyabundant.Regardingthispoint,short-period,low-massOPs(M<6MEarth,a<0.1AU)providethebestcase,becausetheaccretionofgasisnotexpectedtobeimportantforthesemasses(Rafikov,2006),whiletheescapeofH2isfast(muchfasterthanH2O)attheseorbitaldistances(Baraffeetal.,2005).Moremassive,H2-deficientOPsatlongerperiodsmaybedifficulttodistinguishwithinapopulationofplanets(OPsortelluric)withH2envelopes.ItisalsoimportanttonotethatallratiosMIce/MRocksbe-tween∼0and1areexpected.Foragivenplanetarymass,planetaryradiishouldbefoundatanyvaluebetweenthatofasilicateplanetandourOPprototypes(i.e.betweenthetwo JID:YICARAID:8261/FLA[m5+;v1.73;Prn:12/06/2007;8:22]P.14(1-16)14F.Selsisetal./Icarus•••(••••)•••–•••curvesofFig.1).ForM>6MEarthwecanexpectatwodis-tributionsofobjects,hopefullyseparatedbyagap,fortheOPprototypesandthehotNeptune/Saturn/Jupiterobjects.KuchnerandSeager(2005)proposedtheexistenceofcar-bonplanets,formedinenvironmentscharacterizedbyC/O>1.Intheabsenceofdetailedmodelingoftheirstructureandoftheirradiusasafunctionoftheirmass,itisdifficulttoknowifthesecarbonplanetscouldbeidentifiedorconfusedwithsilicate-dominatedplanetsorOPs.Thatisanopenquestionforfurtherstudies.WhatwecansayatthispointisthattheC/Oratiocanbemeasured(withinsomesignificantuncertain-ties)instellarphotospheres,soweshouldbeaware,fortherare(any?)mainsequencestarswithhighC/Oratiowhereatran-sitwillbeobserved,thatacarbonplanetmightbethere.Evenforstarswitha“solar”C/Oratio,planets(e.g.Jupiter)canstillbesignificantlyenrichedincarbon.Inaddition,theicymater-ialfoundbeyondthesnowlinedoesnotconsistsonlyofH2Oice,CO2,CH4andCOarealsoimportantconstituents.Theac-cretionofoneEarthmassofH2Oiceimpliestheaccretionofabout0.1Earthmassofthesecarbon-basedvolatile,whichcanhaveasignificanteffectonthestructureandtheevolutionoftheplanet.Weshouldthusremainawarethatalargediversityofel-ementalcompositioncanexistamongplanetsandthatcannotbecharacterizedonlybythew=water/silicateratio.Never-theless,thedetectionbyCoRoTorKepler,andaroundK,GorFstars,ofshort-periodlow-massplanets(Torb"1month,Mpl"6MEarth),withradiisignificantlylargerthansilicate-dominatedplanetswouldunveilapopulationofvolatile-richplanets.ThebestcandidateforthisvolatilebeingH2O.5.ConclusionWehaverevisitedthetheoreticalestimateofthephysicaldensityofhotocean-planets(OPs)andconcludedthatthere-sultsestablishedforplanetsinthehabitablezone(HZ)arestillvalidmuchclosertotheirstar,becausetheirexpectedswellingshouldremainlimitedwhencomparedtotheplanetaryradius.Curiouslyenough,averyhotOPhasathickwateratmosphereindirectcontactwitha(highpressure)icemantle.ModelsoftheinternalstructureofOPsyieldameandensitysmallerthanthatofrockyplanets.Forthesamemassthereisadifferenceinradiusof∼20%,thatcanreach30%forthehottestOPs,duetotheswellingoftheirwaterenvelopeInthispaperwehaveconsideredwhetheranidentificationofOPsbasedonthemeasurementoftheirdensitiesisactuallypossibleinthenear-tomid-future.Ifstrongenough,migrationcouldbringthesespeculativeplanetstowithintheclosevicinityoftheirparentstar,wheretheirmassesandradiicanbebettermeasuredbyRadialVeloc-ityandbytransitphotometry.CoRoTwilldetectplanetswithperiodslessthan75days,whereasKeplerwillhaveaccesstoperiodslessthan1.33years.Inbothcases,close-inplanetswillbetheeasiestonestodetectandtocharacterize.Wehaveestimatedtheerosionrateofahotocean-planetbythermalescapedrivenbyexosphereheating(withXandUVradiation)andbynon-thermalescapedrivenbyejectedparticlesfromthestar.Wefoundthatthewaterreservoirisonlyweaklyaffected,sothatanOPshouldkeepitspeculiarcompositionevenfairlyclosetoitsstar.Forexample,infrontofasolar-typestar,a6Earth-massOPshouldretainmostofitswatercontentformorethan5Gyrsifitisbeyond0.04AUfromitsstar.Then,wehaveaddressedtheaccuracyneededtodiscrimi-natebetweenOPsandrockyplanets,assumingthattheplan-etarymodelsareadequate.ThesourcesofuncertaintyontheplanetarydensityarethoseonthemassdeterminationbyRa-dialVelocitymeasurements,thestellarradiusdetermination,andthephotometricmeasurementduringthetransits.Asex-pected,theaccuracyoftheKeplerphotometryishigherthanthatofCoRoT.Howeverwiththepresentlyavailableinstru-ments,itistheuncertaintyonRadialVelocitymeasurementsthatisthelimitingfactorforexpecteddetectionsbybothtran-sitmissions.Asaresult,thedeterminationofthenatureoftheseplanetsseemspossibleonlyintheadequatedomainsofthe(mv,a)planethatareshowninFig.6forbothmissions.Theycorre-spondtothebrighteststars(mv<14)andtotheplanetsclosesttotheirstar,especiallyforCoRoT,butbeyondthelimitforthesurvivaloftheirwaterreservoir.Ifeachstarhada6–10Earthmassplanetat∼0.10AU,anarbitraryhypothesis,thenumberofdetectionswithCoRoTwouldbeseveralhundred.Conversely,theabsenceofdetectionofsuchplanetswouldindicatethattheyarenotpresentatthelevelof∼1%.AclearconclusionofourstudyisthatfullbenefitofthehighphotometricprecisionofCoRot,andparticularlyKepler,canbeobtainedonlyifanewgenerationofRadialVelocityinstru-mentsisbuiltthatcanmakeaccuratemeasurementsonfaintstars.Inthatcase,theidentificationofOPscouldbedoneonasignificantlylargersampleofstars.Inthemeantime,asignif-icantfractionofhighperformanceDopplervelocimeterssuchasHARPSshouldbedevotedtothefollow-upofthemostin-terestingcandidatesfoundbyCoRoT,ifwehopetobeabletodiscriminatebetweenthedifferentpossiblecompositionsofter-restrialplanets.AcknowledgmentWearegratefultoOdileDutuit,RolandThissen,Alexan-droMorbidelli,andDavidKochforvaluablediscussionsonthewatermoleculeabsorptionintheEUV,thepossibleabundanceoftheseputativeplanetsandpreciseinformationonKepler,re-spectively.WearealsogratefultoananonymousrefereeandtoMarcKuchnerforvaluablecommentsonourinitialmanu-script.ThisworkwassupportedbyCNRSandCNESinpartbytheProgrammeNationaldePlanétologieandtheGroupedeRecherche“Exobio,”andundercontract1256791withtheJetPropulsionLaboratory(JPL)fundedbyNASAthroughtheMichelsonFellowshipProgram.Wealsoacknowledgetheben-efitfromtheISSITeam“EvolutionofHabitablePlanets.”ReferencesAlibert,Y.,Baraffe,I.,Benz,W.,Chabrier,G.,Mordasini,C.,Lovis,C.,Mayor,M.,Pepe,F.,Bouchy,F.,Queloz,D.,Udry,S.,2006.Formationandstruc-tureofthethreeNeptune-massplanetssystemaroundHD69830.Astron.Astrophys.455,L25–L28. JID:YICARAID:8261/FLA[m5+;v1.73;Prn:12/06/2007;8:22]P.15(1-16)Identificationofhotocean-planets15Alonso,R.,Brown,T.M.,Torres,G.,Latham,D.W.,Sozzetti,A.,Mandushev,G.,Belmonte,J.A.,Charbonneau,D.,Deeg,H.J.,Dunham,E.W.,O’Dono-van,F.T.,Stefanik,R.P.,2004.TrES-1:ThetransitingplanetofabrightK0VStar.Astrophys.J.613,L153–L156.Baraffe,I.,Selsis,F.,Chabrier,G.,Barman,T.S.,Allard,F.,Hauschildt,P.H.,Lammer,H.,2004.Theeffectofevaporationontheevolutionofclose-ingiantplanets.Astron.Astrophys.419,L13–L16.Baraffe,I.,Chabrier,G.,Barman,T.S.,Selsis,F.,Allard,F.,Hauschildt,P.H.,2005.HotJupitersandhotNeptunes:Acommonorigin?Astron.Astro-phys.436,L47–L51.Barman,T.,Hauschildt,P.,Allard,F.,2005.Phase-dependentpropertiesofex-trasolarplanetatmospheres.Astrophys.J.632-2,1132–1139.Bordé,P.,Rouan,D.,Léger,A.,2003.ExoplanetdetectioncapabilityoftheCOROTspacemission.Astron.Astrophys.405,1137–1144.Bouchy,F.,Pepe,F.,Queloz,D.,2001.Fundamentalphotonnoiselimittoradialvelocitymeasurements.Astron.Astrophys.374,733–739.Charbonneau,D.,Brown,T.,Burrows,A.,Laughlin,G.,2006.Whenextra-solarplanetstransittheirparentstars.In:ProtostarsandPlanetsV,astro-ph/0603376.Chassefiére,E.,1996.Hydrodynamicescapeofhydrogenfromahotwater-richatmosphere:ThecaseofVenus.J.Geophys.Res.101,26039–26056.Cox,A.N.,2000.AllenAstrophysicalQuantities,fourthed.Springer-Verlag,NewYork,ISBN0-387-98746-0.p.382.DiBenedetto,G.P.,2005.Predictingaccuratestellarangulardiametersbythenear-infraredsurfacebrightnesstechnique.Mon.Not.R.Astron.Soc.357,174–190.Ehrenreich,D.,Tinetti,G.,LecavelierdesEtangs,A.,Vidal-Madjar,A.,Selsis,F.,2006.ThetransmissionspectrumofEarth-sizetransitingplanets.Astron.Astrophys.448,379–393.Fortney,J.J.,2005.Theeffectofcondensatesonthecharacterizationoftran-sitingplanetatmosphereswithtransmissionspectroscopy.Mon.Not.R.Astron.Soc.364,649–653.Fridlund,M.,Baglin,A.,Conroy,L.,Lochard,J.,2006.TheCoRoTBook.ESA-SP1306,Sect.VII,inpress(http://corotsol.obspm.fr/web-instrum/payload.param/).Griessmeier,J.-M.,Stadelmann,A.,Penz,T.,Lammer,H.,Selsis,F.,Ribas,I.,Guinan,E.F.,Motschmann,U.,Biernat,H.-K.,Weiss,W.W.,2004.Theeffectoftidallockingonthemagnetosphericandatmosphericevolutionof“hotJupiters.”Astron.Astrophys.425,753–762.Griessmeier,J.-M.,Stadelmann,A.,Motschmann,U.,Belisheva,N.K.,Lam-mer,H.,Biernat,H.K.,2005.CosmicrayimpactonextrasolarEarth-likeplanetsinclose-inhabitablezones.Astrobiology5,587–603.Kasting,J.F.,1988.Runawayandmoistgreenhouseatmospheresandtheevo-lutionofEarthandVenus.Icarus74,472–494.Kervella,P.,Thevenin,F.,DiFolco,E.,Segransan,D.,2004.Theangularsizesofdwarfstarsandsubgiants:Surfacebrightnessrelationscalibratedbyin-terferometry.Astron.Astrophys.426,297–307.Khodachenko,M.L.,Lammer,H.,Lichtenegger,H.I.M.,Langmayr,D.,Erkaev,N.V.,Griessmeier,J.-M.,Leitner,M.,Penz,T.,Biernat,H.,K.,Motschmann,U.,Rucker,H.O.,2007.Masslossof“hotJupiters”ImplicationsforCoRoTdiscoveries.PartI:Theimportanceofmagnetosphericprotectionofaplanetagainstionlosscausedbycoronalmassejections.Planet.SpaceSci.55(5),631–642.Koch,D.,and15colleagues,2006.TheKeplerMission:Astrophysicsandeclipsingbinaries.CloseBinariesinthe21thCenturySyros,Greece,27–30June2005.Astrophys.SpaceSci.J.,inpress.Kuchner,M.J.,2003.Volatile-richEarth-massplanetsintheHabitableZone.Astrophys.J.596,L105–L108.Kuchner,M.J.,Seager,S.,2005.Extrasolarcarbonplanets.Astrophys.J.,sub-mittedforpublication,astro-ph/0504214.Lammer,H.,Selsis,F.,Ribas,I.,Guinan,E.F.,Bauer,S.J.,Weiss,W.W.,2003.AtmosphericlossofexoplanetsresultingfromstellarX-rayandextreme-ultravioletheating.Astrophys.J.598,L121–L124.Lammer,H.,Lichtenegger,H.I.M.,Kulikov,Yu.N.,Griessmeier,J.-M.,Terada,N.,Erkaev,N.V.,Biernat,H.K.,Khodachenko,M.L.,Ribas,I.,Penz,T.,Selsis,F.,2006.CMEactivityoflowmassMstarsasanimportantfactorforthehabitabilityofterrestrialexoplanets.PartII.CMEinducedionpickupofEarth-likeexoplanetsinclose-inhabitablezones.Astrobiology,inpress.Léger,A.,Selsis,F.,Sotin,C.,Guillot,T.,Despois,D.,Lammer,H.,Ollivier,M.,Brachet,F.,2003.Anewfamilyofplanets?Ocean-planets.In:Fridlund,M.,Henning,T.(Eds.),ProceedingsoftheConference“TowardsOtherEarthsDarwin/TPFandtheSearchforExtrasolarTerrestrialPlanets.”In:ESA-SP539,pp.253–259.Léger,A.,Selsis,F.,Sotin,C.,Guillot,T.,Despois,D.,Mawet,D.,Ollivier,M.,Labèque,A.,Valette,C.,Brachet,F.,Chazelas,B.,Lammer,H.,2004.Anewfamilyofplanets,“ocean-planets”.Icarus169,499–504.Lovis,C.,and13colleagues,2006.AnextrasolarplanetarysystemwiththreeNeptune-massplanets.Nature441,305–309.Mayor,M.,and31colleagues,2003.SettingnewstandardswithHARPS.TheMessenger114,20–24.Moutou,C.,Pont,F.,Barge,P.,and13colleagues,2005.Comparativeblindtestoffiveplanetarytransitdetectionalgorithmsonrealisticsyntheticlightcurves.Astron.Astrophys.437,355–368.O’Donovan,F.T.,Barman,T.S.,Allard,F.,Hauschildt,P.H.,Lammer,H.,and17colleagues,2006.TrES-2:ThefirsttransitingplanetintheKeplerField.Astrophys.J.651,L61–L64.Papaloizou,J.C.B,Nelson,R.P.,Kley,W.,Masset,F.S.Artmowicz,P.,2006.Disk–planetinteractionsduringplanetformation.In:ProtostarsandPlanetsV.Univ.ofArizonaPress,Tucson.Penz,T.,Langmayr,D.,Erkaev,N.V.,Kulikov,Yu.N.,Lammer,H.,Biernat,H.K.,Selsis,F.,Cecchi-Pestellini,C.,Micela,G.,Barge,P.,Deleuil,M.,Léger,A.,2006.Masslossfrom“hotJupiters”—ImplicationsforCoRoTdiscoveries.PartII.Longtimethermalatmosphericevaporationmodeling.P&SS,submittedforpublication.Pepe,F.,Mayor,M.,Queloz,andcolleagues,2005.Onthetrackofverylow-massplanetswithHARPS.TheMessenger120,22–25.Press,W.,Teukolsky,S.,Vetterling,W.,Flannery,B.,2002.NumericalRecipesinC.CambridgeUniv.Press,Cambridge.Protanov,A.,2002.AnalyseStatistiquedesDonnéesExpérimentales.EDPSci-ence,France,ISBN2-86883-590-2,pp.51–57.Rafikov,R.R.,2006.Atmospheresofprotoplanetarycores:Criticalmassfornucleatedinstability.Astrophys.J.648,666–682.Raymond,S.N.,Mandell,A.M.,Sigurdsson,S.,2006a.ExoticEarths:Forminghabitableworldswithgiantplanetmigration.Science313,1413–1416.Raymond,S.N.,Quinn,T.,Lunine,J.I.,2006b.High-resolutionsimulationsofthefinalassemblyofEarth-likeplanets.I.Terrestrialaccretionanddynam-ics.Icarus183,265–282.Raymond,S.N.,Quinn,T.,Lunine,J.I.,2007.High-resolutionsimulationsofthefinalassemblyofEarth-likeplanets.II.Waterdeliveryandplanetaryhabitability.Astrobiology7(1),66–84.Ribas,I.,Guinan,E.F.,Güdel,M.,Audard,M.,2005.Evolutionofthesolaractivityovertimeandeffectsonplanetaryatmospheres.I.High-energyir-radiances(1–1700A).Astrophys.J.622,680–694.Rouan,D.,Baglin,A.,Barge,P.,Copet,E.,Deleuil,M.,Leger,A.,Schneider,J.,Toublanc,D.,Vuillemin,A.,1999.SearchingforexosolarplanetswiththeCOROTspacemission.Phys.Chem.EarthC24,567–571.Santos,N.C.,Pont,F.,Melo,C.,Israelian,G.,Bouchy,F.,Mayor,M.,Moutou,C.,Queloz,D.,Udry,S.,Guillot,T.,2006.Highresolutionspectroscopyofstarswithtransitingplanets:ThecasesofOGLE-TR-10,56,111,113andTrES-1.Astron.Astrophys.450,825–831.Sato,B.,Fischer,D.A.,Henry,G.W.,and18colleagues,2005.TheN2KCon-sortium.II.AtransitinghotSaturnaroundHD149026withalargedensecore.Astrophys.J.633,465–473.Sotin,C.,Grasset,O.,Moquet,A.,2007.Mass/radiusforextrasolarEarth-likeplanetsandocean-planets.Icarus,inpress.Terada,N.,Machida,S.,Shinagawa,H.,2002.GlobalhybridsimulationoftheKelvin–HelmholtzinstabilityattheVenusionopause.J.Geophys.Res.107,1471–1490.Terquem,C.,Papaloizou,J.C.B.,2006.Migrationandtheformationofsystemsofhotsuper-EarthsandNeptunes.AstrophysJ.,inpress,ArXivAstro-physicse-prints:astro-ph/0609779.Tian,F.,Toon,O.B.,Pavlov,A.A.,DeSterck,H.,2005.Transonichydrody-namicescapeofhydrogenfromextrasolarplanetaryatmospheres.Astro-phys.J.621,1049–1060. Pleasecitethisarticleinpressas:F.Selsisetal.,Couldweidentifyhotocean-planetswithCoRoT,KeplerandDopplervelocimetry?,Icarus(2007),doi:10.1016/j.icarus.2007.04.010JID:YICARAID:8261/FLA[m5+;v1.73;Prn:12/06/2007;8:22]P.16(1-16)16F.Selsisetal./Icarus•••(••••)•••–•••Valencia,D.,O’Connell,R.J.,Sasselov,D.,2006.Internalstructureofmassiveterrestrialplanets.Icarus181,545–554.Vidal-Madjar,A.,LecavelierdesEtangs,A.,Désert,J.-M.,Ballester,G.E.,Ferlet,R.,Hébrard,G.,Mayor,M.,2003.Anextendedupperat-mospherearoundtheextrasolarplanetHD209458b.Nature422,143–146.Vidal-Madjar,A.,Désert,J.-M.,LecavelierdesEtangs,A.,Hébrard,G.,Ballester,G.E.,Ehrenreich,D.,Ferlet,R.,McConnell,J.C.,Mayor,M.,Parkinson,C.D.,2004.Detectionofoxygenandcarboninthehydrody-namicallyescapingatmosphereoftheextrasolarplanetHD209458b.As-trophys.J.604,L69–L72.Watson,A.J.,Donahue,T.M.,Walker,J.C.G.,1981.Thedynamicsofarapidlyescapingatmosphere—ApplicationstotheevolutionofEarthandVenus.Icarus48,150–166.Wood,B.E.,Müller,H.-R.,Zank,G.P.,Linsky,J.L.,2002.Measuredmass-lossratesofsolar-likestarsasafunctionofageandactivity.Astrophys.J.574,412–425.Wood,B.E.,Müller,H.-R.,Zank,G.P.,Linsky,J.L.,Redfield,S.,2005.Newmass-lossmeasurementsfromastrosphericLyαabsorption.Astrophys.J.628,L143–L146.Yelle,R.V.,2004.Aeronomyofextrasolargiantplanetsatsmallorbitaldis-tances.Icarus170,167–179. |
0805.2108 | 5 | 0805 | 2009-04-12T18:47:59 | Transitional solar dynamics, cosmic rays and global warming | [
"astro-ph",
"nlin.CD",
"physics.geo-ph"
] | Solar activity is studied using a cluster analysis of the time-fluctuations of the sunspot number. It is shown that in an Historic period the high activity components of the solar cycles exhibit strong clustering, whereas in a Modern period (last seven solar cycles: 1933-2007) they exhibit a white-noise (non-)clustering behavior. Using this observation it is shown that in the Historic period, emergence of the sunspots in the solar photosphere was strongly dominated by turbulent photospheric convection. In the Modern period, this domination was broken by a new more active dynamics of the inner layers of the convection zone. Then, it is shown that the dramatic change of the sun dynamics at the transitional period (between the Historic and Modern periods, solar cycle 1933-1944yy) had a clear detectable impact on Earth climate. A scenario of a chain of transitions in the solar convective zone is suggested in order to explain the observations, and a forecast for the global warming is suggested on the basis of this scenario. A relation between the recent transitions and solar long-period chaotic dynamics has been found. Contribution of the galactic turbulence (due to galactic cosmic rays) has been discussed. These results are also considered in a content of chaotic climate dynamics at millennial timescales. | astro-ph | astro-ph |
Transitional solar dynamics, cosmic rays and global warming
ICAR, P.O. Box 31155, Jerusalem 91000, Israel
A. Bershadskii
Solar activity is studied using a cluster analysis of the time-fluctuations of the sunspot number.
It is shown that in an Historic period the high activity components of the solar cycles exhibit
strong clustering, whereas in a Modern period (last seven solar cycles: 1933-2007) they exhibit
a white-noise (non-)clustering behavior. Using this observation it is shown that in the Historic
period, emergence of the sunspots in the solar photosphere was strongly dominated by turbulent
photospheric convection. In the Modern period, this domination was broken by a new more active
dynamics of the inner layers of the convection zone. Then, it is shown that the dramatic change of
the sun dynamics at the transitional period (between the Historic and Modern periods, solar cycle
1933-1944yy) had a clear detectable impact on Earth climate. A scenario of a chain of transitions
in the solar convective zone is suggested in order to explain the observations, and a forecast for the
global warming is suggested on the basis of this scenario. A relation between the recent transitions
and solar long-period chaotic dynamics has been found. Contribution of the galactic turbulence
(due to galactic cosmic rays) has been discussed. These results are also considered in a content of
chaotic climate dynamics at millennial timescales.
PACS numbers: 92.70.Qr, 92.70.Mn, 96.60.qd, 98.70.Sa
INTRODUCTION
The sunspot number is the main direct and reliable
source of information about the solar dynamics for his-
toric period. This information is crucial, for instance, for
analysis of a possible connection between the sun activity
and the global warming. In a recent papers [1],[2] results
)
C
0
(
y
l
a
m
o
n
a
e
r
u
t
a
r
e
p
m
e
T
0.4
0.2
0
-0.2
-0.4
-0.6
C
T (land and marine)
B
A
I
II
III
300
250
200
150
100
50
N
S
S
1800
1830
1860
1890
1920
1950
1980
0
2010
Year
FIG. 1: Sunspot number (SSN, monthly) vs time [14]. The
dashed straight lines separate between periods of different in-
tensity of the solar activity. The solid curve shows the global
temperature anomaly (combined land and marine, 7-years
running average) [20].
of a reconstruction of the sunspot number were presented
for the past 11,400 years. The reconstruction shows that
[1]: "..the level of solar activity during the past 70 years
is exceptional, and the previous period of equally high
activity occurred more than 8,000 years ago" (section III
in Fig. 1) .
Figure 1 gives first indication ('by eye'), that the
prominent maxima of the global temperature data (solid
curve in the Fig. 1) correspond to transitions between pe-
riods of different intensity of the Sun activity (character-
ized by the monthly sunspot number-SSN). This observa-
tion can be considered as an indication of a strong impact
of the solar activity transitions on the global Earth cli-
mate. Therefore, understanding of the physical processes
in the Sun, which cause these activity transitions, seems
to be crucial for any serious forecast for global Earth cli-
mate.
SUNSPOTS
When magnetic field lines are twisted and poke
through the solar photosphere the sunspots appear as
the visible counterparts of magnetic flux tubes in the
convective zone of the sun. Since a strong magnetic field
is considered as a primary phenomenon that controls gen-
eration of the sunspots the crucial question is: Where has
the magnetic field itself been generated? The location of
the solar dynamos is the subject of vigorous discussions
in recent years. A general consensus had been developed
to consider the shear layer at the bottom of the convec-
tion zone as the main source of the solar magnetic field
[3] (see, for a recent review [4]). In recent years, however,
the existence of a prominent radial shear layer near the
top of the convection zone has become rather obvious and
the problem again became actual. The presence of large-
scale meandering flow fields (like jet streams), banded
zonal flows and evolving meridional circulations together
with intensive multiscale turbulence shows that the near
surface layer is a very complex system, which can sig-
nificantly affect the processes of the magnetic field and
the sunspots generation. There could be two sources for
the poloidal magnetic field: one near the bottom of the
convection zone (or just below it [3]), another resulting
from an active-region tilt near the surface of the convec-
tion zone. For the recently renewed Babcock-Leighton
[5],[6] solar dynamo scenario, for instance, a combina-
tion of the sources was assumed for predicting future so-
lar activity levels [7], [8].
In this scenario the surface
generated poloidal magnetic field is carried to the bot-
tom of the convection zone by turbulent diffusion or by
the meridional circulation. The toroidal magnetic field is
produced from this poloidal field by differential rotation
in the bottom shear layer. Destabilization and emer-
gence of the toroidal fields (in the form of curved tubes)
due to magnetic buoyancy can be considered as a source
of pairs of sunspots of opposite polarity. The turbulent
convection in the convection zone and, especially, in the
near-surface layer captures the magnetic flux tubes and
either disperses or pulls them trough the surface to be-
come sunspots.
The magnetic field plays a passive role in the photo-
sphere and does not participate significantly in the tur-
bulent photospheric energy transfer. On the other hand,
the very complex and turbulent near-surface layer (in-
cluding photosphere) can significantly affect the process
of emergence of sunspots. The similarity of light elements
properties in the spot umbra and granulation is one of the
indication of such phenomena.
The present paper reports a direct relation between
the fluctuations in sunspot number and the temperature
of photospheric turbulent convection in the Historic pe-
riod. This relation allows for certain conclusions about
the generation mechanisms of the magnetic fields and the
sunspots. In the Modern period (section III in Fig. 1,
cf.
[1],[2]) the relative role of the surface layer (photo-
sphere) in the process of emergence of sunspots decreased
in comparison with the Historic period (section II in Fig.
1), implying a drastic increase of the relative role of the
inner layers of the convection zone in the Modern period.
TIME-CLUSTERING OF FLUCTUATIONS
[10] (see also Ref.
In order to extract new information from sunspot num-
ber data we apply the fluctuation clustering analysis sug-
gested in the Ref.
[11]). For a time
depending signal we count the number of 'zero'-crossings
of the signal (the points on the time axis where the signal
is equal to zero) in a time interval τ and consider their
running density nτ . Let us denote fluctuations of the
running density as δnτ = nτ − hnτ i, where the brackets
2
/
2
1
>
2
nt
<d
1
0.1
0.01
1
-0.37
level SSN=85
10
100
1000
[d]
FIG. 2: The standard deviation for δnτ vs τ (Historic period)
in log-log scales. The straight line (the best fit) indicates the
scaling law Eq. (1).
2
/
1
>
2
nt
<d
1
0.1
0.01
1
-0.50
level SSN=125
10
100
1000
[d]
FIG. 3: The standard deviation for δnτ vs τ (Modern period)
in log-log scales. The straight line (the best fit) indicates the
scaling law Eq. (1).
mean the average over long times. We are interested in
scaling variation of the standard deviation of the running
density fluctuations hδn2
τ i1/2 with τ
hδn2
τ i1/2 ∼ τ −α
(1)
For white noise signal
it can be derived analytically
[12],[13] that α = 1/2 (see also [10]). The same con-
sideration can be applied not only to the 'zero'-crossing
points but also to any level-crossing points of the signal.
One can see that for the Modern period (section III in
Fig. 1) the solar activity is significantly different from
that for the Historic period (section II in Fig. 1). There-
fore, in order to calculate the cluster exponent (if exists)
for this signal one should make this calculation separately
for the Modern and for the Historic periods. We are in-
terested in the active parts of the solar cycles. Therefore,
t
t
3
0.1
0.01
2
/
1
>
2
n
<
-0.37
thermal convection (Ra=1.4* 1011)
0.001
0.01
0.1
1
10
100
1000
t [s]
1
0.1
2
/
1
>
2
n
δ
<
0.01
1
-0.37
Br - 1998-2008yy
10
100
1000
τ [h]
FIG. 4: The standard deviation for δnτ vs τ for the tempera-
ture fluctuations in the Rayleigh-Bernard convection labora-
tory experiment with Ra = 1.4 × 1011 [16]. The straight line
indicates the scaling law Eq. (1).
FIG. 5: The standard deviation for δnτ vs. τ for radial com-
ponent Br of the interplanetary magnetic field, as measured
by the ACE magnetometers for the last solar cycle (hourly
average [26]).
for the Historic period let us start from the level SSN=85.
The set of the level-crossing points has a few large voids
corresponding to the weak activity periods. In the tele-
graph signal Ref.
[10], corresponding to the data set,
the large voids became so clear detectable that there is
no problem to cut them off. Then, the remaining data
have been merged providing a statistically stationary set
(about 104 data points). The robustness of the procedure
has been successfully checked.
Fig. 2 shows (in the log-log scales) dependence of the
standard deviation of the running density fluctuations
hδn2
τ i1/2 on τ for this data set. The straight line is drawn
in this figure to indicate the scaling (1). The slope of
this straight line provides us with the cluster-exponent
α = 0.37 ± 0.02. This value turned out to be insensitive
to a reasonable variation of the SSN level. Results of
analogous calculations (including the large voids 'cut off'
procedure) performed for the Modern period are shown
in Fig. 3 for the SSN level SSN=125. The calculations
performed for the Modern period provide us with the
cluster-exponent α = 0.5 ± 0.02 (and again this value
turned out to be insensitive to a reasonable variation of
the SSN level).
The exponent α ≃ 0.5 (for the Modern period) indi-
cates a random (white noise like) situation. While the
exponent α ≃ 0.37 (for the Historic period) indicates
strong clustering. The question is: Where is this strong
clustering coming from? It is shown in the paper [10]
that turbulence produces signals with strong clustering.
Moreover, the cluster exponents for these signals depend
on the turbulence intensity and they are nonsensitive to
the types of the boundary conditions. Fortunately, we
have direct estimates of the value of the main parame-
ter characterizing intensity of the turbulent convection
in photosphere: Rayleigh number Ra ∼ 1011 (see, for in-
stance [15]). In Fig. 4 we show calculation of the cluster
exponent for the temperature fluctuations in the classic
Rayleigh-Bernard convection laboratory experiment for
Ra ∼ 1011 (for a description of the experiment details
see [16]). The calculated value of the cluster exponent
α = 0.37 ± 0.01 coincides with the value of the cluster-
exponent obtained above for the sunspot number fluctu-
ations for the Historic period. The value of the Rayleigh
number Ra in the photosphere for the Historic period
has the same order as for the Modern period: Ra ∼ 1011
(see next Section). Therefore, the photospheric temper-
ature fluctuations can produce the strong clustering of
the sunspot number fluctuations for the Historic period.
This seems to be natural for the case when the photo-
spheric convection determines the sunspot emergence in
the photosphere. However, in the case when the effect of
the photospheric convection on the SSN fluctuations is
comparable with the effects of the inner convection zone
layers on the SSN fluctuations the clustering should be
randomized by the mixing of the sources, and the cluster
exponent α ≃ 0.5 (similar to the white noise signal). The
last case apparently takes place for the modern period.
Since the Rayleigh number Ra of the photospheric con-
vection preserves its order Ra ∼ 1011 with transition
from the Historic period to the Modern one (see next
Section), we can assume that just significant changes of
the dynamics of the inner layers of the convection zone
(most probably - of the bottom layer) were the main rea-
sons for the transition from the Historic to the Modern
period.
d
t
4
Cumulated aa-index
1000000
800000
600000
400000
200000
a
a
S
1
0.1
2
/
1
>
2
τ
n
δ
<
-0.37
0
1860 1880 1900 1920 1940 1960 1980 2000
t [years]
1875-1932yy
1933-2007yy
0.01
1
10
100
1000
τ [d]
FIG. 6: Cumulative aa-index vs. time. Daily aa-index was
taken from [30]. The arrow indicates beginning of the transi-
tional solar cycle.
FIG. 7: The standard deviation for δnτ vs. τ (in log-log
scales) for the daily aa-index [30]: circles correspond to the
Historic period and crosses correspond to the Modern period.
MAGNETIC FIELDS IN THE SOLAR WIND
AND ON EARTH
Although for the Modern period the turbulent con-
vection in the photosphere has no decisive impact on
the sunspots emergence, the large-scale properties of the
magnetic field coming through the sunspots into the pho-
tosphere and then to the interplanetary space (so-called
solar wind) can be strongly affected by the photospheric
motion. In order to be detected the characteristic scal-
ing scales of this impact should be larger than the scal-
ing scales of the interplanetary turbulence (cf. [24],[25]).
In particular, one can expect that the cluster-exponent
of the large-scale interplanetary magnetic field (if ex-
ists) should be close to α ≃ 0.37.
In Fig. 5 we show
cluster-exponent of the large-scale fluctuations of the ra-
dial component Br of the interplanetary magnetic field.
For computing this exponent, we have used the hourly
averaged data obtained from Advanced Composition Ex-
plorer (ACE) satellite magnetometers for the last so-
lar cycle [26]. As it was expected the cluster-exponent
α ≃ 0.37 ± 0.02. Analogous result was obtained for other
components of the interplanetary magnetic field as well.
At the Earth itself the solar wind induced activity is
measured by geomagnetic indexes such as aa-index (in
units of 1 nT). This, the most widely used long-term ge-
omagnetic index [27], presents long-term geomagnetic ac-
tivity and it is produced using two observatories at nearly
antipodal positions on the Earth's surface. The index is
computed from the weighted average of the amplitude
of the field variations at the two sites. It was recently
discovered that variations of this index are strongly cor-
related with the global temperature anomalies (see, for
instance, [28],[29]).
In this paper, however, we will be
mostly interested in the clustering properties of the aa-
index and their relation to the modulation produced by
the photospheric convective motion. The point is that
the data for the aa-index are available for both the Mod-
ern and the Historic periods. This allows us to check the
suggestion that the Rayleigh number of the photospheric
convection has the same order for the both mentioned
periods. The transition between the two periods can be
seen in Figure 6, where we show the cumulated aa-index
a(t)
Saa(t) = Z t
0
a(t′)dt′
(2)
The arrow in this figure indicates beginning of the tran-
sitional solar cycle.
Figure 7 shows cluster-exponents for both the Modern
and the Historic periods calculated for the low intensity
levels of the aa-index (for the Historic period the level
used in the calculations is aa-index=10nT, whereas for
the Modern period the level is aa-index=25nT, the aa-
index was taken from [30]). The low levels of intensity
were chosen in order to avoid effect of the extreme phe-
nomena (magnetic storms and etc.). The straight line in
Fig. 7 is drawn to show the expected value of the cluster-
exponent α ≃ 0.37 for the both periods. It means that
indeed for both the Historic and the Modern periods the
Rayleigh number Ra ∼ 1011 (cf. Fig. 4) in the solar
photosphere.
IMPACT OF THE SOLAR DYNAMICS ON
EARTH CLIMATE
The transitions of the solar convection zone dynamics
can affect the earth climate through (at least) two
channels. First channel is a direct change in the heat
and light output of the Sun, especially during the
)
l
C
0
(
y
a
m
o
n
a
e
r
u
t
a
r
e
p
m
e
T
T (land and marine)
CO2 (Law Dome)
B
A
0.1
0
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
N
S
S
340
330
320
310
300
290
280
)
m
p
p
(
2
O
C
1850 1870 1890 1910 1930 1950 1970
Year
FIG. 8: Global temperature anomaly (solid curve, combined
land and marine) [20] and atmospheric CO2 (circles) [21] vs
time. Relevant daily SSN are also shown.
)
m
p
p
(
2
O
C
308
306
304
302
300
298
296
294
1900-1930yy (Law Dome)
1977- 2008yy (Mauna Loa)
0
0
5
5
10
10
15
15
20
20
25
25
30
30
t [y]
390
380
370
360
350
340
330
35
35
FIG. 9: Comparison of the CO2 growth for the periods 1900-
1930yy [21] and 1977-2008yy [22]. The solid straight lines
(best fit) indicate a linear growth.
5
)
C
0
(
y
l
a
m
o
n
a
e
r
u
t
a
r
e
p
m
e
T
1975
0
1980
1985
1990
1995
2000
2005
2010
0.5
1912-1942yy
combined land and marine
-0.1
-0.2
-0.3
-0.4
0.4
0.3
0.2
0.1
0
1977- 2007yy
-0.5
1910
1915
1920
1925
1930
1935
1940
-0.1
1945
Year
FIG. 10: Comparison of the temperature anomaly (combined
land and marine) [20] for the periods 1912-1942yy and 1977-
2007yy.
8 shows global temperature anomaly (the solid curve)
[20] and atmospheric CO2 mixing ratios (the circles)
[21] vs.
time. Relevant SSNs are also shown. The
dashed straight line (corresponding to the peak B)
separates between Historic and Modern periods. One
can see, that the transitional solar cycle (1933-1944yy)
is characterized by dramatic changes both in the global
temperature (a huge peak) and in the atmospheric
CO2 (complete suppression of the growth).
After
the transitional period one can observe (during the
three solar cycles: 1944-1977yy) certain growth in the
temperature anomaly and an unusually fast growth in
the atmospheric CO2. These three solar cycles seems
to be an aftershock adaptation of the global climate to
the new conditions. Then, starting from 1977 year, the
atmospheric CO2 growth returns to its linear trend as
before the transition B, but now the rate of the growth
is about four times larger than before the transition (see
Fig. 9). On the other hand, it can be seen from Fig. 1
that the temperature anomaly before the maximum C
returns to about the same growth pattern as it was just
before the transition B. A quantitative comparison of
these patterns, shown in Figure 10, indicates only about
20% difference in the growth rate (cf. also Refs.
[31],
[32]). The thirty year periods used for this comparison
could find a support in the Appendix B.
transitional solar cycle 1933-1944yy. Second channel
is related to the strong increase of the magnetic field
output in the interplanetary space through the sunspots.
The interplanetary magnetic field interacts with the
cosmic rays. Therefore, the change in the magnetic
field intensity can affect the Earth climate through the
change of the cosmic rays intensity and composition
(see,
[17],[18],[19]). Let us look
more closely to the transition B in Fig. 1. The Fig.
for instance, Refs.
If one consider the impact mechanism, related to the
energetic cosmic (charged) particles and the correspond-
ing screen effect of the magnetic fields, one could expect
that this impact mechanism should be also sensitive to
the 11-year solar cycle variability (though considerably
less than to the transitional effects). To find fingerprints
of this variability in the global temperature anomaly data
is not a trivial task due to comparatively (to the 11-
year cycle) short period of observations and due to the
11-year cycle
0.025
0.02
0.015
0.01
0.005
E
global temperature anomaly
sunspot number
0.1
0.2
0.3
0.4
0.5
0
0
f [y-1]
0.03
0.025
0.02
E
0.015
0.01
0.005
0
0
6
11- year cycle
AR spectrum
0.1
0.2
0.3
0.4
0.5
f [y-1]
FIG. 11: The solid curve corresponds to energy spectrum
of the fluctuating (noise-like) component of the decomposed
global temperature anomaly time series (combined land and
marine [20]). The dashed curve corresponds to energy spec-
trum of fluctuating component of the SSN data.
FIG. 12: Energy spectrum of the AR (fluctuating) component
of the decomposed global temperature anomaly time series
(combined land and marine [20]).
statistically non-stationary character of these data (with
considerable time trends). To extract these fingerprints
we will use a time series decomposition method devel-
oped in [33],[34],[35]. A time series is decomposed into a
trend (like that shown in Fig. 1 as the solid curve) and
a fluctuating (noise-like) components (see Appendix A).
The method applies state space modeling and Kalman
filter. Parameters are estimated by maximum likelihood
method. Obtained by this way the fluctuating (noise-
like) component of the (filtered) signal presents a sta-
tistically stationary time series and, therefore, it can be
subject of a spectral analysis. Result of this analysis
for the global temperature anomaly (combined land and
marine [20]) is shown in Figure 11. Although we have
deal with a comparatively short data series one can clear
see the main spectral peak corresponding to the (about)
11-year solar cycle impact (cf. Refs.
If
one also introduces an autoregressive (AR) component
into the time series decomposition (see Appendix A) to
describe the fluctuating part of the global temperature
anomaly, then one can compute energy spectrum of the
AR component. Such spectrum is shown in figure 12.
One can clear see the spectral peak corresponding to the
solar 11-year cycle impact.
[18],[37],[38]).
It is also interesting to look at the reduction in arc-
tic ice extent. Figure 13 shows a trend component of
the anomalies of ice August extent in the Chukchi sea
(the data are taken from [36]) obtained by the time se-
ries decomposition. The dashed straight line indicates
beginning of the transitional solar cycle. Insert to Fig.
13 shows spectrum of corresponding fluctuating (noise-
like) component of the (filtered) signal (cf. Fig. 11).
The dashed straight line indicates beginning of the tran-
sitional solar cycle. This arctic sea is located sufficiently
t
n
e
t
x
e
e
c
i
f
o
)
2
m
k
0
0
0
1
*
(
s
e
i
l
a
m
o
n
A
80
60
40
20
0
-20
-40
-60
1900
E
3000
2000
1000
0
11- year cycle
0
0.1
0.2
f
0.3
0.4
1933 y
1920
1940
1960
1980
2000
Year
FIG. 13: Trend of anomalies of ice August extent in the
Chukchi sea [36]. The insert shows spectrum of fluctuating
(noise-like) component of the time series decomposition.
far from the North Atlantic to diminish its influence,
whereas influence of the Pacific ocean on the Chukchi
sea local climate is not very significant.
A FORECAST
At present time we can already observe a peak C in the
global temperature anomaly (see Fig. 1). The last obser-
vation can be considered as an indication of the current
transition to a new section IV in the solar activity. The
fact that in the 'last' period (1977-2007yy) the temper-
ature anomaly growth returned to the same pattern as
just before the transition peak B (see Fig. 10) provides
an additional support to this suggestion.
)
l
C
0
(
y
a
m
o
n
a
e
r
u
t
a
r
e
p
m
e
T
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
observed
ARMA predicted
I
II
III
IV
1860
1890
1920
1950
1980
2010
Year
250
200
150
100
50
0
N
S
S
7
11 year cycle
-5/3
0
-2
-4
-6
E
n
l
-8
-6
-5
-4
-3
ln f [y-1]
-2
-1
FIG. 14: The upper solid curve corresponds to the observed
yearly global temperature anomaly data (7-year smoothed)
[20]. The lower solid curve corresponds to the observed SSN
data (year-smoothed) [14]. The dashed curves are an ARMA-
prediction: for SSN and for global temperature anomaly.
FIG. 16: Spectrum of the galactic cosmic ray count rate fluc-
tuations in the ln-ln scales (the reconstructed data for pe-
riod 1611-2007yy have been taken from [44]). The straight
dashed line is drawn to indicate the Kolmogorov scaling law
E(f ) ∼ f (−5/3).
E
n
l
2
0
-2
-4
-6
-8
-10
-12
-6
11 - year cycle
-5/3
-5
-4
-3
ln f [y-1]
-2
-1
FIG. 15: ARMA spectrum of the global temperature anomaly
fluctuations in the ln-ln scales. The straight dashed line is
drawn to indicate the Kolmogorov scaling law E(f ) ∼ f (−5/3).
To make a forecast for the new section IV let us as-
sume that there was a storage of certain part of the en-
ergy coming from the radiative zone during the period of
the section II (Fig. 1). Then, in the period of the section
III this stored energy was released and transported to
the Sun surface. This release was not in the form of the
thermal convection, but in the form of strongly concen-
trated toroidal magnetic field structures, which results in
the intense emergence of the sunspots in the active parts
of the Modern period (section III in Fig. 1). It is inter-
esting that the storage and release periods have about
the same length: ≃ 60 years (see also Appendix B). The
storage of energy in a growing magnetic field at the bot-
tom of the convection zone or in the overshoot (by the
convective downflows) area just under the bottom can be
considered as the most plausible for this scenario. Con-
vective overshooting provides means to store magnetic
energy below the convection zone by resisting the buoy-
ancy effect for large time scales. Level of subadiabaticity
in the overshoot layer, needed in order to store the mag-
netic flux for sufficiently long times, can be significant.
Expected intensity of 105 G results an energy density
that is in order of magnitude larger than equipartition
value. Though, suppression of convective motion by the
magnetic flux tubes themselves can be considered as an
additional factor allowing to achieve the sufficient level
of subadiabaticity (see, for instance [3],[4],[23]).
Since the section III in the solar activity was a
it is naturally to assume that the next
release one,
section (IV) in the solar activity will be a 'storage'
one, similar to the 'storage' section II. As for the
corresponding Earth temperature anomaly, the section
IV is also expected to be similar to the section II. But
the temperature anomaly from the beginning of the
section IV will be shifted upward on about 0.6oC in
comparison with the section II. It then will be not
surprising if the total length of the section IV also will
be about 60 years (cf. Appendix B) and the section
will be finished with a maximum D higher than 0.6oC.
A very schematic presentation for a first part of the
section IV is shown in figure 14 (see Appendix A for an
Auto-Regressive-Moving-Average (ARMA) model used
for the prediction, cf. also Ref.
[39]). Figure 15 shows
corresponding ARMA model spectrum of the global
temperature anomaly fluctuations (in ln-ln scales). A
improving the paper. A software provided by K. Yosh-
ioka was used at the model computations.
11 year cycle
8
E
n
l
4
2
0
-2
-4
-6
-8
-10
-6
-5/3
0.4
0.2
0
-0.2
-0.4
-0.6
observed
ARM A predicted
1900
1940
1980
2020
-5
-4
-3
-2
-1
ln f
APPENDIX A
The above applied time series decomposition method
was developed in the Refs. [33],[34],[35]. In this method a
nonstationary mean time series is decomposed into trend,
autoregressive and noise components. A random process
Xt in discrete time t is defined by the following expres-
sion:
Xt = τt + εt + AR(m)
t
(A1)
where τt is a trend component, εt is an observational
noise component, and
AR(m)
t =
m
Xi=1
aiXt−i + η(1)
t
(A2)
is a globally stationary autoregressive (AR) component
of order m (ai are the coefficients of a recursive filter,
and η(1)
is a Gaussian white noise).
t
A k-th order stochastically perturbed difference equa-
tion describes evolution of the trend component
∆kτt = η(2)
t
(A3)
where η(2)
t
is a Gaussian white noise.
In the Ref. [35] algorithms, using state space modeling
and Kalman filter, are suggested for estimation of the
model parameters by the maximum likelihood method.
Using
so-called
autoregressive moving
average
(ARMA) model one could try to make a prediction. For
this model a random process Xt in discrete time t is
defined by the following expression:
Xt = εt + AR(m)
t +
n
Xi=1
biεt−i
(A4)
where εt
is Gaussian white noise with zero mean
[33],[34],[35]. Again, state space modeling and Kalman
filter, are used for estimation of the model parameters
by the maximum likelihood method. Figure 14 shows
an optimistic prediction produced with this model for
a first part of the section IV. Though, continuation of
the analogy to the section II makes the scenario much
less optimistic. A quasi-Newton method and dynamic
restrictions were used for optimization of the model pa-
rameters.
We can also play with the model by cutting the ob-
served time series. This procedure reduces relative con-
tribution of the low-frequency fluctuations, presumably
of Galactic origin (cf. Fig. 17 and Fig. 15). Then, the
insert to the Fig. 17 shows less steep decline of the global
FIG. 17: Spectrum of the global temperature anomaly fluc-
tuations in the ln-ln scales for a cutted observed base ARMA
model. The straight dashed line is drawn to indicate the Kol-
mogorov scaling law E(f ) ∼ f (−5/3).
dashed straight line in this figure indicates a scaling
with '-5/3' exponent: E(f ) ∼ f −5/3. Although, the
scaling interval is short, the value of the exponent is
rather intriguing. This exponent is well known in the
theory of fluid (plasma) turbulence and corresponds to
so-called Kolmogorov's cascade process (see, for instance
[40]). This process is very universal for turbulent fluids
and plasmas [41],[42]. For turbulent processes in Earth
and in Heliosphere the Kolmogorov-like spectra with
such large time scales cannot exist. Therefore, one
should think about a Galactic origin of Kolmogorov
turbulence (or turbulence-like processes [43]) with such
large time scales. This is not surprising if we recall
possible role of the galactic cosmic rays for Earth climate
(see, for instance,
In order to support
this point we show in figure 16 spectrum of galactic
cosmic ray intensity at the Earth's orbit (reconstruction
for period 1611-2007yy [44], cf. also Ref.
[24]). One
can compare Fig. 16 with Figs. 15 corresponding to
the global temperature anomaly fluctuations (see also
Appendix C). Presence of the turbulent-like component
in the low-frequency fluctuations makes any long-range
prediction a very difficult task (see also Appendix B on
a chaotic element in solar dynamics). A chaotic element
in global climate dynamics itself (Appendix C) presents
another problem for the forecast. Such internal forcing
agents as volcanoes and anthropogenic greenhouse gases,
for instance, have a potential to change drastically the
future dynamics of the global climate.
[17],[18],[19]).
The author is grateful to J.J. Niemela, to K.R. Sreeni-
vasan, and to SIDC-team, World Data Center for the
Sunspot Index, Royal Observatory of Belgium for shar-
ing their data and discussions. The author is also grateful
to C.H. Gibson and to R.A. Treumann for their help in
temperature anomaly at this (cutted observed base) ver-
sion of the ARMA model prediction. This game has also
another end. One can assume that using a more long
time series (observed base), than that we have used in
computing the results shown in Fig. 14, one could ob-
tain even more optimistic prediction for the first part of
the section IV.
APPENDIX B
The long-range reconstructions of the sunspot number
fluctuations (see, for instance, Refs.
[1],[2]) allow us to
look on the solar transitional dynamics from a more gen-
eral point of view. In figure 18 we show a spectrum of
such reconstruction for the last 11,000 years (the data,
used for computation of the spectrum,
is available at
[44]). The spectrum was computed using the autore-
gressive model (with maximum entropy method [35]). A
semi-logarithmical representation was used in the figure
to show an exponential law
E(f ) ∼ e−f /fe
(B1)
The straight line is drawn in Fig. 18 to indicate the expo-
nential law Eq. (B1). Slope of the straight line provides
us with the characteristic time scale Te = 1/fe ≃ 176±7y.
The exponential decay of the spectrum excludes the pos-
sibility of random behavior and indicates the chaotic be-
havior of the time series [45],[46]. It is well known that
low-order dynamic (deterministic) systems have as a rule
exponential decay of E(f ) (see, for instance, [46],[47]).
As for infinite dimensional dynamic systems with chaotic
attractors it is interesting to compare Fig. 18 with fig-
ure 3 of the Ref. [48]. It should be noted that the 176y
period is the third doubling of the period 22y. The 22y
period corresponds to the Sun's magnetic poles polarity
switching (see also Appendix C).
The exponential spectrum can be also produced by a
series of Lorentzian pulses with the average width of the
individual pulses equals to τ (though, the distribution of
widths of the pulses should be fairly narrow to result in
the exponential spectrum).
In Fig. 18 a local maximum corresponding to the fre-
quency fe and its first harmonics have been indicated
by arrows.
It should be noted that the harmonic 2fe
corresponds to the well known period T ≃ 88y (see, for
instance, Ref.
[49]), and the harmonic 3fe corresponds
to the period T ≃ 60y (cf. section "A FORECAST").
APPENDIX C
It is also useful to consider the above results in content
of multicentennial and millennial timescales. Figure 19
shows a reconstruction of Northern Hemisphere tempera-
tures for the past 2,000 years (the data for this figure were
9
8
6
4
2
0
-2
0
E~exp(-f/fe)
fe
2fe
3fe
0.1
0.2
0.3
0.4
f [(10y)-1]
)
f
(
E
n
l
FIG. 18: Spectrum of the sunspots number fluctuations in
the ln-linear scales (the reconstructed data for the last 11,000
years have been taken from [44]). The straight line is drawn
to indicate the exponential law Eq. (B1).
1-1980yy
0.5
0
-0.5
-1
)
C
0
(
l
y
a
m
o
n
a
e
r
u
t
a
r
e
p
m
e
T
-1.5
0
500
1000
Year
1500
2000
FIG. 19: A reconstruction of Northern Hemisphere temper-
ature anomaly for the past 2,000 years (the data was taken
from Ref. [50]).
taken from Ref.
[50]). This multi-proxy reconstruction
was performed by the authors of Ref. [51] using combina-
tion of low-resolution proxies (lake and ocean sediments)
with comparatively high-resolution tree-ring data. Fig-
ure 20 shows a power spectrum of the data set calculated
using the maximum entropy method (as for the spectra
shown in Figs. 16 and 18) in the frames of the autore-
gressive model, because it provides an optimal spectral
resolution even for small data sets (see also [47]). The
spectrum exhibits a rather wide peak indicating a peri-
odic component with a period around 22 y, and a broad-
band part with exponential decay. A semilogarithmical
plot was used in Fig. 20 (cf Fig. 18) in order to show
the exponential decay more clearly (at this plot the ex-
2
0
-2
-4
-6
)
f
(
E
n
l
22y
E~exp-f/fe
Te=1/fe=11y
0
0.05
0.1
0.15
f [y-1]
0.2
0.25
0.3
10
)
f
(
E
n
l
1
0
-1
-2
-3
-4
-5/3
22y
-6
-5
-4
ln f [y-1]
-3
-2
FIG. 20: Spectrum of the data, shown in Fig. 19, in semilog-
arithmical scales. The straight line indicates the exponential
decay Eq. (B1).
FIG. 22: Spectrum of the data, shown in Fig. 19, in ln-ln
scales. The dashed straight line indicates the Kolmogorov-
like spectrum: E(f ) ∼ f −5/3. The high-frequency part has
been cutted in order to show the low-frequency part.
)
f
(
E
n
l
10
5
0
-5
-10
-15
-20
0
0.1
0.2
0.3
0.4
0.5
f
FIG. 21: Spectrum of the chaotic fluctuations of the x- com-
ponent for the Rossler system (a = 0.15, b = 0.20, c = 10.0.)
ponential decay corresponds to a straight line). Both
stochastic and deterministic processes can result in the
broad-band part of the spectrum, but the decay in the
spectral power is different for the two cases. The ex-
ponential decay indicates that the broad-band spectrum
for these data arises from a deterministic rather than a
stochastic process. Indeed, for a wide class of determin-
istic systems a broad-band spectrum with exponential
decay is a generic feature of their chaotic solutions (see,
for instance, Refs. [45],[47],[48] and Appendix B).
In order to illustrate this and another significant fea-
ture of the chaotic power spectra we show in figure 21 a
power spectrum for the Rossler system [52]
dx
dt
= −(y + z);
dy
dt
= x + ay;
dz
dt
= b + xz − cz (C1)
chaotic solution, where a, b and c are parameters.
In
this figure one can see a typical picture: a narrow-band
peak (corresponding to the fundamental frequency of the
system) in a low-frequency part and a broad-band expo-
nential decay in a high-frequency part of the spectrum.
Nature of the exponential decay of the power spectra
of the chaotic systems is still an unsolved mathemat-
ical problem. A progress in solution of this problem
has been achieved by the use of the analytical contin-
uation of the equations in the complex domain (see, for
instance, [53],[54]). In this approach the exponential de-
cay of chaotic spectrum is related to a singularity in the
plane of complex time, which lies nearest to the real axis.
Distance between this singularity and the real axis deter-
mines the rate of the exponential decay. If parameters of
the dynamical system periodically fluctuate around their
mean values, then at certain (critical) intensity of these
fluctuations an additional singularity (nearest to the real
time axis) can appear. Distance between this singularity
and the real axis is determined by period of the peri-
odic fluctuation of the system's parameters. Therefore,
exponential decay rate of the broad-band part of the sys-
tem spectrum equals the period of the parametric forcing.
The chaotic spectrum provides two different characteris-
tic time-scales for the system: a period corresponding to
fundamental frequency of the system, Tf un, and a period
corresponding to the exponential decay rate, Te = 1/fe
(cf Eq. (B1)). The fundamental period Tf un can be esti-
mated using position of the low-frequency peak, while the
exponential decay rate period Te = 1/fe can be estimated
using the slope of the straight line of the broad-band part
of the spectrum in the semilogarithmical representation
(Figs. 20 and 21). From Fig. 20 we obtain Tf un ≃ 22±2y
and Te ≃ 11 ± 1y (the estimated errors are statistical
ones). Thus, the solar activity (SSN) period of 11 years
is still a dominating factor in the chaotic temperature
11
fluctuations at the millennial time scales , although it is
hidden for linear interpretation of the power spectrum
(cf Ref.
[55]). In the nonlinear interpretation the addi-
tional period Tf un ≃ 22y might correspond to the funda-
mental frequency of the underlying nonlinear dynamical
system. It is surprising that this period is close to the
22y period of the Sun's magnetic poles polarity switching
(cf Appendix B). It should be noted that the authors of
Ref.
[56] found a persistent 22y cyclicity in sunspot ac-
tivity, presumably related to interaction between the 22y
period of magnetic poles polarity switching and a relic
solar (dipole) magnetic field. Therefore, one cannot rule
out a possibility that the broad peak, in a vicinity of fre-
quency corresponding to the 22y period, is a quasi-linear
response of the global temperature to the weak periodic
modulation by the 22y cyclicity in sunspot activity. I.e.
strong enough periodic forcing results in the non-linear
(chaotic) response whereas a weak periodic forcing results
in a quasi-periodic response.
Finally, figure 22 shows the same power spectrum as in
Fig. 20 but in ln-ln scales, where the dashed straight line
indicates the Kolmogorov-like behavior: E(f ) ∼ f −5/3,
in the lower-frequency part of the spectrum (see section
'A Forecast' and Fig. 16 for relation between this type
of spectrum and galactic turbulence).
[12] G. Molchan, private communication.
[13] M.R. Leadbetter and J.D. Gryer, The variance of the
number of zeros of a stationary normal process, Bull.
Amer. Math. Soc. 71, 561-563 (1965).
[14] The
data
are
available
at
http://sidc.oma.be/sunspot-data/
[15] R.J. Bray, R.E. Loughead, and C.J Durrant, The so-
lar granulation (Cambridge Univ. Press, Cambridge, 2ed,
1984).
[16] J.J. Niemela, L. Skrbek, K.R. Sreenivasan, R. J. Don-
nelly, Turbulent convection at very high Rayleigh num-
bers, Nature, 404, 837-840 (2000).
[17] I. G. Usoskin and G. A. Kovaltsov, Cosmic Ray Induced
Ionization in the Atmosphere: Full Modeling and Practi-
cal Applications, J. Geophys. Res., 111, D21206, (2006).
[18] N.J. Shaviv, On climate response to changes in the cos-
mic ray flux and radiative budget, J. Geophys. Res. 110,
A08105 (2005).
[19] J. Kirkby, Cosmic Rays and Climate, Surveys in Geo-
physics, 28, 333-375 (2007) (see also arXiv:0804.1938).
[20] The data are available at http://www.cru.uea.ac.uk/cru/
data/temperature/ (N.A. Rayner, et al., Improved anal-
yses of changes and uncertainties in marine temperature
measured in situ since the mid-nineteenth century: the
HadSST2 dataset, J. Climate, 19, 446-469 (2006)).
[21] The data are available at http://cdiac.ornl.gov/trends/
co2/lawdome-data.html (D.M. Etheridge, et al., Histori-
cal CO2 records from the Law Dome DE08, DE08-2, and
DSS ice cores).
[22] The data
are
available
at http://cdiac.ornl.gov/
ftp/trends/co2/maunaloa.co2 (C. D. Keeling, and T.P.
Whorf. Atmospheric CO2 records from sites in the SIO
air sampling network).
[1] S. K. Solanki, I. G. Usoskin, B. Kromer, M. Schussler
and J. Beer, Unusual activity of the Sun during recent
decades compared to the previous 11,000 years, Nature,
431, 1084-1087 (2004).
[2] I.G. Usoskin, S.K. Solanki, M. Schussler,
et al.,
Millennium-Scale Sunspot Number Reconstruction: Ev-
idence
the
1940s, Phys. Rev. Lett. 91, 211101 (2003) (see also
arXiv:astro-ph/0310823).
an Unusually Active Sun since
for
[3] E.A. Spiegel, and N.O. Weiss, Magnetic activity and vari-
ations in solar luminosity, Nature, 287, 616-617 (1980)
[4] A. Brandenburg, The Case for a Distributed Solar Dy-
namo Shaped by Near-Surface Shear, ApJ, 625 539-547
(2005) (see also arXiv:astro-ph/0502275).
[5] H.W. Babcock, The topology of the suns magnetic field
and the 22-year cycle, ApJ. 133 572-589 (1961).
[6] R.B. Leighton, A Magneto-Kinematic Model of the Solar
Cycle, ApJ, 156, 1-26 (1969).
[7] M. Dikpati, G.de Toma, and P.A. Gilman, Predicting the
strength of solar cycle 24 using a flux-transport dynamo-
based tool, Geophys. Res. Lett., 33 L05102 (2006).
[8] A.R. Choudhuri, P. Chatterjee, and J. Jiang, Predicting
Solar Cycle 24 With a Solar Dynamo Model, Phys. Rev.
Lett. 98, 131101 (2007).
[9] R.J. Bray, R.E. Loughhead, Sunspots, (Dover Publica-
[23] S.M. Tobias and D.W. Hughes, The Influence of Veloc-
ity Shear on Magnetic Buoyancy Instability in the Solar
Tachocline, ApJ. 603, 785802 (2004).
[24] A. Bershadskii, Multiscaling of Galactic Cosmic Ray
Flux, Phys. Rev. Lett., 90 041101 (2003) (see also
arXiv:astro-ph/0305453).
[25] M.L. Goldstein, Major Unsolved Problems in Space
Plasma Physics, Astrophys. Space Sci. 227, 349369
(2001).
[26] The data are available at http://www.srl.caltech.edu/
ACE/ASC/
[27] P.N. Mayaud, Derivation, Meaning, and Use of Geo-
magnetic Indices, AGU 1129 Geophys. Monograph 22,
(Washington D.C., 1980).
[28] E. W. Cliver, V. Boriakoff, J. Feynman, Solar Variabil-
ity and Climate Change: Geomagnetic aa Index and
Global Surface Temperature, J. Geophys. Res. Lett. 25,
10351038 (1998).
[29] T. Landscheidt, in "The solar cycle and terrestrial cli-
mate"', Vazquez, M. and Schmiedere, E, ed., European
Space Agency, Special Publication, 463, pp 497-500
(2000).
[30] The data are available at http://www.ukssdc.ac.uk/
data/wdcc1/wdc menu.html (World Data Centre for
Solar-Terrestrial Physics, Chilton).
tions, New York, 1979).
[31] C. Fabara
[10] K.R. Sreenivasan and A. Bershadskii, Clustering Proper-
ties in Turbulent Signals, J. Stat. Phys., 125, 1141-1153
(2006).
[11] A. Bershadskii, Transitional dynamics of the solar con-
vection zone, Europhys. Lett., 85, 49002 (2009).
and B. Hoeneisen, Global Warm-
calculations,
back-of-the-envelope
ing:
arXiv:physics/0503119 (2005).
some
[32] G. Gerlich and R. D. Tscheuschner, Falsification Of The
Atmospheric CO2 Greenhouse Effects Within The Frame
Of Physics, arXiv:0707.1161 (2007).
12
[33] G. Kitagawa, and W. Gersch, Smoothness Priors Analy-
(2002)).
sis of Time Series. (Springer, 1996).
[34] P.J. Brockwell, and R.A. Davis, Introduction to Time
Series and Forecasting. (Springer, 1996).
[35] G. Kitagawa, Time Series Analysis Programming,
Iwanami Shoten. (Tokyo, 1993, in Japanese).
[36] The data are available at http://www.frontier.iarc.
uaf.edu:8080/ igor/research/ice/icedata.php (see also I.
Polyakov, et al., Long-Term Ice Variability in Arctic
Marginal Seas, Journal of Climate 16, 2078-2085 (2003)).
[37] N. Scaffeta and B.J. West, Is climate sensitive to solar
variability? , Physics Today, 51, 50-51 (2008).
[38] N. Scaffeta and B.J. West, Estimated solar contribu-
tion to the global surface warming using the ACRIM
TSI satellite composite, Geophys. Res. Lett., 33, L05708
(2006) (see also arXiv:physics/0509248).
[39] The data are available at http://www.sec.noaa.gov/ So-
larCycle/SC24.
[45] D.E. Sigeti, Survival of deterministic dynamics in the
presence of noise and the exponential decay of power
spectrum at high frequencies. Phys. Rev. E, 52, 2443-
2457 (1995).
[46] J. Sun, Y. Zhao, T. Nakamura, and M. Small, From phase
space to frequency domain: A time-frequency analysis for
chaotic time series, Phys. Rev. E 76, 016220 (2007).
[47] N. Ohtomo, K. Tokiwano, Y. Tanaka, A. Sumi, S.
Terachi, and H. Konno, Exponential Characteristics of
Power Spectral Densities Caused by Chaotic Phenomena,
J. Phys. Soc. Jpn. 64 1104-1113 (1995).
[48] J. D. Farmer, Chaotic attractors of an infinite dimen-
sional dynamic system, Physica D, 4, 366-393 (1982).
[49] J. Feynman, and S.B. Gabriel, Period and phase of the
88-year solar cycle and the Maunder Minimum: Evidence
for a chaotic Sun, Solar Physics, 127, 393-403 (1990).
[50] The data are available at http://www.ncdc.noaa.gov/
[40] A.C. Monin and A.M. Yaglom, Statistical Fluid Mechan-
paleo/metadata/noaarecon-6267.html
ics (MIT Press, Cambridge, 1975), Vol. 2.
[41] C.H Gibson, Kolmogorov Similarity Hypotheses for
Scalar Fields: Sampling Intermittent Turbulent Mixing
in the Ocean and Galaxy, Proc. Roy. Soc. Lond. 434, 149
(1991).
[42] J. Cho, A. Lazarian, and E.T. Vishniac, Simula-
tions of MHD Turbulence in a Strongly Magnetized
Medium, Astrophys. J. 564, 291-301 (2002) (see also
arXiv:astro-ph/0205286).
[43] C.H. Gibson, R.N. Keeler, V.G. Bondur,
et al.,
Submerged turbulence detection with optical
satel-
lites, Proc. of SPIE, 6680, 6680-33 (2007) (see also
arXiv:0709.0074v2 [astro-ph]).
[44] The data are available at http://www1.ncdc.noaa.gov/
pub/data/paleo/climate forcing/solar variability/usoskin-
cosmic-ray.txt (see also I.G. Usoskin, K. Mursula, S.K.
Solanki, M. Schuessler, and G.A. Kovaltsov, A physical
reconstruction of cosmic ray intensity since 1610. J.
Geophys. Res., J. Geophys. Res., 107(A11), 1374-1380
[51] A. Moberg, D. M. Sonechkin, K. Holmgren, N. M. Dat-
senko and W. Karlen, Highly variable Northern Hemi-
sphere temperatures reconstructed from low- and high
resolution proxy data Nature, 433, 613-617 (2005).
[52] O.E. Rossler, An equation for continuous chaos, Phys.
Lett. A, 35, 397-398 (1976).
[53] F. Fucito, F. Marchesoni, E. Marianari, G. Parisi,
L.Peliti, S. Ruffo, and V. Vulpiani, Approach to equi-
librium in a chain of nonlinear oscillators, J. Physique,
43, 707-713 (1982).
[54] U. Frisch and R. Morf, Intermittency in non-linear dy-
namics and singularities at complex times, Phys. Rev.
23, 2673 (1981).
[55] A. Bershadskii, Chaotic dynamics of atmospheric CO2 at
millennial timescales, arXiv:0903.2795 (2009).
[56] K. Mursula, I. G. Usoskin, and G. A. Kovaltsov, Per-
sistent 22-year cycle in sunspot activity: Evidence for a
relic solar magnetic field, Solar Phys., 198, 51-56, 2001.
|
astro-ph/9804297 | 1 | 9804 | 1998-04-28T11:46:11 | Towards the absolute planes: a new calibration of the Bolometric Corrections and Temperature scales for Population II Giants | [
"astro-ph"
] | We present new determinations of bolometric corrections and effective temperature scales as a function of infrared and optical colors, using a large database of photometric observations of about 6500 Population II giants in Galactic Globular Clusters (GGCs), covering a wide range in metallicity (-2.0<[Fe/H]<0.0). New relations for BC_K vs (V-K), (J-K) and BC_V vs (B-V), (V-I), (V-J), and new calibrations for T_eff, using both an empirical relation and model atmospheres, are provided. Moreover, an empirical relation to derive the R parameter of the Infrared Flux Method as a function of the stellar temperature is also presented. | astro-ph | astro-ph |
Mon. Not. R. Astron. Soc. 000, 000 -- 000 (1997)
Printed 31 May 2021
(MN LATEX style file v1.4)
Towards the absolute planes: a new calibration of the
Bolometric Corrections and Temperature scales
for Population II Giants ⋆
P. Montegriffo,1 F.R. Ferraro,1 L. Origlia,1 and F. Fusi Pecci2†
1Osservatorio Astronomico di Bologna, Via Zamboni 33, 40126 Bologna, ITALY
2Stazione Astronomica, 09012 Capoterra, Cagliari, ITALY
31 May 2021
ABSTRACT
We present new determinations of bolometric corrections and effective temperature
scales as a function of infrared and optical colors, using a large database of photometric
observations of about 6500 Population II giants in Galactic Globular Clusters (GGCs),
covering a wide range in metallicity ( -- 2.0<[Fe/H]<0.0).
New relations for BCK vs (V -- K), (J -- K) and BCV vs (B -- V), (V -- I), (V -- J), and new
calibrations for Tef f , using both an empirical relation and model atmospheres, are
provided.
Moreover, an empirical relation to derive the R parameter of the Infrared Flux Method
as a function of the stellar temperature is also presented.
Key words: Clusters: Globular -- Stars: Evolution -- Stars: Fundamental Parameters
-- Stars: Hertzsprung -- Russell diagram -- Photometry: IR -- Array
1
INTRODUCTION
A global test of stellar evolutionary models requires a direct
comparison between theoretical tracks and observations for
stars spanning a wide range in stellar parameters, such as
temperature, luminosity and metallicity. In order to achieve
these goals at least two fundamental ingredients are needed:
i) a complete and homogeneous database of photomet-
ric observations;
ii) a suitable set of transformations between observ-
ables and absolute quantities.
GGCs are the best empirical laboratory to obtain com-
plete and homogeneous spectrophotometric information on
Pop. II stars over a wide range of metallicities.
Rewieving the published works on the transformations
to the absolute plane (see Sect.3), that is bolometric correc-
tions (BCs) and temperature scales as a function of differ-
ent colors, it is easy to see that very often these calibrations
are not based on a complete and homogeneous set of data,
spanning a wide range of stellar parameters. Moreover, ad
hoc correction factors are usually adopted to take into ac-
count for example possible systematic differences between
different photometric systems and/or different assumptions
⋆ Based on data taken at the ESO -- MPI 2.2m Telescope equipped
with the near IR camera IRAC2 - ESO, La Silla (Chile).
† on leave from Osservatorio Astronomico di Bologna
c(cid:13) 1997 RAS
for the reference solar quantities, for the adopted model at-
mospheres, and for different laws to extrapolate the data
towards the UV/IR ranges.
Such a scenario indicates that any calibration in the
absolute plane can hardly be fully self -- consistent as it always
depends on the adopted transformations and, more crucial,
the residuals among different scales are very rarely linear
with the involved parameters.
In order to improve the available determinations of
the bolometric corrections and temperature scales, we use
here our IR photometric database on GGC stars combined
with available optical data from the literature to calibrate
new, independent, (hopefully) self -- consistent transforma-
tions, particularly useful to study the red stellar sequences
in Pop II stars.
In Sect.2 we present the complete database used in our
analysis which includes about 6500 RGB and HB stars in
a sample of 10 GGCs observed in both optical and near
IR bands. In Sect.3 we derive the transformations from ob-
served magnitudes and colors to absolute quantities, such
as bolometric corrections and effective temperatures. All
the results are listed in Table 3. In Sect.4 we compare the
inferred scales with existent ones and we give a fully em-
pirical calibration of the R parameter of the Infrared Flux
Method (IRFM) as a function of the effective temperature.
Schematic conclusions are eventually presented in Sect.5.
2
P. Montegriffo et al.
2 THE DATABASE
The IR database used in this study includes photometric ob-
servations in J and K bands of about 17000 stars belonging
to 10 GGCs spanning the whole range in metallicity (from
[Fe/H]= -- 2.15 to [Fe/H]≈0.0). The data were obtained at
ESO, La Silla (Chile), during two different runs (on June
1992 and June 1993), using the ESO -- MPI 2.2m telescope
and the near -- IR camera IRAC -- 2 (Moorwood et al. 1992)
equipped with a NICMOS -- 3 256x256 array detector. The
complete description of the observations and data reduction
can be found in Ferraro et al. (1994a,b) and Montegriffo et
al. (1995) and in a series of forthcoming papers, where each
individual cluster is discussed in more detail.
Our IR database was cross-correlated with other optical
and IR catalogs in the literature in order to provide complete
UBVRIJK photometry of as many stars as possible. The
final sample used in the following analysis includes about
6500 RGB and HB stars.
The reddening correction in each photometric band was
performed starting from the E(B -- V) color excess (as re-
ported by Armandroff (1989) with the exception of M68
and M69 for which we used the values by Walker 1994 and
Ferraro et al. 1994a, respectively) and using the extinction
law proposed by Rieke & Lebovsky (1985). In Table 1 the
main features of the database are listed: the GGC names, the
cluster metallicity as quoted by Zinn (1985), the available
photometric bands, and the adopted E(B -- V) color excess.
In the following the observed colors are always corrected for
reddening.
2.1 Notes on individual clusters
M15: UBVR photometry from Stetson (1994).
M30: UBV photometry from Bergbusch (1996).
M68: BVI photometry from Walker (1994).
M55: BV photometry from Piotto (1996) and VI from Or-
tolani & Desidera (1996): V was averaged.
M4: BV photometry from Lee (1977).
M107: BV photometry from Ferraro et al. (1991).
M69: BV photometry from Ferraro et al. (1994a).
47Tuc: UBV photometry from Auriere & Lauzeral (1996)
and VI from Ortolani & Desidera (1996). The VI photom-
etry was compared with that one obtained by Da Costa &
Armandroff (1990) for the few stars in common. There is
an excellent agreement in V, while a systematic shift in the
zero point by about 0.12 mag in I has been found: we ap-
plied this correction to the I photometry from Ortolani &
Desidera (1996).
NGC6553: VI photometry from Guarnieri et al. (1997) us-
ing HST data.
NGC6528: VI photometry from Ortolani et al. (1995) using
HST data.
3 ABSOLUTE QUANTITIES
Various calibrations of the bolometric correction and tem-
perature scale have been proposed by different authors since
many years.
Table 1. Adopted IR and optical database.
Cluster
NGC7078 M15
NGC7099 M30
NGC4590 M68
NGC6809 M55
NGC6121
M4
NGC6171 M107
NGC6637 M69
NGC104
47Tuc
NGC6553
NGC6528
[Fe/H]a
-- 2.15
-- 2.13
-- 2.09
-- 1.82
-- 1.19
-- 0.89
-- 0.75
-- 0.70
-- 0.29
-- 0.07
Photometryb
E(B -- V)c
UBVRJK
UBVJK
BVIJK
BVIJK
BVJK
BVJK
BVJK
UBVIJK
VIJK
VIJK
0.10
0.04
0.07
0.06
0.40
0.31
0.17
0.04
0.78
0.56
a From Zinn (1985).
b J,K from our IR -- array survey, U,B,V,R,I from the literature
(refers to Sect.2.1 for the bibliographic sources).
c E(B -- V) from Armandroff (1989), but M68 from Walker
(1994) and M69 from Ferraro et al. (1994a), respectively.
Johnson (1966) used a sample of 15 giants observed in
the UBVRIJKLMN to get bolometric corrections in the V
band, integrating the spectral energy distribution and apply-
ing small corrections (<0.1 mag) in the wavelength ranges
not covered by his survey (such as for example the H band)
to take into account possible molecular blending in cool
stars. The zero point is defined as BC⊙
V =0.00.
Carney & Aaronson (1979) reconstructed the energy
distribution of a sample of dwarfs and subdwarfs by means
of polynomial fits to the optical photometry, while the UV
range was extrapolated using the Kurucz (1979) models and
the IR region using blackbodies. Their zero point is BC⊙
V = --
0.12.
Frogel, Persson & Cohen (1981, hereafter FPC81) per-
formed trapezoidal integrations of the energy distribution of
a sample of stars observed in the UBVRIJHKL, and extrap-
olating the UV and IR fluxes. They adopted as zero point
the value BC⊙
V = -- 0.08.
More recently, Tinney et al. (1993) reconstructed the
spectral energy distribution of cool stars combining pho-
tometric data in the VIJHKLL' bands with low resolution
spectra.
Alonso et al. (1995) determined the bolometric correc-
tions in the K band of a sample of F,G,K dwarfs. They
performed trapezoidal integrations using UBVRIJHK pho-
tometric data and included a correction factor C = f (Tef f ,
log g, [Fe/H]) to the UV and IR fluxes as a function of the
stellar parameters, using the Kurucz (1993) models.
Concerning the determination of the stellar effective
temperature, the only pure experimental way to derive it
is to know the star intrinsic luminosity and apparent angu-
lar diameter. Ridgway et al. (1979) for example measured
the angular diameter of a sample of nearby Pop I giants by
means of lunar occultations, while Di Benedetto & Rabbia
(1987, hereafter DBR87) and Di Benedetto (1993, hereafter
DB93) used interferometric techniques.
Once the apparent stellar diameters have been mea-
sured, the effective stellar temperatures can be estimated
assuming a proper scale of bolometric corrections. DBR87
have shown how an error of 0.05 mag in the determination
of the BCK transfers into an error of ≈60 K in the tem-
perature estimates for Tef f around 5000 K and of ≈35 K
for Tef f ≈3000 K. These uncertainties are of the same or-
c(cid:13) 1997 RAS, MNRAS 000, 000 -- 000
Bolometric Corrections and Temperature scales for Population II giants
3
Table 2. Absolute flux calibrations of a zero magnitude star in UBVRIJHKL
bands for different photometric systems. Wavelengths are in µm and fluxes in
units of 10−8 erg s−1 cm−2 µm−1.
Filter
U
B
V
R
I
J
H
K
L
Johnsona
λef f
0.36
0.44
0.55
0.70
0.90
1.25
1.62
2.20
3.50
fλ0
4345
7194
3917
1762
829.9
289.7
107.9
38.02
7.834
Bessell b
λef f
0.366
0.438
0.545
0.641
0.798
1.221
1.632
2.187
3.451
fλ0
4175
6320
3631
2177
1126
314.7
113.8
39.61
7.080
ESOc
λef f
0.363
0.438
0.548
0.641
0.795
1.244
1.634
2.190
3.770
fλ0
4080
6490
3650
2190
1180
312
120
41.7
5.41
a Johnson system revised by Buzzoni (1996).
b Bessell system adopted in BCP97 (Castelli 1997).
c UBVRI from Bessell (1990) and from Megessier (1995).
der of magnitude as the empirical errors, hence an accurate
estimate of BCK is an important issue and severely affects
the overall accuracy of the temperature estimated using this
technique.
Another method which allows one to derive stellar tem-
peratures in more distant objects is represented by the so --
called IRFM proposed by Blackwell, Shallis & Selby (1979)
and Blackwell & Lynas -- Gray (1994).
This method uses as a fundamental parameter the quan-
tity R = fbol/fλ, where fbol is the observed bolometric flux
and fλ is the observed flux in a certain photometric band.
Such a quantity is compared with the corresponding theo-
retical one, derived by means of models of atmospheres, in
order to derive a temperature scale. Tinney et al. (1993) for
their sample of dwarfs (see above) derived Tef f by subse-
quent iterations of the IRFM.
Finally, a fully theoretical approach is the extensive use
of model atmospheres to derive synthetic colors for different
input stellar parameters.
Before describing the adopted procedures to calibrate
our relations, we want to add a further comment on a cru-
cial point: each absolute flux calibration requires a reference
zero -- magnitude star in all the explored photometric bands,
that is in the specific photometric system used. As pointed
out by many authors (see e.g. Bessell & Brett 1988; Buz-
zoni 1989; Bessell 1990, Megessier 1995) the photometric
CCD+filter characterization is often provided with a quite
large uncertainty (typically ≈5%), and this may severely af-
fect the accuracy of the derived quantities, particularly the
bolometric fluxes, based on this instrumental calibration.
Moreover, it is difficult to reconstruct the adopted values
from the various authors since very often they are not tab-
ulated. Since in the calibrations of the present paper we
worked with different photometric systems, in Table 2 we
report all the adopted zero magnitude fluxes.
3.1 The determination of bolometric corrections
3.1.1 The BCK vs (V -- K) and (J -- K)
For all the observed stars in the selected sample of GGCs
we computed the bolometric correction BCK in the K band
c(cid:13) 1997 RAS, MNRAS 000, 000 -- 000
by means of the relation:
BCK = mbol − K
where mbol is the apparent bolometric magnitude.
mbol was computed as follows: first we convert the ap-
parent magnitudes corrected for reddening into fluxes using
the classical formula:
fλ = fλ0 · 10−0.4·mλ
where fλ0 is the absolute flux in a certain photometric band
corresponding to the zero magnitude. We used the UBVJK
filters of Johnson and RI of Cousins in the ESO standard
photometric system. In Table 2 we report the absolute flux
calibrations of a zero magnitude star for all the observed
photometric bands, for different photometric systems. The
values we adopted in our analysis are those listed in columns
#6 and #7.
The energy distribution was reconstructed starting from
a set of Planck functions Bλ(Tc) computed between two ad-
jacent observed bands, where Tc is the corresponding color
temperature. We want to stress that the color temperature
is here simply a "working parameter" and it is not used to
constrain the effective stellar temperature.
The final dense grid was obtained interpolating between
two adjacent bands and extrapolating in the UV and IR
ranges not covered by the observations. The apparent bolo-
metric magnitude was then computed by means of integra-
tion over the energy distribution as follows:
mbol = − 2.5 log10 Z fλ dλ + ZP
The zero point (ZP) of the bolometric scale is given by
ZP = M ⊙
bol + 2.5 log10 (F ⊙
bol)
so its actual value depends upon the assumed M ⊙
bol and
V =4.82 (so BC⊙
F ⊙
bol. We used M⊙
V = -- 0.07),
L⊙
bol = 3.86×1033 erg s−1, as reported by Armandroff (1989),
and the relation: F ⊙
bol/(4πd2), where d=10 pc. We
obtained ZP= -- 11.478.
bol = 4.75, M⊙
bol = L⊙
The accuracy of the inferred bolometric magnitudes pri-
marily depends on the number of observed bands since ex-
trapolations over a wide wavelength range may represent a
severe approximation of the true energy distribution.
4
P. Montegriffo et al.
(a) mbol (tot) -- mbol (UBVRJK), (b) mbol (tot) --
Figure 1.
mbol (UBVIJK), (c) mbol (tot) -- mbol (BVIJK), and (d) mbol
(tot) -- mbol (VIJK) vs the (V -- K) color (see Sect.3.1.1). mbol (tot)
was computed using the complete UBVRIJHKLMN set of filters.
Open circles mark the stars by Morel & Magnenat (1978), while
the continuous lines are our best fits.
Figure 2. Differences between the inferred BCK using our proce-
dure and that of the models, as a function of the (V -- K) color. This
comparison is performed using as reference database the set of
synthetic grids of stellar atmospheres by BCP97. The dashed hor-
izontal line is a systematic zero point shift of ∼0.01 between the
two calibrations (our -- models) (see Sect.3.1.1). At (V -- K)≤ 1 the
Balmer discontinuity strongly affects the accuracy of our BCK .
In order to check the possible corrections to apply to the
inferred values when a small number of bands was observed,
we performed the following experiment: we used the catalog
by Morel & Magnenat (1978) which contains 212 stars with
a complete set of UBVRIJHKL photometry in the Johnson
system and we applied our method to derive the apparent
bolometric magnitudes with varying the number of available
photometric bands, using the Johnson absolute flux calibra-
tion (columns #2 and #3 in Table 2).
In Fig.1 we plot as an example a few simulations of
the differences between mbol computed using all the Morel
& Magnenat (1978) bands and only those available in our
GGC database vs the (V -- K) color. Similar simulations were
performed using (J -- K). The continuous line in the plots is
our numerical best fit and represents the final corrections
we applied to the bolometric magnitudes of the stars in our
database as a function of their (V -- K) and (J -- K) colors and
the available photometric bands.
On average, the discrepancies between the bolometric
corrections computed using the entire set of photometric
data and those computed using only a few bands are always
within a few hundredths of a magnitude as one can see in
Fig.1. Larger discrepancies, up to 0.1 mag, were found in
the case of hot stars which lack U band observations: in this
case, the UV continuum computed with a Planck function
using the (B -- V) color temperature is overestimated.
As a further check on the accuracy of the proposed
method to derive the bolometric corrections we also recon-
structed the energy distribution using direct trapezoidal in-
tegrations as suggested by many authors. The results of such
a computation indicated that the corrections to apply if only
a few colors are available should be even larger than in the
case of integrations by means of Planck functions.
We did not apply any further correction due to the
Balmer discontinuity in the UV and possible molecular
sources of opacity in the IR. Nevertheless, we performed
some tests to evaluate the entity of such an effect using
UBVRIJHKL synthetic color indices computed by Bessell,
Castelli & Plez (1997, hereafter BCP97) and derived from
different grids of model atmospheres, which include both
the ATLAS9 models computed without any overshooting
for the convection (Castelli et al. 1997) and NMARCS mod-
els computed by Plez, Brett, & Nordlund (1992) and Plez
(1995). The synthetic indices derived from Plez (1995) mod-
els are computed for Tef f ranging from 3600 K to 4750
K, gravity log g ranging from from -- 0.5 to 3.5, and metal-
licities [Fe/H]= -- 0.6, -- 0.3,0.0,0.3, 0.6. Synthetic indices from
ATLAS9 models and from Plez et al. (1992) models are tab-
ulated only for the solar metallicity.
We computed the BCK of these synthetic stars follow-
ing the same procedure as for true stars in our database
using the Bessell absolute flux calibration (columns #4 and
#5 in Table 2). We compared the inferred values with those
of the models themselves. The results of this comparison are
plotted in Fig.2 as a function of the (V -- K) color. For (V --
K)>1, the agreement between the two scales is satisfactory
apart from an average shift in the zero point ∆BCK =0.01,
possibly ascribed to the different assumption on M⊙
bol (they
adopt M⊙
bol = 4.74).
For bluer colors the agreement between the two scales
gets exponentially worse and worse since the continuum
c(cid:13) 1997 RAS, MNRAS 000, 000 -- 000
Bolometric Corrections and Temperature scales for Population II giants
5
Figure 3. BCK as a function of (V -- K) and (J -- K) colors for
metal poor (left panels) and metal rich (right panels) stars. Con-
tinuous lines are our best fit.
Figure 4. BCV vs (B -- V), (V -- I) and (V -- J) colors for metal poor
(left panels) and metal rich (right panels) stars. Continuous lines
are our best fits.
shape is strongly affected by the Balmer discontinuity and
a Planck function is unable to properly reproduce it.
In order to investigate a possible dependence of the
bolometric correction vs color relations on the metallicity,
we divided the observed clusters into two groups: one at low
metallicity ([Fe/H]< -- 1.0), using about 2000 stars in M15,
M30, M68, M55 and M4, and the other at higher metallic-
ity ([Fe/H]> -- 1.0), using about 4500 stars in M107, M69, 47
Tuc, NGC6553 and NGC6528). We found a small (a few hun-
dredths of a magnitude) difference when fitting our metal
poor and metal rich stars in the empirical BCK vs (V -- K)
and (J -- K) planes, more evident at (V -- K)<2 and (J -- K)<0.7.
This difference, that we want to stress is purely observa-
tional, could be explained in terms of a small metallicity
dependence of the (V -- K) and (J -- K) colors as suggested by
the model atmospheres whose continuum opacities, particu-
larly at optical wavelengths, are slightly affected by changing
the metal content. A similar behaviour was found by Alonso
et al. (1995) using an independent, semi -- empirical approach
to estimate the bolometric flux (see Sect.1).
Nevertheless, since the difference we found is similar in
size to the intrinsic photometric error, we cannot completely
exclude the possibility that the observed effect is due to a
systematic bias in the data .
In Fig.3, we plot the derived BCK as a function of the
(V -- K) and (J -- K) colors for metal poor and metal rich stars
separately. Our best fits are shown as continuous lines. The
relation with the lowest spread is, as expected, that based on
the (V -- K) color which has the largest wavelength baseline.
In Table 3 we report the corresponding BCK values as
a function of the (V -- K) and (J -- K) colors from our best fits,
both for metal poor and metal rich stars. The BCK is fixed
(in other words we fixed the stellar temperature, see next
section) and the colors slightly change (≤ 0.1 mag) in the
two metallicity regimes.
c(cid:13) 1997 RAS, MNRAS 000, 000 -- 000
3.1.2 Other bolometric corrections
Using our database we also inferred an estimate of BCV as a
function of the (B -- V), (V -- I) and (V -- J) colors, respectively,
for metal poor and metal rich stars, as in the case of the
BCK . In Fig.4, we plot the results and in Table 3 we listed
the values for the corresponding BCK .
The possible small dependence of the (V -- K) color on
metallicity should imply a similar dependence of the BCV ;
hence, BCK could be a purer thermometer than BCV .
Moreover, in the BCV vs (B -- V) plane the difference
between metal poor and metal rich stars is more pronounced
(up to about 0.2 mag), since metallicity effects are more
important at shorter wavelengths.
3.2 The determination of Tef f
Our photometric database does not allow us to calibrate di-
rectly an empirical temperature scale since we do not know
the angular diameter of the observed stars. However DB93
provided a sample of field giants with measured angular di-
ameters by means of Michelson interferometry and lunar
occultations. We used his database, our BC scale and zero
point to calibrate a new empirical relation which links the
effective temperature to the BCK .
We also provided fully theoretical temperature scales as
a function of suitable colors and BC, using the grids of stellar
atmospheres computed by BCP97 at solar metallicity.
3.2.1 Empirical Tef f
We used the DB93's giants listed in his Table 1 and 3 of
DB93, for which accurate estimates of the angular diameter
exist, and by means of a second order polynomial fit we
obtained V fluxes as a function of the (V -- K) color, using our
6
P. Montegriffo et al.
V -- K
Poor
-0.413
-0.338
-0.259
-0.177
-0.091
-0.001
0.092
0.190
0.292
0.399
0.510
0.626
0.745
0.869
0.996
1.123
1.249
1.380
1.506
1.641
1.778
1.925
2.083
2.239
2.415
2.595
2.791
3.006
3.265
3.551
3.860
4.216
4.626
--
--
--
--
--
-0.40
-0.30
-0.20
-0.10
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
1.10
1.20
1.30
1.40
1.50
1.60
1.70
1.80
1.90
2.00
2.10
2.20
2.30
2.40
2.50
2.60
2.70
2.80
2.90
3.00
3.10
3.20
3.30
-0.137
-0.119
-0.099
-0.075
-0.052
-0.030
-0.003
0.026
0.055
0.082
0.113
0.145
0.178
0.210
0.244
0.279
0.316
0.350
0.385
0.423
0.461
0.498
0.532
0.572
0.612
0.653
0.690
0.738
0.790
0.846
--
--
--
--
--
--
--
--
0.013
0.038
0.059
0.077
0.091
0.101
0.108
0.110
0.108
0.101
0.090
0.074
0.055
0.031
0.004
-0.023
-0.049
-0.080
-0.106
-0.141
-0.178
-0.225
-0.283
-0.339
-0.415
-0.495
-0.591
-0.706
-0.865
-1.051
-1.260
-1.516
-1.826
--
--
--
--
--
-0.118
-0.089
-0.062
-0.040
-0.009
0.033
0.068
0.078
0.078
0.100
0.133
0.168
0.191
0.218
0.259
0.307
0.353
0.406
0.451
0.510
0.572
0.651
0.747
0.838
0.961
1.088
1.239
1.416
--
--
--
--
--
--
--
--
--
--
-0.241
-0.204
-0.169
-0.132
-0.098
-0.068
-0.032
-0.004
0.203
0.253
0.293
0.341
0.390
0.444
0.498
0.548
0.592
0.642
0.681
0.731
0.782
0.843
0.915
0.981
1.066
1.152
1.250
1.362
1.512
--
--
--
--
--
--
--
--
--
-0.420
-0.360
-0.300
-0.237
-0.176
-0.120
-0.064
0.157
0.234
0.319
0.365
0.436
0.522
0.615
0.721
0.828
0.918
1.018
1.095
1.193
1.289
1.402
1.531
1.647
1.792
1.934
2.090
2.263
2.481
2.713
2.950
--
--
--
--
--
--
--
Table 3. BCK as a function of (V -- K), (J -- K), BCV , (B -- V), (V -- I), (V -- J) colors for metal poor and metal rich stars, respectively, and
both empirical and theoretical Tef f .
BCK
BCV
B -- V
J -- K
BCV
V -- I
V -- J
J -- K
V -- K
Rich
--
--
--
--
--
--
--
--
--
0.425
0.522
0.621
0.723
0.829
0.938
1.050
1.168
1.289
1.417
1.549
1.688
1.834
1.987
2.147
2.317
2.498
2.691
2.900
3.127
3.390
3.703
4.117
4.696
5.378
6.263
7.485
9.444
19.746
B -- V
V -- I
V -- J
--
--
--
--
--
--
--
--
--
0.082
0.107
0.132
0.159
0.187
0.216
0.245
0.274
0.305
0.336
0.369
0.405
0.444
0.486
0.533
0.582
0.634
0.689
0.747
0.807
0.870
0.937
1.006
1.077
1.152
1.229
1.310
1.393
1.478
--
--
--
--
--
--
--
--
--
0.075
0.078
0.079
0.077
0.071
0.062
0.050
0.032
0.011
-0.017
-0.049
-0.088
-0.134
-0.187
-0.247
-0.317
-0.398
-0.491
-0.600
-0.727
-0.890
-1.103
-1.417
-1.896
-2.478
-3.263
-4.385
-6.244
-16.446
--
--
--
--
--
--
--
--
--
--
--
0.373
0.378
0.393
0.413
0.440
0.476
0.516
0.564
0.614
0.670
0.731
0.795
0.863
0.932
1.005
1.087
1.173
1.263
1.362
1.468
1.612
1.659
1.690
1.732
1.793
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
--
-0.067
0.111
0.321
0.561
0.761
0.900
1.027
1.133
1.226
1.319
1.426
1.559
1.762
2.111
2.564
3.033
3.620
4.443
8.119
--
--
--
--
--
--
--
--
--
--
--
0.563
0.575
0.611
0.662
0.724
0.808
0.897
1.003
1.111
1.228
1.352
1.480
1.610
1.746
1.889
2.038
2.195
2.359
2.544
2.767
3.102
3.637
4.268
5.099
6.258
8.142
18.471
Tef f
Tef f
Emp Theor
10462
10139
9824
9518
9219
8928
8645
8370
8102
7842
7589
7343
7105
6871
6642
6421
6208
6003
5806
5618
5438
5264
5097
4933
4774
4618
4468
4321
4179
4040
3904
3771
3639
3516
3385
3243
2929
2499
--
--
--
--
--
--
--
--
--
7602
7386
7173
6963
6756
6553
6353
6156
5963
5774
5589
5407
5229
5056
4887
4723
4563
4406
4254
4106
3962
3819
3667
3523
3409
3317
3247
3199
3174
zero points. We did not use DB93's linear relations since they
have a discontinuity around (V -- K)=3.7 which propagates
into the relation which gives Tef f as a function of (V -- K)
and BCK .
Following the procedure described in DB93 we derived
the relation:
log Tef f = 3.9619 − 0.0466(V − K) + 0.0038(V − K)2−
0.1BCK
Since DB93's database is at solar metallicity the pre-
vious relation can be regarded as reliable only for metal
rich stars. Hence we computed the effective temperature us-
ing the latter relation only for the metal rich stars in our
database and the results are reported in Fig.5 as a function
of BCK , while the mean relation is listed in Table 3. The
average dispersion around the fiducial line is ∼30 K. Outside
the 1< BCK <3.1 range the relation was extrapolated.
3.2.2 Theoretical Tef f
The calibration of the effective temperature was performed
using the grid of stellar atmospheres by BCP97 at solar
metallicity. In order to evaluate the possible shift with vary-
ing metallicity we also used a grid of color indices for so-
lar and 1/100 solar metallicity computed by Castelli (1997).
These grids are based on the ATLAS9 model atmospheres in
which the modification of the "approximate overshooting" to
the mixing-length treatment of the convection was dropped.
The indices slightly differ from the BCP97 indices owing to
different passbands and zero points. For a given tempera-
ture and metallicity we interpolated the grids in gravity by
means of the VandenBerg (1996) isochrones. For each star
in our database we derived a mean temperature using both
the (V -- K), (V -- J), (J -- K) colors and the BCK .
The relations were derived in the Bessell (1990) pho-
tometric system, as adopted by BCP97 to compute their
stellar atmosphere grids. On the other hand, our observed
database refers to the ESO photometric system. We used
the relations quoted by Bessell & Brett (1988) to transform
the (V -- K), (V -- J) and (J -- K) colors of the models from the
Bessell system to the ESO one. Moreover, since between the
BCK computed with our procedure and those of the models
a systematic shift in the zero point of 0.01 mag was found
(see Sect.3.1.1 and Fig.2), we added this shift to the BCK of
c(cid:13) 1997 RAS, MNRAS 000, 000 -- 000
Bolometric Corrections and Temperature scales for Population II giants
7
Figure 5. Empirical Tef f as a function of BCK (upper panel) for
metal rich stars. The continuous line is our best fit. The dispersion
(on average 30 K) around the fiducial line is also plotted (lower
panel).
Figure 6. Theoretical Tef f as a function of (V -- K), (V -- J), (J -- K)
and BCK .
the models. This avoids to have a systematic drift between
Tef f derived from the colors and the BCK .
In Fig.6 we plot the behaviour of the effective tempera-
ture as a function of both the (V -- K), (V -- J) and (J -- K) colors
in the ESO photometric system and the BCK . Our best fits
(continuous lines) are plotted as well. The spread visible in
the Tef f vs color distributions at the lowest temperature is
due to a gravity effect. It demonstrates how small variations
of such a parameter may strongly affect the accuracy of the
derived temperatures below 3500 K. This effect is less pro-
nounced in the Tef f vs BCK distribution. The latter is also
the most linear and sensitive relation, particularly at low
temperatures, confirming that BCK is probably the purest
stellar thermometer for cool stars.
In Fig.7 the theoretical Tef f derived from averaging the
temperatures obtained by the color and BCK as a function
of the latter are plotted together with the dispersion (on
average 180 K) around the fiducial line and in Table 3 we
listed the values.
3.2.3 Comparison between the two scales
Comparing the empirical and theoretical temperature scales
discussed in the previous sections, we infer the following
behaviours:
Figure 7. Theoretical Tef f derived from averaging the temper-
atures obtained by the color and BCK as a function of the latter
(upper panel). The continuous line is our best fit. The dispersion
(on average 180 K) around the fiducial line is also plotted (lower
panel).
• in the range 4000 -- 7500 K, a general good agreement
has been found, the theoretical scale being systematically
warmer by about 50 K;
• between 3200 and 4000 K, this difference increases up
The behaviour in the coolest temperature range can
be explained as a gravity effect, as already mentioned in
Sect.3.2.2, in the sense that the interpolation in gravity may
be quite uncertain.
to about 100 K;
• below 3200 K, the two scales diverge, the theoretical
one becoming progressively cooler, up to 800 K.
c(cid:13) 1997 RAS, MNRAS 000, 000 -- 000
8
P. Montegriffo et al.
Figure 8. (J -- K) vs (V -- K) color -- color diagram for the mean
loci of the metal rich (continuous line) and metal poor (dashed
line) observed stars and different model atmospheres. ATLAS9
models by Castelli et al. (1997) are without overshooting. The
mean sequences of a sample of field giants and dwarfs by Bessell
& Brett (1988) and the reddening vector are plotted as well.
Figure 9. K vs (J -- K) diagram of the isochrones by Bergbusch &
VandenBerg (1992) transformed into the observational plane at
two different metallicities: dotted lines are the relations inferred
using BG89 transformations, continuous lines represent those cor-
rected as suggested by Bell (1992) to reproduce the observed RGB
distribution, and stars are those obtained using our BCK and
Tef f transformations.
4 DISCUSSION
4.1 A few more tests
In order to check the consistency between our observed
database and the adopted model atmospheres, both used
in the calibrations of the temperature scale (see Sect.3.2.2),
we constructed a (J -- K) vs (V -- K) color -- color diagram in
the ESO photometric standard system for all the stars in our
database (see Fig.8). The mean loci of the observed metal
poor and metal rich stars, different models, and the mean
sequences of a sample of field giants and dwarfs as reported
in Table 2 and 3 of Bessell & Brett (1988) are plotted as
well. The observed and synthetic distributions well overlap
one to each other and also overlap the field giant sequence,
indicating that the two photometric samples are fully con-
sistent.
Another interesting test we performed is to apply our
BCK and Tef f relations to the widely used isochrones com-
puted by Bergbusch & VandenBerg (1992), the only ones
presently published in the IR -- planes. Bell (1992, and refer-
ence therein) has shown how these theoretical isochrones
transformed into the observational plane using the Bell
& Gustafsson (1989, hereafter BG89) model atmospheres
hardly reproduce the sequence of cool stars observed by
Glass (1974a,b), being the former too red by ≈0.1 mag along
the RGB.
The results of our test are plotted in Fig.9: the se-
quences obtained using our new calibrations well overlap
those computed by using BG89 and corrected as suggested
by Bell (1992) along the RGB (continuous lines), while
they overlap the uncorrected ones in the turn -- off region
(dashed lines). Hence, it is clear that the discrepancy found
by Bell (1992) along the RGB was not due to the theoret-
ical isochrones but should rather be a consequence of the
adopted transformations into the IR observational plane.
4.2 Comparison with published BC scales
Our BCs, as reported in Table 3, in both the metal poor
and metal rich regimes, were then compared with published
scales.
In the BCK vs (V -- K) plane, we considered the relations
presented by Johnson (1966), Lee (1970), Ridgway et al.
(1980), DBR87 and BG89, as tabulated by DB93 and scaled
by 0.01 mag to take into account the different BC⊙ adopted.
Moreover, we considered those obtained by FPC81, Bessell
& Wood (1984), Blackwell & Petford (1991) and Alonso et
al. (1995). And, finally, the BCK of the model (BCP97), as
computed to derive Tef f (see Sect.3.2.2), were also included.
The residuals (our -- others) vs the (V -- K) color are plot-
ted in Fig.10 for metal rich (lower panel) and metal poor
(upper panel) stars. As can be seen, all the scales agree one
to each other within 0.1 mag. Interesting enough is the op-
posite behaviour of ∆BCK for metal rich and metal poor
stars. While our BCK are systematically larger than the
other scales in metal rich stars, they are systematically lower
in metal poor ones for a given (V -- K). This may reflect the
fact that all the other scales average the information of the
two metallicity regimes.
Concerning BCK as a function of (J -- K), we compared
our scale with those presented by Frogel, Persson & Cohen
(1980), Bessell & Wood (1984), Bessell (1991) and with the
BCK of the model (BCP97).
c(cid:13) 1997 RAS, MNRAS 000, 000 -- 000
Bolometric Corrections and Temperature scales for Population II giants
9
Figure 12. ∆Tef f vs log R and empirical Tef f between our
empirical scales and the theoretical ones by Blackwell & Lynas --
Gray (1994).
Table 4. R -- parameter of the IRFM as a function of Tef f
and BCK .
Tef f
3250
3500
3750
4000
4250
4500
4750
5000
5250
5500
BCK
3.10
2.82
2.65
2.47
2.30
2.14
1.98
1.83
1.69
1.55
log R†
0.5504
0.6613
0.7303
0.7993
0.8676
0.9327
0.9953
1.0555
1.1134
1.1692
Tef f
5750
6000
6250
6500
6750
7000
7250
7500
7750
BCK i
1.41
1.28
1.15
1.03
0.90
0.78
0.66
0.55
0.43
log R†
1.2235
1.2764
1.3278
1.3781
1.4275
1.4757
1.5232
1.5698
1.6158
† logarithmic value of R = Fbol
F2.2µm
in units of µm.
the adopted colors are plotted in Fig.11 for both metal poor
(left panels) and metal rich (right panels) stars. The average
spread among different scales is larger in these planes than
that inferred in the BCK vs (V -- K) plane.
4.3 The IRFM
Using our BCK and empirical Tef f we can also provide an
independent, empirical calibration of the R parameter. This
parameter was computed using the same ZP of our bolomet-
ric scale and the K band flux calibration of Table 2 (columns
#6 and #7). The results are listed in Table 4.
This empirical scale has been compared with the re-
lation obtained interpolating the theoretical Blackwell &
Lynas -- Gray (1994) grid, which tabulated log R as a func-
tion of Tef f and log g. The interpolation in gravity was
performed using the VandenBerg (1996) isochrones
In Fig.12 we plot the residual Tef f between the two
relations as a function of our empirical log R and Tef f . For
log R>1 there is a good agreement (∆Tef f < ± -- 50 K)
among the two relations. For log R<1 the empirical Tef f
seem to be systematically hotter than those of the Blackwell
& Lynas -- Gray (1994) models.
Figure 10. Comparison between our BCK vs (V -- K) scale as
reported in Table 3 and those obtained by other authors for metal
rich (lower panel) and metal poor (upper panel) stars (see keys
in the plot and Sect.4.2).
Figure 11. Comparison between our BCK vs (J -- K), BCI vs
(V -- I) and BCV vs (B -- V) scales and those by other authors for
metal poor (left panels) and metal rich (right panels) stars. Keys
are as in Fig.9, otherwise indicated.
The relation in the BCI vs (V -- I) plane was compared
with those listed by Bessell & Wood (1984), Da Costa &
Armandroff (1990) and Bessell (1991).
Finally, the BCV vs (B -- V) relation was compared with
that obtained by Blackwell & Petford (1991).
5 CONCLUSIONS
The main results obtained from our analysis can be summa-
rized as follow:
The residuals drawn from each individual comparison vs
• By exploiting the use of a large photometric database
c(cid:13) 1997 RAS, MNRAS 000, 000 -- 000
10
P. Montegriffo et al.
of Pop II stars in GGCs we derived new relations for: BCK
vs (V -- K) and (J -- K), BCV vs (B -- V), (V -- I) and (V -- J), to
infer empirical bolometric corrections from observed colors.
• By making use of both an empirical relation and model
atmospheres we calibrated two different scales to infer reli-
able stellar effective temperatures in the range 3000 -- 7500
K.
• We also calibrated an empirical relation between the R
parameter of the IRFM and the stellar temperature.
All these relations are summarized in Table 3, where for a
given BCK the corresponding colors and BCV in two differ-
ent metallicity regimes and Tef f can be read. These rela-
tions should be the most suitable to calibrate the red stellar
sequences in the color -- magnitude diagrams, like the RGB
and AGB in old GGCs.
ACKNOWLEDGMENTS
We wish to thank all the people who kindly made available
to us photometric data, model atmospheres, and isochrones.
We warmly thank Mike Bessell for his very careful read-
ing of the paper, Fiorella Castelli, Giampaolo Di Benedetto,
Maria Lucia Malagnini and Bob Kurucz for helpful discus-
sions. A special thank to the ADC Service to provide the
Morel & Magnenat (1978) Catalogue. Paolo Montegriffo was
supported by a 1996-grant of the Fondazione del Monte,
Rolo Banca 1473. For their financial support, we also thank
the Ministero della Universit`a e della Ricerca Tecnologica
(MURST) and the Agenzia Spaziale Italiana (ASI).
REFERENCES
Armandroff T.E., 1989, AJ 97, 375
Auriere M., Lauzeral C., 1996, Private Communication
Alonso A., Arribas S., Martinez -- Roger C., 1995, A&A 297, 197
Bell R.A., Gustafsson B., 1989, MNRAS 236, 653 (BG89)
Bell R.A., 1992, MNRAS 257, 423
Bergbusch P.A., 1996, AJ 112, 1061
Bergbusch P.A., VandenBerg D.A., 1992, ApJS 81, 163
Bessell M.S., 1990, PASP 102, 1181
Bessell M.S., 1991, AJ 101, 662
Bessell M.S., Brett J.M., 1988, PASP 100, 1134
Bessell M.S., Castelli F., Plez B., 1997, in preparation (BCP97)
Bessell M.S., Wood P.R., 1984, PASP 96, 247
Blackwell D.E., Shallis M.J., Selby M.J., 1979, MNRAS 188, 847
Blackwell D.E., Petford A.D., 1991, A&A 250, 459
Blackwell D.E., Lynas -- Gray A.E., 1994, A&A 282, 899
Buzzoni A., 1989, AJS 71, 817
Buzzoni A., 1996, Private Communication
Carney B.W., Aaronson M., 1979, AJ 84, 867
Castelli F., Gratton R.G., Kurucz R.L., 1997, A&A 318, 841
Castelli F., 1997, Private Communication
Da Costa G.S, Armandroff T.E., 1990, AJ 100, 162
Di Benedetto G.P., 1993, A&A 270, 315
Di Benedetto G.P., Rabbia Y., 1987, A&A 188, 114 (DBR87)
Ferraro F.R., Clementini G., Fusi Pecci F., Buonanno R., 1991,
MNRAS 252, 357
Ferraro F.R., Fusi Pecci F., Guarnieri M.D., Moneti A., Origlia
L., Testa V., 1994a, MNRAS 266, 829
Ferraro F.R., Fusi Pecci F., Guarnieri M.D., Moneti A., Origlia
L., Testa V., Ortolani S., 1994b, in McLean I.S., ed., Infrared
Astronomy with Arrays: The next Generation. Kluwer, Dor-
drecht, p.235
Frogel J.A., Persson S.E., Cohen J.G., 1980, ApJ 239, 495
Frogel J.A., Persson S.E., Cohen J.G., 1981, ApJ 246, 842
(FPC81)
Glass I.S., 1974a, MNRAS South Afr. 33, 54
Glass I.S., 1974b, MNRAS South Afr. 33, 71
Guarnieri M.D., Ortolani S., Renzini A., Montegriffo P., Moneti
A., Barbuy B., Bica E., 1997, A&A (in press)
Johnson H.L., 1966, ARAA 4, 193
Kurucz R.L., 1979, ApJS 40, 1
Kurucz R.L., 1993, ATLAS9 Stellar Atmosphere Programs and 2
km s−1 grid, Kurucz CD -- ROM No 18
Lee S.W., 1977, A&AS 27, 367
Megessier C., 1995, A&A 296, 771
Montegriffo P., Ferraro F.R., Fusi Pecci F., Origlia L., 1995, MN-
RAS 276, 739
Moorwood A.F.M et al., 1992, The Messenger 69, 61
Morel M., Magnenat P. 1978, A&AS 34, 477
Ortolani S., Desidera S., 1996, Private Communication
Ortolani S. et al., 1995, Nature 377, 701
Piotto G., 1996, Private Communication
Plez B., Brett J.M., Nordlund A., 1992, A&A 256, 551
Plez B., 1995, Private Communication
Ridgway S.T., Wells D.C., Joyce R.R., Allen R.G., 1979, AJ 84,
247
Rieke G.H., Lebovsky M.J., 1985, ApJ 288, 618
Stetson P.B., 1994 PASP 106, 250
Tinney C.G., Mould J.R., Reid I.N., 1993, AJ 105, 1045
VandenBerg D.A., 1996, Private Communication
Walker A.R., 1994, AJ 108, 555
Zinn R.J., 1985, ApJ 293, 424
c(cid:13) 1997 RAS, MNRAS 000, 000 -- 000
|
astro-ph/0205313 | 2 | 0205 | 2003-02-21T00:00:07 | Breaking the "Redshift Deadlock" -- II: The redshift distribution for the submillimetre population of galaxies | [
"astro-ph"
] | In this paper we apply our Monte-Carlo photometric-redshift technique, introduced in paper I (Hughes et al. 2002), to the multi-wavelength data available for 77 galaxies selected at 850um and 1.25mm. We calculate a probability distribution for the redshift of each galaxy, which includes a detailed treatment of the observational errors and uncertainties in the evolutionary model. The cumulative redshift distribution of the submillimetre galaxy population that we present in this paper, based on 50 galaxies found in wide-area SCUBA surveys, is asymmetric, and broader than those published elsewhere, with a significant high-z tail for some of the evolutionary models considered. Approximately 40 to 90 per cent of the sub-mm population is expected to have redshifts in the interval 2 < z < 4. Whilst this result is completely consistent with earlier estimates for the sub-mm galaxy population, we also show that the colours of many (< 50 per cent) individual sub-mm sources, detected only at 850um with non-detections at other wavelengths, are consistent with those of starburst galaxies that lie at extreme redshifts, z > 4. Spectroscopic confirmation of the redshifts, through the detection of rest-frame FIR--mm wavelength molecular transition-lines, will ultimately calibrate the accuracy of this technique. We use the redshift probability distribution of HDF850.1 to illustrate the ability of the method to guide the choice of possible frequency tunings on the broad-band spectroscopic receivers that equip the large aperture single-dish mm and cm-wavelength telescopes. | astro-ph | astro-ph | Mon. Not. R. Astron. Soc. 000, 1 -- ?? (2002)
Printed 31 October 2018
(MN LATEX style file v1.4)
Breaking the "Redshift Deadlock" -- II: The redshift
distribution for the submillimetre population of galaxies
Itziar Aretxaga1, David H. Hughes1, Edward L. Chapin1, Enrique Gaztanaga2,1,
James S. Dunlop3, Rob J. Ivison4
1Instituto Nacional de Astrof´ısica, ´Optica y Electr´onica (INAOE), Aptdo. Postal 51 y 216, 72000 Puebla, Mexico
2IEEC/CSIC, Edifici Nexus-201, c/ Gran Capit´an 2-4, 08034 Barcelona, Spain
3Institute for Astronomy, University of Edinburgh, Blackford Hill, Edinburgh, EH9 3HJ, UK
4Astronomy Technology Center, Royal Observatory, Blackford Hill, Edinburgh, EH9 3HJ, UK
31 October 2018
ABSTRACT
Ground-based sub-mm and mm-wavelength blank-field surveys have identified more
than 100 sources, the majority of which are believed to be dusty optically-obscured
starburst galaxies. Colours derived from various combinations of FIR, submillimetre,
millimetre, and radio fluxes provide the only currently available means to determine
the redshift distribution of this new galaxy population.
In this paper we apply our Monte-Carlo photometric-redshift technique, intro-
duced in paper I (Hughes et al. 2002), to the multi-wavelength data available for 77
galaxies selected at 850µm and 1.25 mm. We calculate a probability distribution for
the redshift of each galaxy, which includes a detailed treatment of the observational
errors and uncertainties in the evolutionary model. The cumulative redshift distribu-
tion of the submillimetre galaxy population that we present in this paper, based on 50
galaxies found in wide-area SCUBA surveys, is asymmetric, and broader than those
published elsewhere, with a significant high-z tail for some of the evolutionary models
considered. Approximately 40 to 90 per cent of the sub-mm population is expected
to have redshifts in the interval 2 ≤ z ≤ 4. Whilst this result is completely consis-
tent with earlier estimates for the sub-mm galaxy population, we also show that the
colours of many (<∼ 50 per cent ) individual sub-mm sources, detected only at 850µm
with non-detections at other wavelengths, are consistent with those of starburst galax-
ies that lie at extreme redshifts, z > 4. Spectroscopic confirmation of the redshifts,
through the detection of rest-frame FIR -- mm wavelength molecular transition-lines,
will ultimately calibrate the accuracy of this technique. We use the redshift probabil-
ity distribution of HDF850.1 to illustrate the ability of the method to guide the choice
of possible frequency tunings on the broad-band spectroscopic receivers that equip the
large aperture single-dish mm and cm-wavelength telescopes.
Key words: submillimetre, millimetre, cosmology, galaxy evolution, star-formation
3
0
0
2
b
e
F
1
2
2
v
3
1
3
5
0
2
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
1
INTRODUCTION
The lack of robust redshift information is one of the out-
standing problems in understanding the nature of the pre-
sumed high-redshift galaxies identified in blank-field sub-
mm and mm wavelength surveys.
In paper I of this series (Hughes et al. 2002) we de-
scribed Monte Carlo simulations of 250 -- 500µm surveys from
the balloon-borne telescope BLAST (Devlin et al. 2001)
and the SPIRE instrument on the Herschel satellite, as
well as longer-wavelength ground-based 850µm surveys with
SCUBA. These simulations included a detailed treatment of
the observational and evolutionary model-dependent errors.
The sub-mm colours of galaxies derived from these mock
surveys demonstrated that it will be possible to derive pho-
tometric redshifts, with a conservative r.m.s. accuracy of
∆z ∼ ±0.5, for thousands of optically-obscured galaxies.
Thus we can look forward to breaking the redshift deadlock
which, at the moment, prevents a reliable estimate of the
evolutionary history of a population that contributes a sig-
nificant fraction (>∼ 50 per cent for S850µm > 2 mJy) to the
sub-mm background (Hughes et al. 1998, Smail et al. 2002).
In this second paper we apply the same Monte Carlo
photometric-redshift technique to the existing FIR -- radio
c(cid:13) 2002 RAS
2
Itziar Aretxaga et al.
multi-wavelength data for 77 sources first identified in blank-
field 850µm (SCUBA) and 1.2 mm (MAMBO) surveys. We
calculate the redshift probability distributions of the indi-
vidual sub-mm and mm galaxies, taking into account obser-
vational and model-dependent uncertainties, and thus pro-
vide a measurement of the cumulative redshift distribution
for the blank-field sub-mm galaxy population. This redshift
distribution is completely insensitive to ambiguities in the
identification of the optical counterparts.
The structure of the paper is as follows: section 2 gives a
brief description of the Monte Carlo simulations, the method
to derive photometric redshifts from the radio -- sub-mm -- FIR
colours of galaxies drawn from the mock catalogues, and the
caveats and advantages of this technique over other redshift
indicators; section 3 presents individual redshift probabil-
ity distributions for blank-field sub-mm galaxies. In addi-
tion, we include a brief discussion of the photometric red-
shifts for radio and FIR-selected galaxies subsequently de-
tected at 850µm. Furthermore, we determine the accuracy
of our method through the comparison of the photometric
redshifts for those 8 sub-mm sources that have published
rest-frame optical spectroscopic redshifts. Finally, in sec-
tion 3 we present the cumulative redshift distribution of
the blank-field sub-mm population. Section 4 discusses the
likelihood of a radio detection of previously detected blank-
field sub-mm sources, based only on the known dispersion
of the 850µm/1.4 GHz colour in starburst galaxies, radio-
quiet AGN and ULIRGs, which are believed to be represen-
tative of the high-z sub-mm galaxy population. Section 5
describes how the determination of redshift probability dis-
tributions for individual sub-mm galaxies is motivating the
efforts to provide spectroscopic confirmation of the redshifts
at mm and cm wavelengths. Ultimately, these spectroscopic
redshifts will accurately calibrate the photometric redshifts
derived from rest-frame FIR -- radio-wavelength data. Ap-
pendix A contains the complete catalogue of spectral en-
ergy distributions (SEDs) and redshift distributions derived
in this paper.
2 PHOTOMETRIC REDSHIFT ESTIMATION
TECHNIQUE
The Monte Carlo simulations described in this paper pro-
duce the redshift probability distribution for an individual
galaxy. The advantage of this technique, compared to the
popular maximum-likelihood methods, is that it provides
more information than a simple estimate of the first and
second moments of the redshift distributions. A detailed de-
scription of the technique can be found in paper I; we only
offer a summary here. Hereafter we will refer to a galaxy de-
tected in a SCUBA or MAMBO survey as a sub-mm galaxy,
and a sub-mm galaxy generated in the Monte Carlo simula-
tions as a mock galaxy.
We choose an evolutionary model for the 60µm lumi-
nosity function that fits the observed 850µm number-counts.
Under this model, and assuming a sub-mm survey of ∼10
deg2, we generate a catalogue of 60µm luminosities and red-
shifts. Randomly selected template SEDs are drawn from a
library of local starbursts, ULIRGs and AGN, to provide
FIR -- radio fluxes, and hence colours, for this mock cata-
logue. The fluxes of the mock galaxies include both photo-
metric and calibration errors, consistent with the quality of
the observational data for the sub-mm galaxy detected in a
particular survey. We reject from the catalogue those mock
galaxies that do not respect the detection thresholds and
upper-limits of the particular sub-mm galaxy under analy-
sis. In this paper the redshift probability distribution of a
sub-mm galaxy is calculated as the normalized distribution
of the redshifts of the mock galaxies in the reduced cata-
logue, weighted by the likelihood of identifying the colours
and fluxes of each mock galaxy with those of the sub-mm
galaxy in question. Thus, we also take into account as much
prior information on the sub-mm galaxy population (and
their uncertainties) as possible. For example, we consider
uncertainties in the sub-mm galaxy number counts, their
favoured evolutionary models, the luminosity function of
dusty galaxies, lensing by a foreground magnifying cluster,
etc. In the case that the sub-mm galaxy is identified as a
lensed source, then the mock catalogue is also amplified by
the appropriate factor, allowing intrinsically fainter objects
to be introduced into the reduced catalogue, using the same
observational measurement errors.
We adopt a flat, Λ-dominated cosmological model
(H0 = 67 km s−1 Mpc−1, ΩM = 0.3, ΩΛ = 0.7). To quan-
tify the sensitivity of the individual redshift distributions
on the assumed evolutionary history of the sub-mm galaxy
population, we consider six different models that are able
to reproduce the observed 850µm number-counts within the
uncertainties. We will refer to them by the following codes:
le1: luminosity evolution of the 60µm luminosity function
Φ[L, z = 0] (Saunders et al. 1990) as (1 + z)3.2 for 0 < z <
2.2. Constant evolution is imposed between z = 2.2 − 10. No
galaxies beyond z = 10, Φ[L, z > 10] = 0.
le1L13: same evolutionary description as in le1, but im-
posing a luminosity cut-off above 1013L⊙.
le2: luminosity evolution of Φ[L, 0] as (1 + z)3 for 0 < z <
2.3. Constant evolution is imposed between z = 2.3 − 6.
Φ[L, z > 6] = 0.
le2L13: same evolutionary description as in le2, but im-
posing a luminosity cut-off above 1013L⊙.
lde1 luminosity evolution of Φ[L, 0] as (1 + z)1.9 and den-
sity evolution as (1 + z)2.5 for 0 < z < 3.5; luminosity and
density evolution as (1 + z)−3.0 for 3.5 ≤ z ≤ 7
lde2 luminosity evolution of Φ[L, 0] as (1 + z)3.5 and den-
sity evolution as (1 + z)3 for 0 < z < 2.0; no further lu-
minosity evolution, and density evolution as (1 + z)−2 for
2 ≤ z ≤ 6
The models used in this analysis provide a wide range of
the many possible evolutionary scenarios that reproduce the
sub-mm to mm number counts. These models will allow us
to test the robustness of the photometric redshifts that we
will derive in the next section.
An extended library of SEDs (compared to paper I)
is used in the Monte Carlo simulations. The library con-
tains 20 starbursts, ULIRGs and AGN that have been either
extensively mapped at FIR -- radio wavelengths, or galax-
ies that are sufficiently distant that single-beam photomet-
ric observations measure the total flux: M 82, NGC 3227,
NGC 2992, NGC 4151, NGC 7469, NGC 7771, NGC 1614,
I Zw 1, Mkn 231, Arp 220, Mkn 273, NGC 6240, UGC 5101,
IRAS 10214+4724,
IRAS 08572+3915,
IRAS 14348−1447
IRAS 15250+3609,
IRAS 05189−2524,
IRAS 12112+0305,
c(cid:13) 2002 RAS, MNRAS 000, 1 -- ??
and Cloverleaf. This selection has the important conse-
quence of reducing any dispersion in the SEDs due to aper-
ture effects. A continuum SED that includes synchrotron
and free-free emission in the radio to mm regime, and ther-
mal emission from dust grains in the mm-to FIR regime, is
fitted to the observational data (Chapin et al. 2003). The
luminosities and temperatures of the dust that dominates
the rest-frame FIR emission are evenly distributed in the
ranges 9.0 <∼ log LFIR/L⊙ <∼ 12.3 and 25 < T /K < 65.
Half of the galaxies in the SED library used in the
derivation of photometric redshifts in this paper are in com-
mon with those considered by Yun & Carilli (2002), here-
after YC02. The temperature range covered by our sample
of galaxies (25 − 65K, median 41K) is broader than that of
YC02 due to the increased fraction of cooler sources, and
is, for the same reason, colder than those of well-studied
ULIRGs (46 − 77K, median 60K, Klaas et al. 2001). Despite
the similar temperatures of our sample to late-type galaxies
selected at 60µm (25 − 55K, median 35K), drawn from the
SLUGS catalogue of Dunne et al. (2000), the shapes of the
SEDs in our library differ significantly from the local galax-
ies in the SLUGS. This illustrates an important point: it is
the shapes of the SEDs, and not the derived dust temper-
atures, that influence the estimated photometric redshifts.
Given the limited FIR -- sub-mm data, there is a degeneracy
between the dust temperatures, grain emissivity index (β),
source-size (Ω) and optical depth (τν). This makes it possible
to fit a single observed SED with a broad range of tempera-
ture, by tuning the choice of β, Ω and τν (e.g. Hughes et al
1993).
It remains unclear whether the lower-luminosity, local
galaxies in the SLUGS are closely related to the blank-field
SCUBA population which have rest-frame FIR luminosities
> 1012L⊙, assuming the sub-mm galaxies lie at z > 1. The
claim that cold SEDs match the number-counts in hydro-
dynamical cosmological simulations (e.g. Fardal et al. 2002)
should not be a basis for favouring cold SEDs, since a differ-
ent evolutionary model with warmer SEDs can fit the counts
equally well.
The strongest caveat in any photometric redshift analy-
sis is the validity of the assumption that the SED templates
of local starbursts, ULIRGs and AGN are good analogs of
the high-redshift sub-mm galaxy population. It is not until
future rest-frame FIR -- sub-mm observations from sensitive
balloon-borne or satellite experiments (paper I) accurately
measure the variety of SEDs present in the sub-mm galaxy
population, and spectroscopic redshifts for these are avail-
able, that it will possible to improve the error estimates
presented in this paper.
The technique adopted in this paper differs philosoph-
ically from those developed in other photometric-redshift
studies. Our redshift estimates do not rely on the match of
the observed data to a single template SED, and our errors
are not drawn from the departure of other local SEDs from
the adopted standard template, which has become the com-
mon practice (Carilli & Yun 1999, 2000, hereafter referred to
as CY99, CY00; Gear et al. 2000; YC02). Instead, we allow
the test galaxies to be matched against the whole variety
of template SEDs that can be detected at any given red-
shift. This point should be emphasized, since although we
assign our template SEDs at random, not all SED shapes
(scaled to the same FIR luminosity) are detected with equal
c(cid:13) 2002 RAS, MNRAS 000, 1 -- ??
Breaking the "Redshift Deadlock" -- II
3
probability at any given redshift and wavelength. The red-
shift distributions derived in this paper are often asymmet-
ric, and broader than those inferred from the single SED
template analysis. Despite the efforts of some authors to in-
clude realistic error-bars in their single-template estimates,
we consider these to have been underestimated: error bars
derived from 68% of the galaxies in a template catalogue, is
not equivalent to the 68% dispersion in the fluxes that the
whole catalogue can create, since some of the outliers con-
tribute very significantly to the departures from the mean.
Our redshift estimation method also has the advan-
tage of including a common-sense evaluation of whether a
source at a given redshift is simultaneously consistent with
the observed colours and fluxes. The probability of gener-
ating a certain luminosity at a certain redshift (i.e. flux)
is introduced by the luminosity function and the evolution-
ary model adopted. As shown below, when sub-mm galaxies
with detections at more than two wavelengths are consid-
ered, the results are quite insensitive to the differences in
the evolutionary models adopted to generate the mock cata-
logue. The consideration of reasonable evolutionary models,
however, allow plausible colour-based redshift solutions to
be excluded.
3 MONTE CARLO BASED PHOTOMETRIC
REDSHIFTS
The available data-set for the sub-mm population of galax-
ies is usually restricted to a low-S/N detection at 850µm
and an upper-limit at 450µm (from SCUBA surveys), or a
1.2 mm detection (from MAMBO surveys or an IRAM PdB
follow-up observation), with, perhaps, radio observations at
1.4, 5 or 8.6 GHz. Although a number of the SCUBA surveys
have been conducted in fields previously observed by the In-
frared Space Observatory (ISO) with the ISOPHOT camera
at 170 and 90µm, e.g. HDF (Hughes et al. 1998; Borys et
al. 2002), ELAIS N2 and the Lockman Hole (Scott et al.
2002; Fox et al. 2002), the ISO measurements are generally
too insensitive to provide any additional constraint on the
photometric redshifts.
The method described in section 2 produces mock-
catalogues that replicate the sensitivities and calibration ac-
curacies of the original surveys in which each sub-mm galaxy
was detected. The colours and fluxes of the mock galaxies
provide diagnostic diagrams to illustrate the derivation of
the most probable redshift: colour-colour-redshift (C −C −z)
diagrams, when more than 3-band detections are available;
colour-flux-redshift (C − f − z) diagrams, when just two
bands are available; and flux-redshift (f −z) diagrams; when
just one band detection is available. From these diagrams
we derive the corresponding redshift probability distribu-
tion (P − z) for each sub-mm galaxy. The distributions for
a selection of sub-mm galaxies considered in this paper are
shown in Appendix A.
3.1 Specific examples: LH850.1 and BCR11
LH850.1 is the brightest 850µm source identified in the wide-
area UK 8 mJy SCUBA survey of the Lockman Hole (Scott
et al. 2002), and has one of the most complete FIR -- radio
SEDs (Lutz et al. 2001). BCR11, named after source 11 in
4
Itziar Aretxaga et al.
Figure 1. Colour-colour-redshift (C − C − z) plot for LH850.1
(Lutz et al. 2001, Scott et al. 2002). The flux ratios of the mock
galaxies, generated under the evolutionary model le2 (section 2),
are represented as diamonds, and their redshifts are coded in
colour according to the scale shown to the right of the panel
(colours can be seen in the electronic version of the paper). The
cross represents the measured colours of LH850.1, and the dashed
box shows the 1σ uncertainty in each colour.
the sub-mm catalogue of Barger, Cowie & Richards (2000),
is a low S/N, but nonetheless representative, 850µm source
detected in a follow-up SCUBA photometry survey of the
micro-Jansky 1.4 GHz radio sources in the Hubble Deep
Field (Richards 2000).
Even though BCR11 was not first identified in a blank-
field sub-mm survey, and hence is not included in the com-
bined redshift distribution of the sub-mm selected galaxy
population, it provides a useful source to discuss. The sub-
mm photometry observations of BCR11 at the position of a
known radio source (Richards 2000) guarantees that we have
the correct association of a radio and sub-mm source, and
hence the correct SED to analyse. BCR11 allows us to confi-
dently derive the photometric redshift and accuracy that we
can expect from only the 850µm/1.4 GHz colour, which is
commonly used to measure redshifts of blank-field sub-mm
galaxies with follow-up radio observations (CY99, CY00).
Figs. 1,2 and 3, respectively, show the C − C − z and
P −z distributions for the multi-wavelength data of LH850.1,
and the comparison of the observed SED and template SEDs
(redshifted to the mode of the P − z distribution). Figs. 4,5
and 6 show the same distributions for BCR11. In both ex-
Figure 2. Discrete and cumulative redshift probability distribu-
tions of LH850.1. The six estimates, plotted with lines of differ-
ent style, correspond to the six evolutionary models introduced
in section 2.
Figure 3. The observed SED of LH850.1 normalised to the flux
density at 850µm is shown as squares and arrows. The arrows indi-
cate 3σ upper limits. The squares denote detection at a level ≥ 3σ,
with 1σ error bars. The template SEDs (lines) are redshifted to
z = 2.6, the mode of the redshift probability distribution of most
models considered (see table 1). The template SEDs at this red-
shift compatible within 3σ error bars with the SED of LH850.1
are displayed as darker lines (blue in the electronic version).
c(cid:13) 2002 RAS, MNRAS 000, 1 -- ??
Breaking the "Redshift Deadlock" -- II
5
Figure 4. Colour-flux-redshift (C −f −z) plot for BCR11 (Barger
et al. 2000). The description of the symbols is as in Fig. 1
Figure 5. Redshift probability distributions for BCR11. The de-
scription of the plot is as in Fig. 2
amples (Figs. 2 and 5) the peak of the redshift probability
distribution depends little on the details of the evolution-
ary model used to compute the mock catalogues. In both
cases, the models that invoke pure luminosity evolution (le1,
le1L13, le2, le2L13) have more extended wings, since they
provide a significant population of sources at redshifts be-
yond 3 that can reproduce the colours of the observed galax-
ies. On the other hand, the combined luminosity and den-
sity evolution models (led1, led2) have a very strong de-
cline in the density of sources at redshifts beyond 2 and 3.5,
and thus, the corresponding redshift distributions are much
more concentrated to values between 2 and 3. Model led1
is the most extreme of these two evolutionary models, since
the very strong negative luminosity and density evolution
does not allow any source beyond redshift z = 3.5 to be de-
tected at any reasonable depth. The concentration effect is
more dramatic on BCR11, where the colour constraints are
the weakest, and the wings of the redshift distribution are
strongly dependent on the prior information on the galaxy
population that has been used. The sharpness of this dis-
tribution is, thus, due to the prior, and not to the colour
constraints. In contrast, the sharpness of the distribution of
LH850.1 is intrinsic to the well determined colours of the
source, and is almost independent of the prior.
The comparison of LH850.1 and BCR11 demonstrates
the improved redshift accuracy that, in general, one can
expect for those sources with the greatest combination of
c(cid:13) 2002 RAS, MNRAS 000, 1 -- ??
Figure 6. Comparison of the observed SED of BCR11 with those
of template galaxies, redshifted to z = 2.7 (table 2). The descrip-
tion of the plot is as in Fig. 3. Note that the selection of SEDs to
represent BCR11 is different from that in LH850.1.
6
Itziar Aretxaga et al.
deep multi-wavelength data. In the case of BCR11, it also
illustrates the difficulty of constraining individual redshifts
with observations at 1.4 GHz and 850µm only. The redshift
probability distributions of both sources have very similar
modes (z = 2.6−2.8), yet for the less constraining evolution-
ary models (le1, le2, le1L13, le2L13) there remains 40 -- 50%
and 25 -- 35% probabilities that BCR11 has a redshift > 3
and > 4 respectively. The corresponding probabilities for
LH850.1 are 25-30% and 5% respectively. For models with a
strong decline in the number of sources with redshift (led1,
led2), the high-z tails are not produced. The sharpness of
the derived redshift distributions is due to the lack of bright
sources at high redshift, and not by the colours of the galaxy.
Fig. 7 shows the manner in which the redshift distribu-
tion derived for LH850.1 (using model le2) degrades as dif-
ferent observational constraints are successively taken into
account. When the 450, 850µm, 1.2mm and 1.4GHz detec-
tions are included in the analysis, the addition of upper lim-
its to the SED makes no difference to the derived redshift
distribution. It is also apparent that the distribution does
not change significantly whilst the detection at 450µm is
used in the analysis, in combination with at least one ad-
ditional band in the mm -- sub-mm and radio regime. How-
ever, as soon as the 450µm data is excluded, a high redshift
tail appears. This is due to the large intrinsic scatter in the
1.4GHz/850µm ratio amongst the template SEDs which, in
the absence of the 450µm data, allows galaxies over a wide
range of redshifts to satisfy the colour constraints. The elim-
ination of the radio data from the analysis allows more low-
redshift sources to be possible counterparts of the detected
source, and hence this flattens the resulting redshift distri-
bution.
The improvement in constraint on the photometric red-
shift is extremely important if it is to be used to deter-
mine the tuning of any broad-band mm -- cm spectrograph,
in an attempt to detect redshifted molecular CO-lines. This
is discussed further in section 5. Higher frequency radio ob-
servations at ∼ 5 − 15 GHz and deep 450µm observations,
including sensitive upper-limits, are essential ground-based
data that can provide the additional diagnostic power to
constrain redshifts (e.g. LH850.1).
3.2 Individual redshift distributions
Before we calculate the cumulative redshift-distribution for
the 850µm sources detected in the wide-area (> 200 sq. ar-
cmin) blank-field SCUBA surveys, which include the Lock-
man Hole and ELAIS N2 regions (Scott et al. 2002; Fox et
al. 2002), and the CUDSS fields (Lilly et al. 1999; Eales et
al. 2000; Webb et al. 2002), it is instructive to discuss some
of the typical redshift distributions that can be derived from
data of different quality (i.e. sensitivity and wavelength cov-
erage) for individual sources.
Therefore, we compare the redshift distributions of
wide-area blank-field sub-mm sources with those detected in
various lensing-cluster surveys (Smail et al. 1997; Smail et
al. 1999; Bertoldi et al. 2000; Chapman et al. 2002a). We also
consider radio, FIR and UV-selected objects with follow-up
sub-mm data (Barger et al. 2000; Scott et al. 2000; Chap-
man et al. 2000), and two bright sub-mm sources with ex-
tensive multi-wavelength observations, HDF850.1 (Hughes
et al. 1998; Downes et al. 1999; Dunlop et al. 2002), and the
extremely-red object HR10 (Dey et al. 1999).
Tables 1, 2, and 3 summarize the photometric redshifts
of all the above sources: the modes of their redshift distri-
butions, 68% and 90% confidence intervals derived from the
six models studied in this paper, and other relevant infor-
mation. Appendix A contains the C − C − z, C − f − z, P − z
diagrams and SEDs of the galaxies listed in Tables 1 and 2,
and also a selection of those listed in Table 3 for model le2,
which is one of the less constraining models considered and
serves as an illustration of the points made here.
When we have the most complete multi-wavelength
data-set (rest-frame radio -- sub-mm -- FIR,
including upper
limits) then photometric redshifts can be constrained to a
68% confidence band of width ∆z ∼ ±0.5 (Table 1 and
Fig.A1). Those sub-mm galaxies for which just two detec-
tions, and perhaps some shallow upper-limits are available,
do much worse, with accuracies at a 68% confidence level
of width ∆z >∼ ±1 (Table 2 and Figs. A2). A merit of the
Monte Carlo photometric-redshift method is that it can de-
termine redshift probability distributions even in the case
that the galaxies have just been detected in one band (Ta-
ble 3 and Fig. A3), the result being very flat redshift distri-
butions, that are highly dependent on the assumed evolu-
tionary model. This is especially important because half of
the sub-mm sources in the blank-field surveys belong to this
category, although it is worth noting that several of these
previously reported sub-mm sources have now been found
to be spurious (Ivison et al. 2002).
Often, the well constrained distributions are asymmet-
ric, skewed towards lower redshifts with a high-redshift tail
(e.g. LH850.1, N2850.2, HR10, ... in Figs. 2 and A1). This is
a natural consequence of the ability of the data (particularly
with a deep 450µm detection or upper limit, < 30 mJy) to
strongly reject the possibility of sub-mm galaxies lying at
z <∼ 2, whilst the same observational data cannot reject the
possibility that the sub-mm galaxies lie at a much higher
redshift, z ≫ 3. Examples of the power with which the
450µm detections or upper-limits can reject z <∼ 2 can be
seen in the SEDs of LH850.1 and HDF850.1 (Figs. 3 and
A1), respectively. On the other hand, the bright ISOPHOT
detections at 170µm (> 120mJy) make a strong case that the
sources lie at z <∼ 2, for example N1 -- 40, N1 -- 64 (Fig. A1) and
N1 -- 8 (Fig. A2), whilst the ISOPHOT non-detections are
too shallow to offer any additional constraints on the SEDs
of the majority of the sub-mm population (e.g. LH850.1,
HDF850.1 in Figs. 3 and A1).
It is also possible to obtain a bimodal redshift proba-
bility distribution for some galaxies (e.g. SMMJ00266+1708
in Fig. A1, BCR33 in Fig. A2), an effect created by two or
more template SEDs reproducing the same observed colours
at very different redshifts.
The figures in appendix A show the insensitivity of the
850µm/1.4GHz ratio to redshift when the sources are at
z >∼ 2 (a point also made by CY99, CY00). The redshift dis-
tributions in these cases are usually shallow, with long high-
redshift tails (e.g. BCR11, BCR13, MMJ154127+6615, in
Fig. A2), although exceptions also occur (SMMJ16403+4644
in Fig. 2A). The distributions are even broader in the cases
when only a sub-mm detection and a radio upper-limit are
available (Fig. A3). All these redshift distributions are nar-
rower for those evolutionary models with a strong decline
c(cid:13) 2002 RAS, MNRAS 000, 1 -- ??
Breaking the "Redshift Deadlock" -- II
7
Figure 7. Redshift distributions and compatible template SEDs for LH850.1, when different photometry bands are taken into account.
All estimates are based on a Monte Carlo produced with the same evolutionary model (le2), The SED models represented in the right-hand
panels have been redshifted to the mode of distributions represented in the left. Symbol and line codes are as in Fig. 3.
c(cid:13) 2002 RAS, MNRAS 000, 1 -- ??
8
Itziar Aretxaga et al.
beyond z ∼ 2 − 3, as can be seen from the measurements in
Tables 2 and 3, especially for lde1 which essentially contains
no sources at high-z.
A quick browse through Appendix A illustrates the va-
riety of template SEDs that reproduce the colours of an
individual sub-mm galaxy, which lends support to the use
of a template library in the derivation of the photometric
redshifts. A few sub-mm sources, however, still cannot be
reproduced at any redshift with the template SEDs avail-
able (e.g. LH850.12, SMMJ16403+46437). A possible expla-
nation for this problem is the contribution of AGN emission
to the radio-fluxes at a level not included in the SEDs of
the template library, or a mis-identification of a sub-mm
galaxy with a nearby radio-source (see also section 4). A
wider range of SEDs, including modest fractions of galaxies
with an excess of radio-emission, could partially solve this
problem but, on the other hand, it would also introduce an
extra error term in the redshift computation for the rest
of the sub-mm galaxies. Other sub-mm galaxies have only
a small representation of analogs among the mock galax-
ies (e.g. N1-40, SMMJ14009+0252), most probably because
they are rare objects in the Universe, either by their intrinsi-
cally high luminosity or their low redshift. This latter point
is illustrated by the rarity of mock galaxies in the corre-
sponding C − C − z or C − f − z diagrams, assuming any of
the evolutionary scenarios considered in this paper.
3.2.1 Accuracy of the photometric-redshift method
Fig. 8 shows the comparison of photometric and pub-
lished spectroscopic redshifts for 8 galaxies: in order of in-
creasing spectroscopic redshift, N1 -- 40, CUDSS14.18, N1 --
64,
HR10,
SMMJ14011+0252,
SMMJ02399−0136, and W−MM11.
SMMJ02399 -- 0134,
The two galaxies which depart most from the zphot =
zspec line are SMMJ02399-0134 and W−MM11. The first
has a photometric redshift calculated only on the basis of its
1.4 GHz/850µm colour (Fig. A2), and the latter on a 850µm
detection with a 450µm upper limit (Fig. A3). These two
objects illustrate the large confidence intervals that have to
be considered when the redshifts are derived from a single
colour. Specifically, SMMJ02399−0134 illustrates the uncer-
tainties attached to the large dispersion of 1.4GHz/850µm
colours implied from a local sample of galaxies. The agree-
ment between spectroscopic and photometric redshifts for
the remaining six galaxies (with more than one colour avail-
able) is remarkably good. The asymmetry of the error bars
is a reflection of the skewness of their respective P −z distri-
butions (Figs. A1, A2). These sources illustrate the impor-
tance of obtaining the description of the whole redshift dis-
tribution instead of relying on measurements of the median,
which departs significantly from the most probable redshift
in skewed distributions.
A larger sample of spectroscopic redshifts is obviously
essential to assess the statistical significance of the good-
ness of this technique, and the accuracy of the error bars
we derive. For the sources with more than one measured
colour, these errors are mainly the result of the scatter in
the shapes of the template SEDs. A larger sample of spec-
troscopic redshifts would allow us to fine-tune the selection
of SEDs and priors in order to trim the error bars if, for
instance, the modes of the distributions do not populate
Figure 8. Comparison of the photometric-redshift estimates and
the true redshifts of sub-mm sources with published optical/IR
spectroscopy. The diamonds represent the modes of the redshift
distributions for le2. The solid and dashed error bars represent
68% and 90% confidence intervals, respectively. Sources repre-
sented in black (in increasing redshift: N1 -- 40, CUDSS14.18, N1 --
64, HR10, SMMJ14011+0252, and SMMJ02399−0136) have pho-
tometric redshifts derived from a combination of colours, and are
the most precise. Sources represented in grey (in increasing red-
shift: SMMJ02399−0134, and W−MM11) have photometric red-
shifts derived from only one colour (or limit), and have shallow
probability distributions, still compatible with their spectroscopic
redshifts. Note that the asymmetry of the error-bars is a reflec-
tion of the asymmetry of the redshift probability distributions,
which in cases (CUDSS14.18 and HR10) are significantly skewed.
the envelope of error bars implied from the analysis. The
six best constrained redshift distributions (shown in black
in Fig. 7) do show a tendency to cluster more around the
zspec = zphot line than the collective extension of their error
bars. However, fine-tuning the inputs of the Monte-Carlo at
this early stage of confirmation of the photometric redshifts
is premature.
3.2.2 Notes on individual sources
LH850.1: Other redshift estimates are 2.95±1.49
0.98 based
on the 1.4GHz/850µm spectral index technique; and 2.72 ±
0.37 based on a χ2 minimization with one template SED
(YC02) . Our own estimate z = 2.6±0.4
0.5 (le2) is consistent
with these measurements.
LH850.12: No SED can reproduce the observed pho-
tometry of this source. Ivison et al. (2002) find evidence of
variability at 1.4GHz, and report associated X ray emission.
They conclude that LH850.12 might be a radio-loud QSO.
This kind of objects is actually not represented among our
SEDs, and thus the redshift we derive is unreliable. Note
that there are no counterpart mock-galaxies in the vicinity
of the error box of its C − C − z diagram in Fig. A1.
HDF850.1: Other redshift estimates are > 2.6, based
on the 1.4GHz/850µm spectral index technique; and 4.11 ±
c(cid:13) 2002 RAS, MNRAS 000, 1 -- ??
Breaking the "Redshift Deadlock" -- II
9
Table 1. Sources detected in three or more passbands at a ≥ 3σ level. Column 1 gives the name of the source; column 2 the
references for the SED data; columns 3-8 the mode of the photometric-redshift distributions and 68% confidence intervals for
different evolutionary models, and in parenthesis below, the corresponding 90% confidence intervals; and column 9 notes for each
object. The references are coded as follows: (1) Lutz et al. 2001, (2) Scott et al. 2002, (3) Ivison et al. 2002, (4) Downes et al. 2001,
(5) Dunlop et al. 2002; (6) Eales et al. 2000, (7) Smail et al. 1999, (8) Frayer et al. 2000, (9) Ivison et al. 1998, (10) Ivison et al.
2000, (11) Dey et al. 1999, (12) Chapman et al. 2002b. (13) Cowie et al. 2002 The A in the notes denotes the amplification factor
reported in the literature for the lensed sources.
zle1
phot
zle1L13
phot
zle2
phot
zle2L13
phot
zlde1
phot
zlde2
phot
notes
object
LH850.1
LH850.3
LH850.8
LH850.12
N2850.1
N2850.2
N2850.4
HDF850.1
CUDSS14.1
ref
1,2,3
2,3
2,3
2,3
2,3
2,3
2,3
4,5
6
SMMJ00266+1708
7,8
SMMJ02399−0136
8,9
SMMJ09429+4658
7,13
SMMJ14009+0252
10
SMMJ14011+0252
8
HR10
N1-40
N1-64
11
12
12
2.6±0.4
0.6
(2.0 -- 4.1)
2.7±0.2
(2.1 -- 3.0)
3.8±1.2
1.4
(2.0 -- 6.5)
0.6±0.3
0.1
(0.1 -- 1.0)
2.9±0.6
(2.0 -- 4.1)
2.3±0.6
0.3
(1.5 -- 3.4)
2.8±1.7
0.3
(1.3 -- 4.5)
4.2±0.5
0.7
(3.3 -- 5.5)
3.8±0.2
0.7
(2.5 -- 5.2)
2.7±2.2
0.2
(2.1 -- 5.5)
3.5±0.0
1.0
(2.3 -- 5.0)
2.8±0.7
(2.0 -- 4.5)
2.8±1.7
0.3
(2.0 -- 5.6)
4.4±0.7
0.9
(2.7 -- 5.5)
2.8±0.5
0.8
(2.0 -- 4.5)
1.5±1.0
0.0
(1.5 -- 3.1)
1.2 ± 0.2
(0.6 -- 1.5)
1.2±0.3
0.8
(0.5 -- 1.8)
2.6±0.4
0.5
(2.0 -- 3.8)
2.7±0.2
(1.8 -- 3.0)
3.8±1.3
1.2
(1.6 -- 6.0)
0.6±0.3
0.1
(0.1 -- 1.0)
2.8±0.7
0.4
(2.0 -- 3.7)
2.3±0.5
0.3
(1.5 -- 3.2)
2.7±0.7
(2.0 -- 4.5)
4.1±0.6
(3.1 -- 5.0)
3.8±0.8
0.7
(2.5 -- 5.5)
2.7±2.0
0.2
(2.2 -- 5.5)
3.1±0.8
0.6
(2.0 -- 4.5)
2.9±0.6
(2.0 -- 4.4)
2.8±0.6
(1.9 -- 4.0)
4.1±0.8
0.6
(3.0 -- 5.1)
2.8±0.5
0.8
(2.0 -- 4.5)
1.5±0.5
0.0
(1.5 -- 2.5)
0.7±0.6
0.1
(0.5 -- 1.5)
1.2±0.3
(0.5 -- 1.5)
2.6±0.4
0.5
(2.0 -- 3.7)
3.0±0.0
0.5
(2.1 -- 3.0)
3.7±1.5
0.7
(2.3 -- 6.0)
0.2±0.8
0.2
(0.0 -- 1.3)
2.8±0.7
0.5
(2.0 -- 4.1)
2.3±0.6
0.3
(1.5 -- 3.5)
2.9±1.2
0.9
(1.2 -- 4.5)
4.1±0.6
(3.0 -- 5.1)
3.8±0.7
0.8
(2.0 -- 4.9)
2.7±1.6
0.7
(2.0 -- 5.4)
3.9±0.9
(2.5 -- 5.0)
2.8±1.3
0.3
(2.0 -- 5.0)
2.8±1.5
0.3
(2.0 -- 4.9)
4.1±0.8
(3.1 -- 5.0)
2.7±0.6
0.7
(2.0 -- 4.4)
1.5±0.8
0.0
(1.5 -- 2.9)
0.6±0.6
0.1
(0.5 -- 1.5)
1.2±0.8
0.3
(0.5 -- 2.1)
2.6±0.4
0.5
(2.0 -- 3.9)
2.5±0.0
1.3
(1.0 -- 3.0)
3.8±1.2
0.8
(2.0 -- 5.3)
0.2±0.6
0.2
(0.0 -- 1.2)
2.8±0.3
0.8
(2.0 -- 5.3)
2.3±0.5
0.3
(1.5 -- 3.0)
2.8±0.9
0.8
(1.6 -- 4.5)
4.0±0.7
0.5
(3.0 -- 5.0)
3.7±0.8
(2.2 -- 5.0)
2.8±2.2
0.2
(1.0 -- 5.3)
3.1±0.7
0.6
(2.5 -- 4.6)
2.8±0.7
0.5
(2.0 -- 4.3)
2.8±0.7
0.4
(1.9 -- 4.0)
4.1±0.8
(2.5 -- 5.2)
2.7±0.6
0.7
(2.0 -- 4.5)
1.5±0.7
0.0
(1.5 -- 2.6)
0.6±0.2
(0.3 -- 1.0)
1.2±0.3
0.6
(0.5 -- 1.8)
2.7±0.3
(2.0 -- 3.1)
2.9±0.6
1.2
(1.0 -- 3.5)
3.5±0.0
0.6
(2.6 -- 3.5)
0.7±0.2
(0.5 -- 1.0)
2.9±0.3
0.4
(2.4 -- 3.5)
2.3±0.3
(2.0 -- 3.4)
2.8±0.5
0.3
(2.2 -- 3.5)
3.5±0.0
0.6
(2.7 -- 3.5)
2.9±0.4
(2.1 -- 3.5)
2.8±0.5
0.3
(2.2 -- 3.5)
3.5±0.0
0.8
(2.2 -- 3.5)
2.8±0.4
0.3
(2.0 -- 3.5)
3.0±0.5
0.1
(2.3 -- 3.5)
3.5±0.0
0.9
(2.1 -- 3.5)
2.7±0.6
0.2
(2.3 -- 3.5)
1.5±1.0
0.0
(1.5 -- 3.1)
0.7±0.2
(0.5 -- 1.0)
1.2±0.3
0.6
( -- )
2.7±0.3
0.2
(2.3 -- 3.5)
3.0±0.0
0.5
(2.1 -- 3.0)
3.0±0.0
0.9
(2.0 -- 4.3)
1.0±0.5
(0.8 -- 1.5)
2.7±0.3
0.4
(2.0 -- 3.5)
2.6±0.4
0.1
(2.1 -- 3.0)
2.7±0.3
0.7
(1.5 -- 3.4)
4.0±0.5
0.6
(3.2 -- 5.0)
2.8±1.1
0.8
(2.0 -- 4.8)
2.9±0.5
0.4
(2.5 -- 5.3)
3.8±1.7
zspec = 2.8, A = 2.4,
0.8
(2.4 -- 5.0) AGN, 3.4cm too low?
2.8±0.8
0.3
(2.0 -- 4.4)
2.7±0.3
0.6
(2.0 -- 4.3)
4.1±0.4
1.2
(2.0 -- 5.0) AGN contribution?
2.7±0.3
zspec = 2.57, A = 3.
0.4
(2.0 -- 3.5)
1.5±0.7
0.0
(1.5 -- 2.4)
1.0±0.5
0.0
(0.5 -- 1.5)
1.2±0.6
0.2
(0.7 -- 2.0)
zspec = 1.44
zspec = 0.45
A = 1.5
A = 1.5
A = 3
A = 1.6
zspec = 0.91
radio-loud QSO(3), zphot is
unreliable
disregarding the 3.4cm upper limit
0.51 based on a χ2 minimization with one template SED
(YC02). Our own estimate z = 4.1±0.6
0.5 (le2) is in good
agreement with these measurements, and is unaffected by
the inclusion or exclusion of the possible lensing amplifica-
tion experienced by this source (A ≈ 3, Dunlop et al. 2002).
CUDSS14.1: Other published redshift estimates are
2.01±1.10
0.71, based on the 1.4GHz/850µm spectral index tech-
nique; and 2.06 ± 0.31 based on a χ2 minimization with
one template SED (YC02). Both these estimates are in the
low-redshift tail of our redshift distribution, which at a 68%
significance level places this object at z = 3.8±0.7
0.8 (le2).
CUDSS14.18: The inclusion of the 450 and 850µm
fluxes as detections at 2.3 and 2.8σ (instead of upper limits)
gives a similar distribution to the one shown in Figure A2.
Our redshift estimate places this object at z = 0.7±1.3
0.2 with
a 68% confidence level (le2), in good agreement with its true
spectroscopic redshift, zspec = 0.66. Other redshift estimates
c(cid:13) 2002 RAS, MNRAS 000, 1 -- ??
are 1.12±0.53
0.45, based on the 1.4GHz/850µm spectral index
technique; and 1.11 ± 0.21 based on a χ2 minimization with
one template SED (YC02).
SMMJ00266+1708: Other redshift estimates are
3.49±2.03
1.23 based on the 1.4GHz/850µm spectral index tech-
nique; and 3.50 ± 0.45 based on a χ2 minimization with one
template SED (YC02). Our estimate z = 2.7±2.3
0.2 (le2) is
consistent with these measurements. The redshift distribu-
tion is bimodal, due to the range of SEDs used in the redshift
estimation technique.
SMMJ02399−0136: The complete SED reported in
Frayer et al. (2000), Ivison et al. (1998) is inconsistent with
the SEDs of our template library at any redshift. This is
reported to be an AGN (Smail et al. 1999). One possible
explanation for the failure to detect SMMJ02399-0136 at
3.4cm is a high-frequency break (between 5.5cm and 0.9cm)
in the synchrotron spectrum. Such sharp spectral features
10
Itziar Aretxaga et al.
Table 2. Sources detected in two pass-bands at ≥ 3σ level. The columns are identical to those described in
Table 1. The reference codes are as follows: (1) Scott et al. 2002, (2) Ivison et al. 2002 (3) Barger et al. 2000,
(4) Eales et al. 2000, (5) Webb et al. 2002, (6) Smail et al. 1999, (7) Chapman et al. 2002a, (8) Bertoldi et
al. 2000, (9) Dannerbauer et al. 2002, (10) Chapman et al. 2002b.
object
LH850.6
LH850.7
LH850.14
LH850.16
LH850.18
N2850.7
N2850.13
BCR7
BCR11
BCR13
BCR33
BCR49
CUDSS14.3
CUDSS14.9
CUDSS14.18
CUDSS3.1
CUDSS3.8
CUDSS3.9
CUDSS3.10
CUDSS3.15
SMMJ02399−0134
SMMJ10236+0412
SMMJ10237+0410
SMMJ16403+4644
SMMJ16403+46440
SMMJ16403+46437
MMJ154127+6615
MMJ154127+6616
MMJ120546−0741.5
MMJ120539−0745.4
ref
1,2
1,2
1,2
1,2
1,2
1,2
1,2
3
3
3
3
3
4
4
4
5
5
5
5
5
6
7
7
7
7
7
8
8
9
9
N1−008
10
zle1
phot
zle1L13
phot
zle2
phot
zle2L13
phot
zlde1
phot
zlde2
phot
notes
2.7±2.0
0.7
(2.0 -- 6.4)
3.6±1.9
0.8
(1.6 -- 5.5)
2.5±1.9
0.4
(1.5 -- 6.3)
3.6±1.0
1.6
(1.5 -- 6.0)
3.2±1.1
(1.5 -- 5.8)
2.8±1.5
0.8
(1.9 -- 5.5)
2.7±0.8
1.0
(1.5 -- 5.0)
2.7±0.8
1.1
(1.5 -- 4.8)
2.6±0.9
(1.5 -- 4.7)
2.3±1.9
0.3
(1.5 -- 5.5)
2.9±1.0
(1.5 -- 5.2)
2.3±1.1
0.5
(1.6 -- 4.5)
3.1±0.9
(1.7 -- 5.0)
2.6±0.8
0.6
(1.5 -- 4.1)
0.7±0.8
0.4
(0.0 -- 1.8)
2.1±0.4
0.6
(1.0 -- 2.5)
2.1±0.9
0.2
(1.4 -- 3.0)
1.0±0.2
0.5
(0.5 -- 1.5)
2.6±0.8
(1.4 -- 4.5)
2.2±0.7
(1.0 -- 3.7)
2.2±0.7
(1.0 -- 3.6)
2.3±0.8
(1.3 -- 4.0)
2.0±1.1
0.5
(1.0 -- 3.5)
2.2±0.8
(0.5 -- 3.1)
2.8±0.7
(1.5 -- 4.1)
2.2±0.8
0.6
(1.0 -- 3.3)
2.8±1.7
0.5
(2.0 -- 6.0)
2.8±1.7
0.3
(2.0 -- 6.0)
4.7±1.3
1.9
(2.5 -- 7.8)
2.8±1.7
0.6
(2.0 -- 6.3)
0.0±0.5
0.0
(0.0 -- 0.5)
2.7±2.0
0.7
(2.0 -- 6.2)
3.5±0.5
1.8
(1.5 -- 5.1)
2.4±2.0
0.4
(1.5 -- 5.5)
3.5±1.0
1.5
(1.5 -- 6.1)
3.4±1.0
1.4
(1.5 -- 5.6)
2.8±1.2
0.6
(2.0 -- 5.3)
3.0±0.5
1.5
(1.5 -- 5.0)
2.7±0.8
1.0
(1.5 -- 4.7)
2.7±0.8
(1.5 -- 4.7)
2.3±2.0
0.3
(1.5 -- 5.6)
3.0±1.0
(1.5 -- 5.3)
2.4±0.5
(1.5 -- 4.2)
3.1±0.9
(1.5 -- 4.8)
2.6±0.9
0.6
(1.5 -- 4.0)
0.7±0.8
0.4
(0.0 -- 1.8)
1.8±0.7
0.3
(0.7 -- 2.5)
2.0±0.5
(1.4 -- 3.0)
1.0±0.4
0.0
(1.0 -- 2-0)
2.6±0.9
(1.3 -- 4.5)
2.3±0.7
(1.5 -- 4.0)
2.3±0.8
(1.4 -- 4.0)
2.3±0.8
(1.3 -- 4.0)
2.0±1.2
0.5
(1.0 -- 3.7)
2.2±0.8
(0.8 -- 3.5)
2.8±0.7
(1.5 -- 4.1)
2.3±0.7
0.8
(1.1 -- 3.5)
3.6±0.9
1.4
(2.0 -- 5.9)
2.8±2.0
0.3
(2.0 -- 6.0)
3.4±2.7
1.9
(2.5 -- 7.0)
2.8±1.7
0.8
(2.0 -- 6.0)
0.0±0.5
0.0
(0.0 -- 0.5)
2.7±1.8
0.7
(2.0 -- 5.4)
3.6±0.4
1.9
(1.5 -- 5.3)
2.4±1.9
0.4
(2.0 -- 5.4)
3.3±1.1
1.3
(2.0 -- 5.5)
3.0±1.1
1.0
(1.5 -- 4.9)
3.0±1.0
(1.5 -- 4.9)
2.7±0.8
1.0
(1.5 -- 4.6)
2.2±1.1
0.7
(1.5 -- 4.8)
2.7±0.8
0.9
(1.5 -- 4.6)
2.3±1.7
0.3
(1.5 -- 4.7)
3.0±1.0
(1.5 -- 4.6)
2.4±1.1
0.6
(1.0 -- 4.0)
3.2±0.8
1.0
(2.0 -- 5.0)
2.5±0.5
1.0
(1.5 -- 4.5)
0.7±0.8
0.4
(0.0 -- 1.8)
1.7±0.7
0.2
(1.5 -- 2.5)
2.0±0.8
0.5
(1.0 -- 3.0)
1.1±0.8
0.1
(0.5 -- 2.0)
2.5±1.0
(1.5 -- 4.5)
2.2±0.9
0.7
(1.5 -- 4.0)
2.3±0.8
(1.0 -- 3.8)
2.3±0.8
(1.5 -- 4.0)
2.1±1.0
0.6
(1.0 -- 3.6)
2.1±0.4
1.1
(1.0 -- 3.5)
2.7±0.8
(1.5 -- 4.1)
2.6±0.4
1.1
(1.0 -- 3.3)
2.8±1.7
0.3
(2.0 -- 5.0)
2.8±1.8
0.3
(2.5 -- 5.0)
3.8±2.1
0.3
(3.0 -- 6.0)
2.8±1.7
0.3
(2.0 -- 4.9)
0.0±0.4
0.0
(0.0 -- 0.5)
2.7±1.8
0.7
(2.0 -- 5.4)
3.5±0.5
1.9
(1.5 -- 5.3)
2.3±2.0
0.3
(2.0 -- 5.4)
3.6±0.8
1.5
(2.0 -- 5.4)
3.2±1.0
1.2
(1.8 -- 4.9)
3.0±0.9
(1.7 -- 5.0)
2.8±0.7
1.2
(1.5 -- 4.8)
2.7±0.6
1.2
(1.5 -- 4.7)
2.6±0.7
1.1
(1.5 -- 4.7)
2.3±1.7
0.3
(1.5 -- 4.7)
3.0±1.0
(1.5 -- 4.5)
2.5±1.0
0.5
(1.0 -- 3.7)
3.1±0.9
(1.7 -- 5.0)
2.6±0.4
1.1
(1.5 -- 4.0)
0.7±1.1
0.2
(0.0 -- 2.0)
1.8±0.5
0.3
(0.9 -- 2.5)
2.0±0.5
(1.2 -- 3.0)
0.7±0.3
0.5
(0.0 -- 1.4)
2.5±0.9
(1.0 -- 4.2)
2.3±0.7
(1.0 -- 3.7)
2.3±0.8
(1.0 -- 3.7)
2.3±0.8
(1.3 -- 4.0)
2.0±1.1
0.5
(1.0 -- 3.6)
2.2±0.7
(0.5 -- 3.5)
2.8±0.7
(1.5 -- 4.1)
2.6±0.4
1.0
(1.2 -- 3.5)
3.7±0.8
1.2
(2.0 -- 5.0)
2.8±2.2
0.3
(2.1 -- 5.0)
3.8±2.0
0.3
(2.5 -- 6.0)
2.5±1.7
0.5
(2.0 -- 5.3)
0.0±0.4
0.0
(0.0 -- 0.5)
2.7±0.6
0.2
(2.3 -- 3.5)
3.5±0.0
1.0
(1.7 -- 3.5)
2.7±0.3
0.5
(2.0 -- 3.4)
2.8±0.5
0.3
(2.0 -- 3.5)
3.2±0.3
0.7
(2.0 -- 3.5)
3.0±0.3
0.5
(2.5 -- 3.5)
2.8±0.5
0.3
(2.0 -- 3.5)
2.8±0.5
0.3
(1.9 -- 3.5)
2.7±0.6
0.2
(2.1 -- 3.5)
2.8±0.7
0.3
(2.2 -- 3.5)
3.0±0.5
0.3
(2.0 -- 3.5)
2.7±0.3
0.7
(1.7 -- 3.5)
3.5±0.0
0.8
(2.2 -- 3.5)
2.7±0.8
0.2
(2.0 -- 3.4)
0.7±0.7
(0.0 -- 1.9)
1.9±0.4
(1.5 -- 2.6)
2.6±0.4
0.2
(2.0 -- 3.0)
1.0±0.4
0.0
(1.0 -- 1.5)
2.7±0.8
0.2
(1.9 -- 3.5)
2.5±0.5
(1.9 -- 3.5)
2.6±0.7
0.5
(1.7 -- 3.5)
2.7±0.3
0.7
(2.0 -- 3.5)
2.6±0.6
(1.7 -- 3.5)
2.4±0.6
0.8
(1.2 -- 3.5)
2.8±0.6
0.3
(2.1 -- 3.5)
2.6±0.5
(1.8 -- 3.5)
2.7±0.3
(2.0 -- 3.1)
3.0±0.5
0.4
(2.2 -- 3.5)
3.0±0.2
0.5
(2.5 -- 3.4)
2.5±0.3
(2.0 -- 3.3)
0.0±0.4
0.0
(0.0 -- 0.5)
2.8±0.2
0.6
(2.0 -- 4.4)
2.5±0.2
0.5
(1.5 -- 3.8)
3.0±0.0
0.7
(2.0 -- 3.5)
3.5±0.0
0.8
(1.5 -- 3.5)
3.0±0.0
0.7
(2.0 -- 3.3)
2.5±0.5
(2.0 -- 4.3)
2.5±0.3
0.5
(2.0 -- 3.5)
2.4±0.4
(1.5 -- 3.2)
2.4±0.4
(1.5 -- 3.0)
2.9±0.1
0.6
(2.0 -- 3.3)
3.0±0.0
0.7
(2.0 -- 3.0)
2.8±0.2
0.6
(1.6 -- 3.0)
3.0±0.0
0.8
(2.2 -- 4.1)
2.8±0.2
0.6
(1.9 -- 3.5)
1.2±0.5
0.7
(0.2 -- 2.0)
1.8±0.4
0.3
(1.2 -- 2.5)
1.9±0.4
(1.2 -- 2.5)
1.3±0.5
0.3
(0.6 -- 2.0)
2.5±0.0
0.9
(1.0 -- 2.5)
2.5±0.0
0.7
(1.2 -- 2.5)
2.5±0.0
0.8
(1.2 -- 2.5)
2.8±0.2
0.5
(1.7 -- 3.0)
2.4±0.4
(1.6 -- 3.0)
2.5±0.4
(1.6 -- 3.0)
2.9±0.1
0.7
(1.5 -- 3.1)
2.5±0.5
(1.5 -- 3.0)
2.8±0.3
(2.0 -- 4.1)
2.9±1.0
0.4
(2.6 -- 5.3)
3.5±1.0
(2.5 -- 6.0)
2.8±0.2
0.6
(2.0 -- 3.7)
0.0±0.4
0.0
(0.0 -- 0.5)
zspec = 0.66
high 1.4GHz, AGN?
A = 2.5, zspec = 1.06
A = 3.2
A = 3.7
A = 3.0
A = 3.6
A = 3.2, 1.4GHz high, AGN?
A = 1
A = 1
c(cid:13) 2002 RAS, MNRAS 000, 1 -- ??
Breaking the "Redshift Deadlock" -- II
11
Table 3. Sources detected at 850µm or 1.2mm at a ≥ 3.5σ level, and undetected at any other wavelength.
The columns are identical to those described in Table 1. The reference codes are as follows: (1) Scott et
al. 2002, (2) Eales et al. 2000, (3) Webb et al. 2002, (4) Smail et al. 1999, (5) Chapman et al. 2002a, (6)
Dannerbauer et al. 2002. (7) Ivison et al. 2002
object
LH850.2
LH850.4
LH850.5
LH850.13
N2850.3
N2850.5
N2850.6
N2850.8
N2850.9
N2850.10
N2850.11
CUDSS14.2
CUDSS14.4
CUDSS14.5
CUDSS14.6
CUDSS14.8
CUDSS14.10
CUDSS3.2
CUDSS3.3
CUDSS3.4
CUDSS3.5
CUDSS3.6
CUDSS3.7
CUDSS3.11
CUDSS3.12
CUDSS3.13
CUDSS3.16
SMMJ02400−0134
SMMJ04431+0210 (N4)
SMMJ04541−0302
c(cid:13) 2002 RAS, MNRAS 000, 1 -- ??
ref
1,7
1,7
1,7
1,7
1,7
1,7
1,7
1,7
1,7
1,7
1,7
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
5
zle1
phot
zle1L13
phot
zle2
phot
zle2L13
phot
zlde1
phot
zlde2
phot
notes
5.9±1.3
2.4
(3.5 -- 9.0)
5.5±2.0
1.8
(3.0 -- 8.8)
5.3±2.2
1.3
(3.3 -- 9.0)
4.9±2.6
1.4
(3.0 -- 8.8)
4.8±2.8
0.8
(3.5 -- 9.0)
5.0±2.0
(2.5 -- 8.6)
5.0±2.0
(2.5 -- 8.7)
4.0±2.7
1.5
(2.1 -- 8.5)
5.2±2.0
1.7
(3.0 -- 9.0)
4.7±2.3
1.9
(2.1 -- 8.5)
5.5±1.9
(3.0 -- 8.9)
5.2±2.4
1.7
(2.5 -- 8.7)
5.0±2.6
1.5
(2.5 -- 8.8)
4.8±2.6
1.3
(2.8 -- 9.0)
4.7±2.8
1.5
(2.5 -- 8.8)
4.7±2.6
1.7
(2.5 -- 9.0)
4.7±2.6
1.7
(2.5 -- 9.0)
4.5±1.8
2.2
(2.0 -- 8.5)
3.8±2.5
1.5
(2.0 -- 8.4)
3.2±2.9
1.2
(2.0 -- 8.5)
3.7±2.6
1.7
(1.8 -- 8.5)
3.7±2.6
1.7
(2.0 -- 8.7)
3.7±2.6
1.2
(2.0 -- 8.4)
3.8±3.0
1.3
(1.9 -- 8.5)
3.7±2.6
1.7
(1.9 -- 8.5)
4.0±2.3
2.0
(1.8 -- 8.5)
3.5±2.9
1.5
(1.6 -- 8.5)
5.0±2.6
1.5
(3.0 -- 9.1)
4.5±2.5
1.6
(2.5 -- 8.8)
4.7±2.0
2.2
(2.2 -- 8.5)
5.5±1.5
1.9
(3.3 -- 8.5)
5.3±2.2
1.4
(3.0 -- 8.5)
6.1±1.5
1.7
(3.0 -- 8.5)
5.1±2.1
1.6
(3.0 -- 8.4)
4.6±2.9
0.6
(3.5 -- 8.8)
4.7±2.3
1.4
(2.7 -- 8.5)
4.8±1.9
2.3
(2.5 -- 8.3)
4.7±2.0
2.2
(2.0 -- 8.4)
5.0±2.0
1.8
(2.8 -- 8.5)
4.5±2.5
1.7
(2.1 -- 8.5)
4.62.9
0.9
(3.0 -- 8.7)
5.0±2.5
1.6
(2.5 -- 8.8)
5.1±2.5
1.6
(2.5 -- 8.8)
4.9±2.7
1.4
(2.8 -- 9.0)
4.7±3.0
1.2
(2.5 -- 8.8)
4.7±2.0
2.2
(2.5 -- 9.0)
4.8±2.5
1.8
(2.5 -- 9.0)
3.8±2.7
1.6
(1.9 -- 8.5)
3.7±2.4
1.7
(2.0 -- 8.4)
3.7±2.8
1.7
(2.0 -- 8.3)
3.7±2.6
1.4
(1.8 -- 8.5)
4.0±2.4
2.0
(2.0 -- 8.8)
3.7±2.6
1.2
(2.0 -- 8.1)
4.5±1.7
2.5
(2.0 -- 8.5)
3.2±2.1
1.2
(1.9 -- 8.5)
3.2±2.2
1.5
(1.7 -- 8.5)
3.2±2.2
1.2
(1.6 -- 8.5)
5.2±2.3
1.7
(3.0 -- 9.1)
4.6±2.2
1.6
(2.5 -- 8.7)
4.0±2.6
1.5
(2.3 -- 8.5)
5.8±0.2
1.7
(3.3 -- 6.0)
4.7±1.3
0.5
(3.5 -- 6.0)
4.7±1.3
0.5
(3.4 -- 6.0)
5.0±1.0
(3.0 -- 6.0)
5.8±0.2
1.4
(3.6 -- 6.0)
5.0±0.7
1.5
(2.9 -- 6.0)
4.8±0.8
1.3
(3.0 -- 6.0)
4.5±1.0
1.5
(2.6 -- 6.0)
5.9±0.1
1.8
(3.2 -- 6.0)
4.7±0.8
1.6
(2.6 -- 6.0)
5.0±1.0
(3.2 -- 6.0)
4.7±0.8
1.0
(2.9 -- 6.0)
4.6±1.4
0.8
(2.9 -- 6.0)
4.8±1.2
0.8
(3.0 -- 6.0)
5.2±0.8
1.3
(2.9 -- 6.0)
4.5±1.3
1.0
(2.6 -- 6.0)
3.9±2.0
0.4
(2.6 -- 6.0)
3.5±1.7
1.0
(2.0 -- 5.8)
4.0±1.2
1.5
(2.2 -- 6.0)
4.3±0.9
1.2
(2.3 -- 6.0)
3.8±1.5
1.3
(2.0 -- 5.9)
3.7±1.5
1.2
(2.0 -- 5.8)
4.0±1.5
1.2
(2.3 -- 6.0)
3.8±1.7
1.0
(2.3 -- 6.0)
3.5±1.8
1.0
(2.0 -- 5.8)
3.5±1.8
1.0
(2.0 -- 5.8)
3.7±1.5
1.2
(2.1 -- 6.0)
4.9±1.1
0.9
(3.1 -- 6.0)
4.5±1.4
1.0
(2.7 -- 6.0)
4.7±1.2
(2.6 -- 6.0)
5.1±0.4
1.5
(3.3 -- 6.0)
4.6±1.4
0.5
(3.3 -- 6.0)
6.0±0.0
1.9
(3.4 -- 6.0)
5.0±1.0
(3.3 -- 6.0)
4.8±0.9
(3.6 -- 6.0)
4.5±1.5
0.6
(2.9 -- 6.0)
4.8±1.2
(2.5 -- 6.0)
4.8±1.2
(2.6 -- 6.0)
5.0±1.0
(3.1 -- 6.0)
4.8±1.2
(2.6 -- 6.0)
4.81.2
0.8
(3.2 -- 6.0)
5.0±0.7
1.5
(2.8 -- 6.0)
4.7±1.3
0.9
(2.8 -- 6.0)
5.3±0.7
1.2
(3.0 -- 6.0)
5.2±0.8
1.3
(2.9 -- 6.0)
4.7±1.9
1.2
(2.6 -- 6.0)
4.3±1.2
(2.6 -- 6.0)
3.9±1.4
(2.2 -- 6.0)
4.0±1.3
1.5
(2.0 -- 5.8)
3.9±1.4
(2.2 -- 6.0)
3.9±1.4
(2.0 -- 5.8)
3.9±1.4
(2.0 -- 5.8)
3.7±1.3
(2.0 -- 5.7)
3.7±1.8
0.9
(2.0 -- 5.7)
3.9±1.3
(2.0 -- 5.8)
3.9±1.4
(2.1 -- 6.0)
3.5±1.9
1.0
(2.0 -- 5.9)
5.2±0.8
1.2
(3.1 -- 6.0)
4.2±1.6
0.7
(2.7 -- 6.0)
4.7±0.8
1.6
(2.6 -- 6.0)
3.5±0.0
0.5
(2.7 -- 3.5)
3.5±0.0
0.6
(2.6 -- 3.5)
3.5±0.0
0.6
(2.6 -- 3.5)
3.5±0.0
0.6
(2.6 -- 3.5)
3.5±0.0
0.5
(2.7 -- 3.5)
3.5±0.0
0.7
(2.5 -- 3.5)
3.5±0.0
0.8
(2.5 -- 3.5)
3.0±0.5
0.3
(2.3 -- 3.5)
3.5±0.0
0.6
(2.6 -- 3.5)
3.5±0.0
0.8
(2.3 -- 3.5)
3.5±0.0
0.7
(2.6 -- 3.5)
3.5±0.0
0.7
(2.5 -- 3.5)
3.5±0.0
0.8
(2.4 -- 3.5)
3.5±0.0
0.8
(2.5 -- 3.5)
3.5±0.0
0.8
(2.5 -- 3.5)
3.5±0.0
0.8
(2.3 -- 3.5)
2.9±0.5
(2.3 -- 3.5)
2.9±0.5
(2.2 -- 3.5)
2.9±0.4
(2.1 -- 3.5)
2.9±0.4
(2.2 -- 3.5)
2.9±0.5
(2.1 -- 3.5)
2.9±0.4
(2.2 -- 3.5)
2.8±0.5
0.3
(2.0 -- 3.5)
2.9±0.4
(2.2 -- 3.5)
2.9±0.5
(2.2 -- 3.5)
2.9±0.5
(2.1 -- 3.5)
2.9±0.4
(2.2 -- 3.5)
3.5±0.0
0.7
(2.5 -- 3.5)
3.5±0.0
0.7
(2.5 -- 3.5)
3.0±0.5
0.3
(2.4 -- 3.5)
4.3±1.2
0.8
(3.2 -- 6.0)
3.9±1.7
0.4
(3.0 -- 6.0)
4.5±1.0
(3.0 -- 6.0)
4.6±1.2
(2.7 -- 6.0)
4.3±1.0
(3.2 -- 6.0)
2.5±2.0
0.5
(3.0 -- 6.0)
2.5±1.9
0.5
(2.0 -- 5.6)
2.5±0.3
1.0
(2.5 -- 6.0)
3.7±1.5
0.7
(2.5 -- 6.0)
2.5±0.5
1.0
(1.5 -- 4.8)
4.1±1.2
(2.6 -- 6.0)
2.5±1.2
1.0
(1.5 -- 5.1)
2.5±1.2
1.0
(1.5 -- 5.2)
2.5±1.9
0.5
(2.0 -- 5.7)
2.5±1.4
1.0
(1.5 -- 5.3)
2.5±0.8
1.0
(1.5 -- 5.0)
2.5±0.5
1.0
(1.5 -- 4.8)
2.5±0.0
0.9
(1.0 -- 4.0)
2.5±0.0
0.9
(1.0 -- 4.0)
2.5±0.0
1.0
(1.0 -- 4.0)
2.5±0.0
1.0
(1.0 -- 3.9)
2.5±0.0
1.0
(1.0 -- 3.9)
2.5±0.0
1.0
(1.4 -- 4.5)
2.5±0.0
1.0
(1.0 -- 4.1)
2.5±0.0
1.0
(1.0 -- 3.9)
2.5±0.0
1.0
(1.0 -- 4.0)
2.5±0.0
1.0
(1.0 -- 3.8)
4.0±0.7
2.0
(2.0 -- 5.5)
2.4±1.0
(1.5 -- 4.9)
2.5±0.6
1.0
(1.5 -- 4.7)
A > 1.9 (2.5?)
A > 1.
A = 2.6
12
Itziar Aretxaga et al.
Table 3. (cont.)
object
ref
zle1
phot
zle1L13
phot
zle2
phot
zle2L13
phot
zlde1
phot
zlde2
phot
notes
SMMJ04542−0301
SMMJ04543+0256
SMMJ04543+0257
SMMJ10237+0412
SMMJ14573+2220
SMMJ17223+3207
SMMJ22471−0206
MMJ120517−0743.1
W−MM11
5
5
5
5
5
5
4
6
5
4.7±2.2
(2.5 -- 8.9)
3.9±2.7
1.4
(2.0 -- 8.4)
4.7±2.2
1.7
(2.0 -- 8.2)
4.0±2.9
1.5
(2.0 -- 8.7)
4.0±2.7
1.5
(2.0 -- 8.5)
4.0±2.5
1.5
(2.0 -- 8.4)
4.8±2.4
1.8
(2.5 -- 8.8)
7.0±1.5
2.4
(4.3 -- 10.)
4.2±2.0
2.2
(2.0 -- 8.7)
4.3±2.7
1.5
(2.1 -- 8.5)
3.8±2.7
1.6
(2.0 -- 8.5)
4.5±2.1
1.5
(2.5 -- 8.3)
3.9±3.0
1.4
(2.0 -- 8.7)
4.2±2.6
1.7
(2.0 -- 8.6)
4.2±2.6
1.7
(2.0 -- 8.4)
4.8±2.3
1.8
(2.5 -- 8.8)
7.1±1.4
2.6
(4.3 -- 10.)
3.4±2.9
1.5
(2.0 -- 8.6)
4.2±1.7
0.7
(2.7 -- 6.0)
4.0±1.5
1.1
(2.4 -- 6.0)
4.2±1.3
1.1
(2.6 -- 6.0)
4.5±0.7
2.0
(2.0 -- 6.0)
4.0±1.6
1.0
(2.0 -- 6.0)
3.9±1.6
1.0
(2.4 -- 6.0)
4.7±1.3
0.9
(2.8 -- 6.0)
6.0±0.0
1.6
(3.6 -- 6.0)
3.7±1.6
1.2
(2.0 -- 5.8)
4.5±1.4
1.0
(2.6 -- 6.0)
4.0±1.5
1.1
(2.0 -- 6.0)
4.5±1.0
1.4
(2.6 -- 6.0)
4.7±0.8
1.8
(2.6 -- 6.0)
4.2±1.8
0.8
(2.5 -- 6.0)
4.0±1.1
1.5
(2.5 -- 6.0)
4.9±1.1
(2.8 -- 6.0)
6.0±0.0
1.6
(3.5 -- 6.0)
3.9±1.6
1.2
(2.2 -- 6.0)
3.3±0.2
0.6
(2.4 -- 3.5)
2.9±0.4
(2.0 -- 3.5)
3.5±0.0
0.8
(2.3 -- 3.5)
2.9±0.4
(2.2 -- 3.5)
2.9±0.4
(2.2 -- 3.5)
3.0±0.3
0.5
(2.2 -- 3.5)
3.4±0.1
0.6
(2.5 -- 3.5)
3.5±0.0
0.4
(2.7 -- 3.5)
2.9±0.4
(2.0 -- 3.5)
2.5±0.9
1.0
(1.5 -- 4.9)
2.5±0.0
1.0
(1.5 -- 4.5)
2.5±0.7
1.0
(1.5 -- 4.8)
2.5±0.0
1.0
(1.0 -- 4.1)
2.5±0.1
1.0
(1.5 -- 4.7)
2.5±0.1
1.1
(1.5 -- 4.6)
2.9±1.3
1.0
(1.5 -- 5.2)
4.4±1.4
0.4
(3.5 -- 6.0)
2.5±0.0
0.9
(1.0 -- 4.0)
A = 4.5
A = 3.1
A = 3.4
A = 3.5
A = 2.8
A = 2.8
A > 1.9 (2.5?)
zspec = 2.98
are not taken into account in the fitting of the radio spec-
trum. We note however that excluding the 3.4cm upper-limit
from the modelling of the multi-wavelength data provides a
redshift estimate z = 2.8±1.3
0.3 (le2) that is in good agreement
with the true redshift of SMMJ02399−0136: z = 2.80. Other
redshift estimates are 1.65±0.80
0.53 based on the 1.4GHz/850µm
spectral index technique; and 2.83 ± 0.38 based on a χ2 min-
imization with one template SED (YC02).
SMMJ09429+4658 (H5): Other redshift estimates
are ≥ 3.6 based on the 1.4GHz/850µm spectral index tech-
nique; and 3.86 ± 0.49 based on a χ2 minimization with one
template SED (YC02). Our estimates actually place this
source at a slightly lower redshift z = 2.4±1.5
0.4 (le2). Given
the sub-mm detection, the low flux density at 1.4 GHz ex-
cludes almost all of the template SEDs in our library.
SMMJ14009+0252: Other redshift estimates are
1.27±1.59
0.54 based on the 1.4GHz/850µm spectral index tech-
nique; and 1.30±0.23 based a on a χ2 minimization with one
template SED (YC02). However large discrepancies (∼ 3σ)
exist between the template SED of YC02 and the observed
450µm and 1.4GHz fluxes, suggesting the fit will have a high
reduced-χ2, and that their errors (inferred for good matches)
might be under-estimated. The flattening of the spectrum
at 450µm, relative to the Raleigh-Jeans tail defined by the
850µm and 1.35mm fluxes, supports a higher redshift esti-
mate. Our own fit indicates that this object has a redshift
z ∼ 4.1 ± 0.8 (le2). The discrepancies between different red-
shift estimators might be attributed to the possible AGN
nature of this object (Ivison et al. 2000).
SMMJ14011+0252: Other redshift estimates are
2.53±1.24
0.82 based on the 1.4GHz/850µm spectral index tech-
nique; and 2.73 ± 0.37 based on a χ2 minimization with one
template SED (YC02). Our own estimate is z = 2.7±0.6
0.7
(le2) in good agreement with these estimates, and with the
spectroscopic redshift of the optical counterpart zspec =
2.57.
HR10: Other redshift estimates are 1.49±0.75
0.53 based on
the 1.4GHz/850µm spectral index technique; and 1.75±0.28
based on a χ2 minimization with one template SED (YC02).
Our own estimate z = 1.5±0.9
0.0 (le2) is in good agreement
with their measurements, and with the true spectroscopic
redshift of this source zspec = 1.44.
SMMJ02399−0134: Other redshift estimates are
1.17±0.53
0.45 based on the 1.4GHz/850µm spectral index tech-
nique; and 0.93 ± 0.19 based on a χ2 minimization with one
template SED (YC02). The spectroscopic redshift of this
source (zspec = 1.06) is inconsistent with our best estimate
of the redshift, which within a 68% confidence interval places
it at z = 2.3 ± 0.8 (le2). Redshifts in the range 1.0 ≤ z ≤ 1.5
have a 10% integrated probability.
SMMJ16403+46437: The detection at 1.4GHz
(593µJy) is high when compared with the 850µm level, and
there is no acceptable correspondence with any of the SEDs
in our template library. This, plausibly, might be a misiden-
tification or a radio-quiet AGN, with enhanced radio lumi-
nosity compared to the rest-frame FIR luminosity.
SMMJ16403+46440: Treating the 1.35mm observa-
tion as a detection (2.2σ) provides no additional constraint
and gives a similar redshift distribution to the one derived
from the 450µm/850µm colour and the 1.4GHz and 1.35mm
upper limits, shown in Figure A2.
3.3 Cumulative redshift distribution of the
sub-mm galaxy population
It is straightforward to calculate the cumulative redshift dis-
tribution for the sub-mm galaxy population as the coaddi-
tion of the individual probability distributions. It should be
noted that the calculation carried out in this paper has the
advantage of including the whole redshift probability dis-
tribution of each individual galaxy, and not just the mean
values of the distributions. We have included in this cal-
culation of the cumulative distribution only those sub-mm
sources identified in wide-area blank-field surveys (UK 8mJy
survey and the CUDSS) at a ≥ 3.5σ level: 50 sources. We
do not include sub-mm galaxies identified in surveys car-
ried out towards lensing clusters, or targeted observations
c(cid:13) 2002 RAS, MNRAS 000, 1 -- ??
Breaking the "Redshift Deadlock" -- II
13
Figure 9. Combined redshift distribution for the population of
sub-mm galaxies in the 8mJy SCUBA survey with one or more
measured colours (the 14 galaxies contained in Tables 1 & 2). The
different lines represent different evolutionary models assumed for
the sub-mm galaxy population (section 2). Upper panel: discrete
bin probabilities. Lower panel: cumulative distribution.
of other populations of galaxies (radio, ISO, Lyman-break
galaxies or extremely red objects), or catalogs that have not
been fully published (MAMBO surveys), since they can dis-
tort and bias the characteristic redshift distribution of the
sub-mm galaxy population.
Half of sources in complete flux-limited sub-mm sam-
ples, such as the UK 8mJy and CUDSS surveys, have a
single 850µm detection with one, or more, additional upper-
limits at other wavelengths: 28/50 sources detected (> 3.5σ)
at 850µm belong to this category. Since the redshift distribu-
tions of these sources are the most dependent on the priors
used, the interpretation of the combined redshift distribu-
tions should be based on the range of results given by the
different evolutionary models.
Fig. 9 shows the cumulative distribution and the com-
bined redshift distribution of the 14 galaxies in the UK 8mJy
survey with redshift distributions derived from SEDs which
contain at least 2-band detections (objects in Tables 1 and
2). This figure indicates that 65 -- 90 per cent of the sub-
mm galaxies could have redshifts in the range 2 <∼ z <∼ 4.
The results are consistent with the expectation that sub-
mm galaxies detected in the SCUBA (and also MAMBO)
surveys represent a high-z population (Dunlop 2001, Blain
et al. 2002).
c(cid:13) 2002 RAS, MNRAS 000, 1 -- ??
Figure 10. Combined redshift distribution for the population of
all 50 sub-mm galaxies detected in wide-area surveys at 850µm
(> 3.5σ) and listed in Tables 1,2, and 3. The different lines repre-
sent different evolutionary models assumed for the sub-mm galaxy
population (see text). Beware that the extended tail towards high-
redshifts is due to sources which are just detected at 850µm. The
upper limits at other wavelengths do not usually help to constrain
their redshifts between 2 ≤ z ≤ 10, and hence their probability
distributions are very flat in those regimes. Upper panel: discrete
bin probabilities. Lower panel: cumulative distribution.
The sources in the CUDSS sample with one or more
measured colours (9 sources) appear to have a significantly
different lower redshift distribution, with 40 per cent of the
galaxies at z <∼ 2 and approximately 50 per cent of the
galaxies at 2 <∼ z <∼ 4. The high redshift tail of the CUDSS
sample is similar to that shown by the 8mJy survey with ∼10
per cent probability at z > 4. The two galaxies responsible
for the low-redshift difference are CUDSS3.1 and 3.9, both
with relatively high 450µm flux densities S450µm ≥ 77 mJy,
compared with the 30 -- 40 mJy detection levels for the UK
8mJy survey sources. Taking the data for the CUDSS sample
at face-value suggests that it includes an over-density of low-
z galaxies at z <∼ 2, compared to the galaxies in the UK 8mJy
survey, and vice-versa. This discrepancy, however, could also
be explained by a difference in the flux-calibration of the
450µm data in the two surveys.
Fig. 10 shows the cumulative redshift distributions of
the sub-mm galaxy population for the 50 galaxies in the
8mJy and CUDSS surveys detected at a ≥ 3.5σ level. This
extended sample includes the 27 sub-mm galaxies with a
14
Itziar Aretxaga et al.
Figure 11. Distribution of intrinsic fluxes at 1.4GHz correspond-
ing to a population of galaxies discovered at 850µm, at a level
≥ 8mJy. Depending on the evolutionary model adopted, 80% of
the sources would have a flux in excess of 2 to 5µJy, and hence
in order to conduct a parallel 1.4GHz imaging follow-up program
with 80% completeness, the 1σ depth of the images should be 1
to 2µJy.
single detection at 850µm and one, or more, upper limits
at other wavelengths (Table 3). The combined redshift dis-
tributions, for the different evolutionary models under con-
sideration, imply that the majority of the sub-mm galax-
ies, approximately 50 -- 90 per cent, lie between z = 2 − 4.
The remainder of the galaxies, ∼ 10 per cent, lie at z < 2.
Fig. 10 also shows that <∼ 50 per cent of the galaxies have
colours that are consistent with z > 4. The distribution of
the high-redshift tail is very sensitive to the choice of evo-
lutionary model, since the individual redshift distributions
for the 27 sources with only a single detection are neces-
sarily very flat over the redshift range z >∼ 2 (see Fig. A3).
For these galaxies, the constraint on their redshift distribu-
tions comes mainly from the flux detected at 850µm and the
ability to reproduce it with the evolution of the 60µm local
luminosity function. More extensive photometry, especially
including shorter sub-mm wavelengths, is crucial to better
constrain these flat distributions, as illustrated by Fig. 7 (see
also Fig.6 in paper I).
4 THE LIKELIHOOD OF MIS-IDENTIFYING
THE RADIO COUNTERPARTS OF HIGH-Z
SUB-MM GALAXIES
The strength of the correlation between the FIR and radio
luminosities in starburst galaxies (Helou et al. 1985, Con-
don 1992) has supported the expectation that all high-z
star forming sub-mm galaxies should also have a detectable
radio-counterpart, provided the radio observations are deep
enough.
Although originally encouraged by this assumption,
some of the deepest VLA searches for radio emission as-
sociated with SCUBA and MAMBO sources have met with
varying degrees of success.
The deep VLA surveys of the Lockman Hole and ELAIS
N2 fields (Ivison et al. 2002) have a 1σ noise limit ≥ 5−9µJy
at 1.4 GHz. These surveys have identified 11/16 and 8/14
sub-mm sources at 1.4GHz, respectively, with peak radio
fluxes above 4σ and integrated fluxes above 3σ, after reject-
ing 5 and 1 sub-mm sources from the parent samples, re-
spectively, because of their high sub-mm noise levels. From
these, there are 9/16 and 5/14 sources in the Lockman Hole
and ELAIS N2 surveys, respectively, with a final S/N>3
when one considers the additional errors due to calibration
and fitting a gaussian to the peak of the radio emission.
If one limits the analysis to sources ≥ 8 mJy at 850µm,
then 4/10 and 4/8 sources have been simultaneously de-
tected with S/N> 3 at 850µm and 1.4GHz. This is in clear
contrast to the claims of previous authors (Fox et al. 2002)
that bright sub-mm sources (> 8mJy at 850µm) should be
detected at flux densities of ∼ 100µJy at 1.4 GHz.
In order to understand the properties of the radio asso-
ciations, we have calculated the probability of detection at
1.4GHz for a mock 8mJy survey, following the same method-
ology as in paper I, but using the extended library of SEDs
(section 2) . The intrinsic 850µm fluxes of the galaxies are
convolved with 1σ ∼ 2.6mJy measurement errors, and 9 per
cent calibration uncertainty. Fig.11 shows the distribution of
intrinsic 1.4GHz fluxes for the detected sources (including
errors) above the 8mJy threshold at 850µm.
We conclude that future surveys at 1.4 GHz must reach
a 3σ sensitivity of 2 -- 5µJy if radio observations are to pro-
vide counterparts to 80% of the high-z sub-mm galaxies with
850µm flux densities ≥ 8 mJy, according to the redshift dis-
tributions assumed in the alternative evolutionary models.
Thus, current radio observations have insufficient sensitivity
to detect the bulk of the sub-mm population with adequate
S/N, and we must wait for the next generation of new and
upgraded facilities (eg. e-VLA, e-MERLIN, SKA).
In the case of the UK 8mJy survey, Fig. 10 shows that
40 -- 50% (Lockman Hole) and 50 -- 60% (ELAIS N2) of the
sources should have an intrinsic 1.4GHz flux lower than the
3σ threshold, in the absence of any radio-loud AGN emis-
sion. These levels agree well with the number of radio coun-
terparts to ≥ 8mJy sources (Scott et al. 2002, Ivison et al.
2002), bearing in mind the actual variation in sensitivity
across the SCUBA wide-area survey fields.
This potential
for misidentifying the correct radio
sources associated with blank-field sub-mm galaxies will in-
evitably lead to incorrect estimates of photometric redshifts
if they are derived in part from the 1.4 GHz fluxes. In the cal-
culation of the cumulative redshift distribution (section 3.3)
we have treated all of the published associations of 1.4 GHz
and 850µm sources as the correct radio counterparts to the
sub-mm galaxies.
Despite the rapid acceptance of the radio -- sub-mm flux
density ratio, α1.4
345 (CY99, CY00), as the preferred method
with which to estimate the redshift distribution of the sub-
mm galaxy population, this diagnostic ratio loses its power
to discriminate between photometric redshifts at z > 2
(CY99, CY00). We have already demonstrated in paper I
(Fig. 11) the detrimental effect of the large dispersion in
radio -- sub-mm colours of the local template galaxies on the
accuracy of the derived photometric redshifts of the sub-mm
blank field sources. Furthermore, the existing radio interfer-
ometers will remain limited in their ability to map the large
areas ∼ 1 -- 40 sq. deg. that will be surveyed by the next
c(cid:13) 2002 RAS, MNRAS 000, 1 -- ??
Breaking the "Redshift Deadlock" -- II
15
(Downes et al. 1999) and the unprecedented depth of
the multi-wavelength HDF surveys (Williams et al. 1996;
Thompson et al. 1999; Richards 1999; Hornschmeier et al.
2000). It is only after the accurate subtraction of the ellip-
tical galaxy, 3−586, at z ∼ 1, however, that a faint lensed
IR counterpart (K = 23.5, H − K = 1.5, I − K > 5.1) to
HDF850.1 has been discovered (Dunlop et al. 2002).
This IR detection now gives greater confidence to the
likelihood that the coincident radio emission, with a flux
density of 16±4µJy at 1.4 GHz, from the combined data of
MERLIN and the VLA, is genuinely associated with the
sub-mmm source (Dunlop et al. 2002). Furthermore this
strengthens the suggestion that a marginal 3.3σ detection
at 8.4 GHz (VLA 3651+1226, Richards, priv. comm.) that
may also associated with HDF850.1.
The redshift probability distribution for HDF850.1,
based on the multi-wavelength 170µm -- 1.4 GHz data, indi-
cates z is comprised between 3.3 and 5.4 with a 86% to 90%
confidence depending on the adopted prior (Fig.12). This
high photometric redshift supports the earlier decision to
reject the z ∼ 1 elliptical galaxy 3−586, as the optical coun-
terpart. Unfortunately the IR counterpart, HDF850.1K, is
too faint to attempt IR spectroscopy with even the 10-m
class of telescopes.
The most efficient means to determine the spectroscopic
redshifts of optically-obscured sub-mm galaxies is the de-
tection of millimetre and centimetre wavelength molecular
lines using large single-dish telescopes (e.g. GBT or LMT)
with broad-band receivers (Townsend et al. 2001, Carilli &
Blain 2002). In this way, it is possible to eliminate the neces-
sity for an optical/IR counterpart, or a position accurate to
< 3 arcsec. Given the predicted accuracy of the photomet-
ric redshift for HDF850.1, and taking all of the uncertainties
discussed in section 2 into account, we demonstrate in Fig.12
that the J=1 -- 0 CO line in HDF850.1 has a 86-90% probabil-
ity of detection in the K-band receiver (18.0 -- 26.5 GHz) on
the GBT. Observations to conduct searches for CO emission
in high-z sub-mm galaxies, guided only by their photometric
redshifts, are currently underway on the GBT.
6 CONCLUSIONS
We have derived meaningful photometric-redshift probabil-
ity distributions for individual sub-mm galaxies using Monte
Carlo simulations that take into account a range of measure-
ment errors and model-dependent uncertainties. From the
individual redshift probability distributions, we have mea-
sured the cumulative redshift distribution for the sub-mm
population of galaxies identified in blank field surveys. We
find that the multi-wavelength FIR -- radio data for 50 sub-
mm galaxies detected (> 3.5σ) at 850µm are consistent with
∼ 50−90% of the population lying at redshifts z = 2−4, with
only a small fraction, ∼ 10% of the population at z < 2. We
also show that up to 50% of the galaxies have colours that
are consistent with z > 4. The possibility that the sub-mm
galaxy population contains a significant fraction undergoing
a major burst of star formation at such early epochs has im-
portant consequences for models of galaxy formation. This
suggestion is still highly dependent on the prior assumption
of the evolutionary model for the sub-mm population which
Figure 12. Redshift probability distributions for HDF850.1. The
different line types and colours correspond to the different evo-
lutionary models considered. Model led1 is not plotted, since it
excludes sources at z > 3.5 due to the strong prior decline of
the density and luminosity of sources. The unshaded area corre-
sponds to the redshift range where the 12CO (J = 1 − 0) line lies
in the observable K-band window (18.5 -- 26.0 GHz), accessible
with the 100m GBT. There is an 86-90% probability of detecting
the J=1-0 CO-line in this GBT window
generation of sub-mm and FIR experiments (e.g. BLAST,
Herschel, SIRTF) down to the necessary sensitivity levels
(3σ ∼ 2 − 5µJy at 1.4GHz), making the use of α345
1.4 an im-
practical measure of redshift for the majority of the popula-
tion. Deep interferometric radio observations, however, re-
main a practical way to pinpoint the source of the sub-mm
emission with sufficient positional accuracy to investigate
galaxy morphologies and exploit optical/IR spectroscopic
line diagnostics.
This
the
limitation of
radio -- sub-mm technique
prompted the design and construction of a Balloon-borne
Large Aperture Submillimetre Telescope, BLAST (Devlin et
al. 2001). The primary scientific goal of BLAST is to break
the current "redshift deadlock" by providing photometric
redshifts with sufficient accuracy, without the prior neces-
sity for secure optical, IR or radio counterparts. BLAST will
use rest-frame sub-mm and FIR colours to derive redshifts
with an accuracy ∆z ∼ ±0.5 for > 1000 galaxies in the red-
shift range 0 < z < 6 detected in wide-area (> 0.5 −10 deg2)
surveys at 250, 350 and 500µm (Hughes et al. 2002).
This redshift accuracy is sufficient to guide the choice of
possible front-end spectrographs, and their receiver tunings,
on large single-dish ground-based mm and cm telescopes
(e.g. 100-m GBT -- Green Bank Telescope and 50-m LMT --
Large Millimetre Telescope) in the effort to detect redshifted
CO emission-lines. We conclude this paper with the descrip-
tion of one application of this technique.
5 BROAD-BAND MM -- CM SPECTROSCOPY
Until recently the brightest sub-mm source in the Hubble
Deep Field, HDF850.1 (Hughes et al. 1998), has avoided
the detection of its X-ray, optical, IR or radio counterpart,
despite an accurate interferometric position at 236 GHz
c(cid:13) 2002 RAS, MNRAS 000, 1 -- ??
16
Itziar Aretxaga et al.
affects most strongly those sub-mm sources detected only at
850µm.
Shorter-wavelength sub-mm data (250 − 500µm) from a
future balloon-borne experiment, BLAST, will provide pow-
erful additional constraints (∆z ∼ ±0.5) on the redshift dis-
tribution of all the SCUBA and MAMBO galaxies at z > 2.
A photometric-redshift accuracy of ±0.5 for an individual
galaxy will be sufficient to provide robust statistical mea-
surements of the redshift distribution of the entire popula-
tion, the global history of dust-obscured star formation and
the clustering properties of sub-mm galaxies.
We have argued that the large dispersion in the radio
to sub-mm SEDs (or 850µm/1.4GHz colour) of the template
galaxies naturally translates into an imprecise measure of
the photometric redshift, and that radio observations can-
not be used efficiently to measure the individual redshifts
of thousands of galaxies detected in future wide-area sub-
mm surveys. The sub-mm/radio spectral index is still able
to discriminate between the z < 2 and z > 2 regime.
We have an on-going programme to conduct centimetre-
wavelength observations of CO J=1-0 molecular line tran-
sitions in high-z sub-mm galaxies on the 100-m GBT. The
observations, however, first require an accurate photometric-
redshift probability distribution to guide the choice of re-
ceiver and tuning parameters. We illustrate the power of
our photometric-redshift technique with the example of
HDF850.1, one of the best-studied sub-mm galaxies (Hughes
et al. 1998; Downes et al. 1999; Dunlop et al. 2002).
Within the next few years, the combination of ground-
based, balloon-borne and satellite experiments will provide
accurately calibrated rest-frame sub-mm -- FIR data for a sig-
nificant population of sub-mm galaxies. The centimetre and
millimetre wavelength spectroscopic redshift confirmation of
their photometric redshifts will calibrate the method out-
lined in this paper. Thus we can finally look forward to
breaking the "redshift deadlock" that currently prevents an
accurate understanding of the nature of the sub-mm galaxy
population and their evolutionary history.
7 ACKNOWLEDGMENTS
This work has been partly supported by CONACYT grants
32180-E and 32143-E. We thank the anonymous referee for
the critical comments and careful reading of this manuscript.
REFERENCES
Barger A.J., Cowie, L.L, Richards, E.A., 2000, A.J., 119, 2092
Bertoldi F. et al., 2000, A&A, 360, 92.
Blain A.W. et al., 2002, Physics Reports,
in press (astro-
ph/020228)
Borys, C, Chapman S.C, Halpern M., Scott D., 2002, MNRAS,
330, 63.
Carilli C.L. & Blain , 2002, ApJ, 569 605.
Carilli C.L. & Yun M.S., 1999, Ap.J., 513, L13 (CY99)
Carilli C.L. & Yun M.S., 2000, Ap.J., 530, 618 (CY00)
Chapin E. et al, 2003, MNRAS, in preparation.
Chapman S.C. et al. 2000, MNRAS, 319, 318.
Chapman S.C., Scott D., Borys C., & Fahlman G.G., 2002, MN-
RAS, 330, 92.
Chapman S.C., Smail I., Ivison R.J., Helou G., Dale D.A., La-
gache G., 2002b, ApJ, in press (astro-ph/0203068)
Condon J.J., 1992, ARAA, 30, 575
Dannerbauer H. et al., 2002, ApJ, in press (astro-ph/0201104)
Devlin M. et al., 2001, in J. Lowenthal & D.H. Hughes eds., Deep
Millimetre Surveys: Implications for Galaxy Formation and
Evolution World Scientific, p. 59 (astro-ph/0012327)
Dey A., Graham J.R., Ivison R.J., Smail I., Wright G.S. & Liu
M.C., 1999, ApJ, 519, 610.
Downes D., et al., 1999, A&A, 347, 809
Dunlop J.S., 2001, in FIRSED2000, eds. I.M. van Bremmel et al.,
Elsevier New Astronomy Reviews (astro-ph/0101297).
Dunlop J.S. et al., MNRAS, 2002, submitted
Dunne L., Eales S.A., Edmunds M., Ivison R., Alexander P., &
Clements D.L., 2000, MNRAS, 315, 115
Cowie L., Barger A.J. & Kneib J., 2002, AJ, 123, 2197.
Eales S.A., et al., 1999, Ap.J., 515, 518
Eales S.A., et al., 2000, A.J., 120, 2244
Fardal M.A., Katz N., Weinberg D.H., Dav´e R., Hernquist L.,
2002, ApJ, submitted (astro-ph/0107290)
Fox M.J., et al., 2002, MNRAS, 331, 839.
Frayer D.T., et al., 2000, A.J., 120, 1668
Gear W.K., Lilly S. J., Stevens J.A., Clements D.L., Webb T.M.,
Eales S.A., Dunne, L., 2000, MNRAS, 316, 51.
Helou P. et al. 1985, ApJ, 440, 35
Hughes D.H., Robson E.I., Dunlop J.S., Gear W.K., 1993, MN-
RAS, 263, 607
Hughes D.H., et al., 1998, Nature, 394, 241
Hughes D.H., et al., 2002, MNRAS, submitted, (paper I, astro-
ph/0111547)
Hornschmeier A., et al., 2000, Ap.J., 541, 49
Ivison R.I. et al., 1998, MNRAS, 298, 583
Ivison R.J., et al., 2000, MNRAS, 315, 209
Ivison R.J., et al., 2002, MNRAS, in press (astro-ph/0206432)
Klaas U. et al. 2001. A&A, 379, 823
Lilly S.J., et al., 1999, Ap.J., 518, 441
Lutz D., et al., 2001, A&A, 378, 70.
Richards E.A., et al., 1998, Ap.J., 116, 1039
Richards E.A., 1999, Ap.J., 513, L9
Richards E.A., 2000, Ap.J., 533, 611
Saunders D.B. Rowan-Robinson M., Lawrence A., Efstathiou G.,
Kaiser N., Ellis R. S., Frenk C. S., 1990, MNRAS, 242, 318
Scott D. et al., 2000, A&A, 357, 5.
Scott S., et al., 2002, MNRAS, 331, 817.
Smail I., et al., 1997, MNRAS, 490, L5
Smail I., et al., 1999, MNRAS, 308, 1061
Smail I., Ivison R.J., Blain A.W., Kneib J.-P., 2002, MNRAS,
331, 495.
Thompson R., et al., 1999, A.J., 117, 17
Townsend et al. 2001, MNRAS, 328, L17
Webb T.M.A., et al., 2002, ApJ, submitted (astro-ph/0201181).
Williams R.E., et al., 1996, A.J., 112, 1335
Yun M.S. & Carilli C.L., 2002, Ap.J., in press (YC02, astro-
ph/0112074).
APPENDIX A: REDSHIFT ESTIMATES FOR
INDIVIDUAL SUB-MM GALAXIES
A catalogue of redshift probability distributions and SEDs
for the galaxies detected in more than 2 bands (Tables 1 and
2) is included. We also include the corresponding colour-
colour-redshift or colour-flux-redshift plots from which they
have been derived. All estimates are derived for model le2.
For galaxies detected in just one band (Table 3) we
present a selection of the results derived for 6 out of the
46 galaxies included in Table 3. The general uniformity of
their redshift distributions is a fair representation of those
galaxies not included in this appendix.
c(cid:13) 2002 RAS, MNRAS 000, 1 -- ??
|
0710.3826 | 1 | 0710 | 2007-10-20T09:03:32 | R fluids | [
"astro-ph"
] | A theory of collisionless fluids is developed in a unified picture, where nonrotating figures with anisotropic random velocity component distributions and rotating figures with isotropic random velocity component distributions, make adjoints configurations to the same system. R fluids are defined and mean and rms angular velocities and mean and rms tangential velocity components are expressed, by weighting on the moment of inertia and the mass, respectively. The definition of figure rotation is extended to R fluids. The generalized tensor virial equations are formulated for R fluids and further attention is devoted to axisymmetric configurations where, for selected coordinate axes, a variation in figure rotation has to be counterbalanced by a variation in anisotropy excess and vice versa. A microscopical analysis of systematic and random motions is performed under a few general hypotheses, by reversing the sign of tangential or axial velocity components of an assigned fraction of particles, leaving the distribution function and other parameters unchanged (Meza 2002). The application of the reversion process to tangential velocity components, implies the conversion of random motion rotation kinetic energy into systematic motion rotation kinetic energy. The application of the reversion process to axial velocity components, implies the conversion of random motion translation kinetic energy into systematic motion translation kinetic energy, and the loss related to a change of reference frame is expressed in terms of systematic (imaginary) motion rotation kinetic energy. A procedure is sketched for deriving the spin parameter distribution (including imaginary rotation) from a sample of observed or simulated large-scale collisionless fluids i.e. galaxies and galaxy clusters. | astro-ph | astro-ph |
R fluids
R. Caimmi∗
December 19, 2018
Abstract
A theory of collisionless fluids is developed in a unified picture,
(σ11 = σ22 = σ33) random velocity component distributions and ro-
where nonrotating (fΩ1 = fΩ2 = fΩ3 = 0) figures with anisotropic
tating (fΩ1 6= fΩ2 6= fΩ3) figures with isotropic (σ11 6= σ22 6= σ33) ran-
dom velocity component distributions, make adjoints configurations
to the same system. R fluids are defined as ideal, self-gravitating
fluids satisfying the virial theorem assumptions, in presence of sys-
tematic rotation around each principal axis of inertia. To this aim,
mean and rms angular velocities and mean and rms tangential veloc-
ity components are expressed, by weighting on the moment of inertia
and the mass, respectively. The figure rotation is defined as the mean
angular velocity, weighted on the moment of inertia, with respect to
a selected axis. The generalized tensor virial equations (Caimmi &
Marmo 2005) are formulated for R fluids and further attention is de-
voted to axisymmetric configurations where, for selected coordinate
axes, a variation in figure rotation has to be counterbalanced by a
variation in anisotropy excess and vice versa. A microscopical analy-
sis of systematic and random motions is performed under a few general
∗Dipartimento di Astronomia, Universit`a di Padova, Vicolo Osservatorio 2, I-35122
Padova, Italy - email: [email protected]
1
hypotheses, by reversing the sign of tangential or axial velocity com-
ponents of an assigned fraction of particles, leaving the distribution
function and other parameters unchanged (Meza 2002). The appli-
cation of the reversion process to tangential velocity components, is
found to imply the conversion of random motion rotation kinetic en-
ergy into systematic motion rotation kinetic energy. The application
of the reversion process to axial velocity components, is found to im-
ply the conversion of random motion translation kinetic energy into
systematic motion translation kinetic energy, and the loss related to
a change of reference frame is expressed in terms of systematic (imag-
inary) motion rotation kinetic energy. A number of special situations
are investigated with further detail. It is found that a R fluid always
admits an adjoint configuration where figure rotation occurs around
only one principal axis of inertia (R3 fluid), which implies that all
the results related to R3 fluids (Caimmi 2007) may be extended to
R fluids. Finally, a procedure is sketched for deriving the spin pa-
rameter distribution (including imaginary rotation) from a sample of
observed or simulated large-scale collisionless fluids i.e. galaxies and
galaxy clusters.
keywords - galaxies: clusters - galaxies: haloes - stars: stellar systems.
1
Introduction
11 = σ2
33), where σ2
Due to particle shocks, collisional fluids (e.g., stars, gas clouds) exhibit an
isotropic stress tensor (σ2
pp is the rms random ve-
locity component on the axis, xp. The absence of particle shocks (leaving
aside extreme situations, such as high-density galactic nuclei) makes a dif-
ferent situation in collisionless fluids (e.g., galaxies, galaxy clusters), where
the stress tensor is - in general - anisotropic (σ2
33). The shape
of the body is determined by systematic rotation, which is quantified by a
spin parameter (null for nonrotating configurations), and/or by the difference
between stress tensor diagonal components, or any equivalent anisotropy in-
dicator, related to the rotation and an equatorial principal axis of inertia,
respectively (null for configurations where the random velocity component
distribution is isotropic). A description of collisionless fluids based on the
equivalence of systematic and random motions with respect to the shape,
appears to be highly rewarding and it would provide further insight on the
11 6= σ2
22 6= σ2
22 = σ2
2
properties of stellar and galaxy systems.
In an earlier attempt (Caimmi 1996) the stress tensor has been expressed
as the sum of two terms, one related to a random (isotropic) velocity com-
ponent distribution, and one other to anisotropic internal motions within
the system. Further investigation has been devoted to the simplest situation
where the system is made of two equal components, which are rotating at
the same rate but in opposite sense. Then it has been recognized that the
anisotropy excess may be related to real rotation, if the shape is flattened,
and to imaginary rotation, if the shape is elongated, with respect to the
rotation axis.
A latter approach (Caimmi & Marmo 2005) has been restricted to homeoidally
striated density profiles, for which the tensor virial equations were formulated
and generalized to unrelaxed configurations. The kinetic-energy tensor has
been expressed as the sum of two terms, one related to systematic rotation
obeying an assigned law, and one other to the remaining motions e.g., ran-
dom motions, streaming motions, radial motions. Finally, an expression of
the spin parameter in terms of the anisotropy excess, has shown the role of
systematic and remaining motions in flattening or elongating the shape.
The above mentioned results have been improved and extended in sub-
sequent work (Caimmi 2006, hereafter quoted as C06), where imaginary ro-
tation has been related to negative anisotropy excess. Then sequences of
configurations for which the generalized tensor virial equations hold, have
been determined for homeoidally striated Jacobi ellipsoids including prolate
shapes induced by imaginary rotation. The results from numerical simula-
tions on the stability of rapidly rotating spherical configurations (Meza 2002)
have been interpreted in the light of the theory. To this respect, the key argu-
ment is that the reversion (from clockwise to counterclockwise or vice versa)
of tangential velocity components related to an assigned fraction of partices,
preserves the potential energy, the kinetic energy, and the distribution func-
tion (Lynden-Bell 1960, 1962; Meza 2002).
The study on homeoidally striated Jacobi ellipsoids has been extended to
a more general class of bodies (R3 fluids) in a recent paper (Caimmi 2007,
hereafter quoted as C07), where the contribution of radial and tangential
velocity components on the equatorial plane was investigated with further
detail. In addition, mean and rms (weighted on the moment of inertia) angu-
lar velocity have been defined, and related to systematic and random motion
tangential kinetic-energy tensor components, respectively. Also for R3 fluids,
3
it has been realized that the effect of (positive or negative) anisotropy excess
is equivalent to additional (real or imaginary) figure rotation.
The current attempt is aimed to extend the above mentioned results to a
still more general class of bodies, R fluids, defined as ideal, self-gravitating,
collisionless fluids where rotation occurs around each principal axis of iner-
tia.
It will be found that R fluids always admit an adjoint configuration
where figure rotation occurs around a single principal axis, that is a R3 fluid.
Accordingly, all the results which hold for R3 fluids may be extended to R
fluids.
The work is organized as follows. A number of basic definitions are pro-
vided in Sect. 2, including the inertia tensor, the angular-velocity tensor, and
the angular-momentum tensor. The generalized tensor virial equations for
R fluids are formulated in Sect. 3. The microscopical analysis of systematic
and random motions , for a collisionless fluid made of N identical parti-
cles, is performed in Sect. 4, where a velocity component reversion process
is defined, and a number of special situations are analysed in detail with
respect to kinetic energy changes from random to systematic motions and
vice versa. A procedure aimed to the derivation of the spin parameter distri-
bution (including imaginary rotation) from an assigned sample of observed
or simulated objects, is outlined in Sect. 5. Some concluding remarks are
reported in Sect. 6.
2 Angular-velocity and angular-momentum ten-
sor
In the special case of solid bodies, rotation is rigid and occurs around a single
axis which, in turn, can remain fixed or change its direction. Accordingly,
the angular momentum and the rotation kinetic energy read (e.g., Landau &
Lifchits 1966, Chap. VI, §§ 31-33; hereafter quoted as LL66):
Jr =
Trot =
3Xs=1
1
2
I ′
rsΩs
;
r = 1, 2, 3 ;
3Xr=1
3Xs=1
I ′
rsΩrΩs
;
4
(1)
(2)
where ~J = (J1, J2, J3) is the angular-momentum vector,
the angular-velocity vector, and I ′ the inertia tensor:
−→
Ω = (Ω1, Ω2, Ω3)
I ′
rs =ZS
ρ(x1, x2, x3)"δrs
3Xr=1
r − xrxs# d3S ;
x2
(3)
related to the density profile, ρ, within the volume, S, being δrs the Kronecker
33, are the
symbol. The diagonal components of the inertia tensor, I ′
moments of inertia with respect to the axes, x1, x2, x3, respectively.
22, I ′
11, I ′
In addition, the inertia tensor is symmetric of second rank, which implies
the existence of a reference frame, (O′ X1 X2 X3), where the inertia tensor is
diagonal (LL66, Chap. VI, § 32):
I ′
rs = δrsI ′
rr
;
(4)
the coordinate axes coincide with the principal axes of inertia, and the di-
agonal components define the principal moments of inertia. Accordingly,
Eqs. (1) and (2) reduce to:
3Xs=1
2(cid:16)I ′
1
Jr =
I ′
rrΩr
;
r = 1, 2, 3 ;
Trot =
11Ω2
1 + I ′
22Ω2
2 + I ′
33Ω2
3(cid:17) ;
where, in addition (LL66, Chap. VI, § 32):
rr ≤ I ′
I ′
ss + I ′
tt
;
r = 1, 2, 3 ;
s = 2, 3, 1 ;
t = 3, 1, 2 ;
(7)
with regard to the body under consideration.
The inertia tensor has been defined in a different way, as (e.g., Chan-
drasekhar 1969, Chap. 2, § 9; Binney & Tremaine 1987, Chap. 4, § 3):
Irs =ZS
ρ(x1, x2, x3)xrxs d3S ;
and the combination of Eqs. (3) and (8) yields:
I ′
rs = δrs
Irr − Irs
;
3Xr=1
I ′
rr = Iss + Itt
I ′
rs = −Irs
;
;
r 6= s 6= t ;
r 6= s ;
5
(5)
(6)
(8)
(9)
(10)
(11)
or:
ss + I ′
tt − I ′
rr
;
r 6= s 6= t ;
2Irr = I ′
Irs = −I ′
rs
;
r 6= s ;
(12)
(13)
which translates one formulation into the other (e.g., Bett et al. 2007).
In the general case of (collisional or collisionless) fluids, rotation could
occur different from solid-body, and around each principal axis of inertia.
Let (O x1 x2 x3) be a generic reference frame and (O′ X1 X2 X3) a reference
frame where the origin coincides with the centre of inertia, and the coordinate
axes coincide with the principal axes of inertia. Let the coordinate axes, X1,
−→
X2, X3, be defined as the principal axes. Let
Ω3, be the angular-
velocity vectors (to be specified later) related to the principal axes of inertia.
−→
Ωr, on the coordinate axis, xs. The
Let Ωrs be the component of the vector,
(3 × 3) tensor, Ωrs, is defined as the angular-velocity tensor of the system
under consideration, with respect to the reference frame, (O x1 x2 x3). Let
−→ω1, −→ω2, −→ω3, be the angular-velocity vectors related to the coordinate axes, x1,
x2, x3. The following relation holds:
−→
Ω1,
−→
Ω2,
ωs =
3Xr=1
Ωrs
;
s = 1, 2, 3 ;
(14)
and the angular-velocity tensor, ωrs = Ωrs, can formally be defined. Simi-
larly, the (3 × 3) angular-momentum tensor is expressed as:
J ′
rs = I ′
rsωrs
;
J ′
s =
3Xr=1
I ′
rsωrs
;
(15)
(16)
where the inertia tensor is related to the reference frame, (O x1 x2 x3).
In the special case where the reference frame, (O x1 x2 x3), coincides with
(O′ X1 X2 X3), then Ωrs = δrsΩrr. Accordingly, Eqs. (14)-(16) reduce to:
ωs = Ωs = Ωss = ωss
J ′
rs = δrsI ′
J ′
s = Js = I ′
rrΩrr
ssΩs
;
;
;
(17)
(18)
(19)
where I ′
tt = Irr + Iss, r 6= s 6= t, represents the moment of inertia with respect
to the principal axis of inertia, xt. From this point on, it shall be intended
6
that the origin coincides with the centre of inertia, and the coordinate axes
coincide with the principal axes of inertia.
The rotation kinetic-energy tensor is defined as:
(Trot)rs =
1
2
I ′
rsΩrΩs =
1
2
δrsI ′
rrΩ2
r
;
(20)
where the diagonal components of the angular-velocity tensor are expressed
as (C07):
1
I ′
rr ZS(cid:12)(cid:12)(cid:12)
Ωr = fΩr =
Ωr(x1, x2, x3, t)(cid:12)(cid:12)(cid:12) =
(cid:12)(cid:12)(cid:12)
wr = (x2
s + x2
t )1/2 ;
−→
−→
Ωr(x1, x2, x3, t)(cid:12)(cid:12)(cid:12) w2
vφr(x1, x2, x3, t)
;
wr
r ρ(x1, x2, x3, t) d3S; r 6= s 6= t;
(21)
(22)
(23)
and Ωr(x1, x2, x3, t) is the mean value related to all the particles at the time, t,
within the infinitesimal volume element, d3S = dx1 dx2 dx3, centred on the
point, P(x1, x2, x3), vφr is the tangential velocity component on the (O xs xt)
principal plane, and the moment of inertia, I ′
rr = Iss + Itt, reads:
I ′
rr =ZS
w2
r ρ(x1, x2, x3, t) d3S ;
r 6= s 6= t ;
(24)
as expected from the theorem of the mean, in connection with Eq. (21).
Similarly, the mean square diagonal components of the angular-velocity
tensor are expressed as:
w2
r ρ(x1, x2, x3, t) d3S; r 6= s 6= t; (25)
(26)
r) =
Ω2
r = g(Ω2
1
I ′
rr ZSh−→
and the related variance reads:
Ωr(x1, x2, x3, t)i2
(cid:16)σfΩrfΩr(cid:17)2
= g(Ω2
r) − (fΩr)2 ;
as known from statistics.
At this stage, it may be useful to extend and generalize the definition of
figure rotation.
Figure rotation. Given a R fluid, the figure rotation is defined as the mean
angular velocity, weighted on the moment of inertia, with respect to a selected
principal axis.
7
In terms of tangential velocity components, the counterparts of Eqs. (21),
(25), and (26) read:
1
1
−→
φr ) =
M ZS(cid:12)(cid:12)(cid:12)
Ωr(x1, x2, x3, t)(cid:12)(cid:12)(cid:12) wrρ(x1, x2, x3, t) d3S ;
gvφr =
M ZSh−→
Ωr(x1, x2, x3, t)i2
g(v2
(cid:16)σfvφrfvφr(cid:17)2
= g(v2
φr ) − (gvφr )2 ;
w2
r ρ(x1, x2, x3, t) d3S; r 6= s 6= t;
r 6= s 6= t ;
(27)
(28)
(29)
and the combination of Eqs. (25) and (28); (21) and (27); (26) and (29);
yields:
r) ;
φr ) = I ′
M g(v2
rr g(Ω2
rr(fΩr)2 ;
M(gvφr )2 = I ′
M(cid:16)σeφreφr(cid:17)2
rr(cid:16)σfΩrfΩr(cid:17)2
= I ′
(30a)
(30b)
(30c)
;
which relate tangential velocity components on the (O xs xt) principal plane,
to angular velocity components on the xr principal axis.
3 The generalized tensor virial equations for
R fluids
Let R fluids be defined as (collisional or collisionless) ideal self-gravitating
fluids where figure rotation occurs around all the three principal axes of
inertia. Let (O x1 x2 x3) be a reference frame where the origin coincides with
the centre of inertia, and the coordinate axes coincide with the principal axes
of inertia. Then the mean radial velocity components must necessarily equal
zero:
vwr = 0 ;
(v2
wr ) = (σwrwr )2
;
(31)
where vwr is the radial velocity component on the (O xs xt) principal plane,
perpendicular to the xr principal axis. Let positive and negative radial ve-
locity components be defined as directed outwards and inwards, respectively.
The same holds for the mean tangential velocity components:
vφr = 0 ;
(v2
φr ) = (σφrφr )2
;
(32)
8
even in presence of systematic rotation. Let positive and negative tangential
velocity components be defined as rotating counterclockwise and clockwise,
respectively.
The kinetic-energy tensor may be expressed as the sum of two contribu-
tions: one, related to systematic motions, and one other, related to random
motions (C07). The result is:
Tkskt = (Tsys)kskt + (Trdm)kskt
;
k = w, φ ;
(33)
where the terms on the right-hand side, using Eqs. (30) and (31), can be
expressed as:
(Tsys)wswt = 0 ;
1
2
(Trdm)wswt =
δstM (σwsws)2
(Tsys)φsφt =
(Trdm)φsφt =
1
2
1
2
δstI ′
δstI ′
ss(fΩs)2 ;
ss(cid:16)σfΩsfΩs(cid:17)2
;
;
(34a)
(34b)
(35a)
(35b)
keeping in mind that nondiagonal components are null in the case under
discussion, only diagonal components shall be considered from this point on.
The combination of Eqs. (26) and (35) yields:
Tφrφr =
1
2
r) ;
I ′
rr g(Ω2
(36)
which depends on the density profile via the moment of inertia, I ′
rr, and
the tangential velocity component distribution via the mean square angular
r), regardless from the fraction of systematic and random motions.
In terms of the contributions related to the axial components of the
velocity, g(Ω2
kinetic-energy tensor, Tss and Ttt, Eqs. (33), (35), and (36) read:
(Tφrφr )ℓℓ =
1
2
[(Tsys)φrφr ]ℓℓ =
[(Trdm)φrφr ]ℓℓ =
2
r) ;
Iℓℓ g(Ω2
Iℓℓ g(Ωr)
Iℓℓh g(Ω2
1
2
1
2
r) − (fΩr)2i
9
ℓ = s, t ;
;
ℓ = s, t ;
(37a)
(37b)
;
ℓ = s, t ;
(37c)
where Eq. (10) has been used.
The invariance of a vector with respect to a change of the reference frame,
implies the validity of the relations (C07):
s ) + (v2
wr) + (v2
φr) = (v2
(v2
(vwr)2 + (vφr )2 = (vs)2 + (vt)2 ;
(σwrwr)2 + (σφr φr)2 = (σss)2 + (σtt)2
t ) ;
(38)
(39)
(40)
;
where the velocity components on the xs and xt principal axes are labelled
by the indices, s and t, respectively.
The combination of Eqs. (26), (29), and (38)-(40) yields:
(σwrwr )2 = (σss)2 + (σtt)2 −
which makes Eqs. (33) and (34) translate into:
(Twrwr )ℓℓ = [(Trdm)wrwr]ℓℓ =
1
2
Mσ2
ℓℓ −
1
2
[(Tsys)wrwr ]ℓℓ = 0 ;
ℓ = s, t ;
;
I ′
rr
(41)
M h g(Ω2
Iℓℓh g(Ω2
r) − (fΩr)2i
r) − (fΩr)2i ; ℓ = s, t ; (42a)
(42b)
in terms of the contributions related to the axial components of the kinetic-
energy tensor, Tss and Ttt.
The generalized tensor virial equations of the second order can be formu-
lated, extending the procedure used for R3 fluids (C07). The result is:
Irrh(fΩs)2 + (fΩt)2i + Mζrrσ2 + (Epot)rr = 0 ;
11 + σ2
22 + σ2
σ2 = σ2
33 ;
ζpp =
(eTrdm)pp
Trdm
=
σ2
pp
σ2
;
p = 1, 2, 3 ;
ζ11 + ζ22 + ζ33 = eTrdm
Trdm
= eσ2
σ2 = ζ ;
(43)
(44)
(45a)
(45b)
where mean angular velocity components on the xr principal axis are due
to systematic rotation around xs and xt axes, (Epot)rr is the self potential-
energy tensor, ζrr may be conceived as generalized anisotropy parameters
10
(Caimmi & Marmo 2005; C06; C07), and eTrdm is the effective random kinetic
energy i.e. the right amount needed for an instantaneous configuration to
satisfy the usual tensor virial equations of the second order, defined by the
effective anisotropy parameters (C07):
=
ζpp
ζ
(eTrdm)pp
eTrdm
eζpp =
eζ11 + eζ22 + eζ33 = 1 ;
;
p = 1, 2, 3 ;
(46a)
(46b)
and the condition, ζ = 1, or ζpp = eζpp, p = 1, 2, 3, makes Eqs. (43) reduce to
their standard counterparts. To get further insight, a microscopical analysis
is needed.
In the special case of axisymmetric configurations, I11 = I22, (Epot)11 =
(Epot)22, and the combination of the related tensor virial equations, expressed
by Eq. (43), yields:
p = 1, 2 ;
q = 2, 1 ;
(47)
Ipph(fΩq)2 − (fΩp)2i = Mσ2(ζqq − ζpp) ;
[(fΩq)2 − (fΩp)2], is counterbalanced by an
where a figure rotation excess,
anisotropy excess, (ζqq − ζpp);
in particular, a null figure rotation excess
implies a null anisotropy excess and vice versa. Accordingly, a flattening
on the (O xp xr) principal plane, induced by the figure rotation excess, has
to be counterbalanced by an elongation on the xq principal axis, induced by
the anisotropy excess, to yield an axisymmetric configuration with respect
to the xr principal axis.
4 Microscopical analysis of systematic and ran-
dom motions
Given a collisionless R fluid, let N be the total number of particles and
m the mean particle mass in absence of mass segregation i.e.
local and
global mean particle mass coincide. For simplicity, the equivalent description
(C07) involving N identical particles of mass, m, shall be considered. With
regard to the (O xs xt) principal plane, let vφr be the tangential velocity
component on the above mentioned plane. It is worth noting (Meza 2002)
that the distribution function is independent of the sign of vφr , and the
11
whole set of possible configurations are characterized by an equal amount of
both kinetic and potential energy. Numerical simulations show that spherical
systems, even if rapidly rotating, are dynamically stable after reversion of
the tangential velocity component in an assigned fraction of particles (Meza
2002).
For sake of simplicity, let the initial configuration be nonrotating (vφr = 0)
and with isotropic random velocity component distribution (ζ11 = ζ22 = ζ33).
In the case under discussion of identical particles, m(i) = m, 1 ≤ i ≤ N,
the centre of inertia velocity components, vCr, equal the related arithmetic
means:
vCr =
= vr
;
(48)
m
=
v(i)
r
NXi=1
Nm
m(i)v(i)
r
m(i)
NXi=1
NXi=1
and the moments of inertia, I ′
rr, reduce to:
I ′
rr =
NXi=1
r i2
m(i)hw(i)
= m
r i2
NXi=1hw(i)
= MR2
Gr
;
(49)
where RGr = [w2
r ]1/2 is the curl radius with respect to the xr axis.
The weighted mean, mean square, and rms tangential velocity compo-
nents, expressed by Eqs. (27)-(29), read:
m(i)v(i)
φr =
v(i)
φr = vφr
;
m
M
NXi=1
φri2
m(i)hv(i)
=
m
M
φri2
NXi=1hv(i)
= (v2
φr) − (vφr )2 = (σφrφr )2
NXi=1
NXi=1
1
M
1
M
φr ) =
gvφr =
g(v2
(cid:16)σeφreφr(cid:17)2
and Eqs. (30) reduce to:
= (v2
φr ) ;
(50)
(51)
(52)
(53a)
(53b)
(53c)
;
;
φr ) = R2
(v2
(vφr )2 = R2
(σφrφr )2 = R2
r) ;
Gr g(Ω2
Gr(fΩr)2 ;
Gr(cid:16)σfΩrfΩr(cid:17)2
12
which relate weighted angular velocities around the xr axis to mean tangential
velocity components on the (O xs xt) principal plane.
At this stage, let the tangential velocity component of a fraction, n/N,
of particles, be reversed in equal sense (from clockwise to counterclockwise
or vice versa), according to the following assumptions.
(i) Both the number, n, of particles where the tangential velocity component
has been reversed, and the number, N − n, of particles which remain
unchanged, are sufficiently large, 1 ≪ n ≪ N, 0 ≤ n ≤ Int(N/2).
(ii) The fraction, nk/Nk, of particles where the tangential velocity compo-
nent has been reversed, within a generic volume element, Sk, is inde-
pendent of the volume element, nk/Nk = n/N.
(iii) The system is made of identical particles, m(i) = m, M = mN.
(iv) After tangential velocity components have been reversed in nk particles
within a generic volume element, Sk, on a total of Nk, a second set of
nk particles (among the remaining Nk −nk) exists, where the tangential
velocity component of any particle equals its counterpart belonging to
the first set.
Obviously, mean square tangential velocity components, (v2
In the following, the above process shall be quoted as "the reversion process".
φr ), are left un-
changed by the reversion process. On the contrary, mean tangential velocity
components after the reversion process read:
vφr =
1
N
NXi=1
v(i)
φr =
(54)
1
N
2nXi=1
v(i)
φr +
NXi=2n+1
v(i)
φr ;
where the first sum within brackets relates to particles where the reversion
process has occurred and their counterparts with equal tangential velocity
components, while the second sum comprises the remaining particles and
necessarily equals the mean tangential velocity component before the occur-
rence of the reversion process, which is null in the case under discussion.
Accordingly, Eq. (54) reduces to:
vφr =
2n
N
(vφr )n =
(vφr )n ;
1
2n
2nXi=1
v(i)
φr =
1
n
nXi=1
v(i)
φr
;
13
(55a)
(55b)
keeping in mind that the first sum is performed on couples of particles with
equal tangential velocity components.
The validity of Eqs. (54) and (55) still maintains if tangential velocity
components, vφr , are replaced by axial velocity components, vr. The combi-
nation of Eqs. (48) and (55a) yields:
vCr = vr =
2n
N
(vr)n ;
(56)
which is the velocity component of the centre of inertia with respect to the
xr principal axis, after the reversion process.
The total kinetic energy is left unchanged by the reversion process but,
on the other hand, a fraction of random motion kinetic energy is turned into
systematic motion kinetic energy. In the following, the reversion process shall
be discussed with further details for a number of different situations.
4.1 Tangential velocity component reversion
Performing the reversion process on a given fraction of particles, n/N, with
respect to tangential velocity components, implies the conversion of random
(rotation) motion kinetic energy into systematic (rotation) motion kinetic
energy, as:
∆(Trdm)φrφr = −∆(Tsys)φrφr = −
1
2
M(vφr )2 = −
2n
N
nm[(vφr )n]2 ;
(57)
where the remaining parameters are left unchanged.
The occurrence of the reversion process implies the following energy
changes:
Trdm → Trdm −
2n
N
nm[(vφr )n]2 ;
(Trdm)φrφr → (Trdm)φrφr −
nm[(vφr )n]2 ;
2n
N
nm[(vφr )n]2 ;
ℓ = s, t ;
1
2
2n
N
(Trdm)ℓℓ → (Trdm)ℓℓ −
Tsys → 0 +
2n
N
nm[(vφr )n]2 ;
(Tsys)φrφr → 0 +
2n
N
nm[(vφr )n]2 ;
14
(58)
(59)
(60)
(61)
(62)
(Tsys)ℓℓ → 0 +
1
2
2n
N
nm(vφr )2
n ;
ℓ = s, t ;
(63)
while the contributions from random radial motions along the equatorial
plane, (Trdm)wrwr , and the rotation axis, (Trdm)rr, remain unchanged.
With the system being relaxed, ζ = 1, in the case under discussion, the
generalized and effective anisotropy parameters, ζpp and eζpp, coincide with
their counterparts related to the usual tensor virial equations, and Eqs. (45a)
and (46a) take the explicit form (C06):
ζℓℓ =
ζrr =
(1/3)Trdm − (2n/N)(n/2)m[(vφr )n]2
Trdm − (2n/N)nm[(vφr )n]2
(1/3)Trdm
Trdm − (2n/N)nm[(vφr )n]2
;
;
ℓ = s, t ;
(64a)
(64b)
where the special case, n = 0, relates to the initial configuration, character-
ized by isotropic random velocity component distributions (ζpp = 1/3) and
no figure rotation.
In the extreme case where the reversion process is completed, n = N/2,
the changes expressed by Eqs. (58)-(63) take the form:
Trdm → Trdm −
N
2
m[(vφr )N/2]2 ;
(Trdm)φrφr → (Trdm)φrφr −
N
2
m[(vφr )N/2]2 ;
(Trdm)ℓℓ → (Trdm)ℓℓ −
N
4
m[(vφr )N/2]2 ;
ℓ = s, t ;
Tsys → 0 +
N
2
m[(vφr )N/2]2 ;
(Tsys)φrφr → 0 +
(Tsys)ℓℓ → 0 +
N
4
m[(vφr )N/2]2 ;
N
2
m[(vφr )N/2]2 ;
ℓ = s, t ;
(65)
(66)
(67)
(68)
(69)
(70)
similarly, Eqs. (64) take the form:
ζℓℓ =
ζrr =
(1/3)Trdm − (N/4)m[(vφr )N/2]2
Trdm − (N/2)m[(vφr )N/2]2
(1/3)Trdm
Trdm − (N/2)m[(vφr )N/2]2
;
;
ℓ = s, t ;
(71a)
(71b)
15
in any case, the anisotropy excess, ζℓℓ − ζrr < 0, is counterbalanced by figure
rotation.
4.2 Axial velocity component reversion
Performing the reversion process on a given fraction of particles, n/N, with
respect to axial velocity components, implies the conversion of random (trans-
lation) motion kinetic energy into systematic (translation) motion kinetic
energy, as:
∆(Trdm)rr = −∆(Tsys)rr = −
1
2
M(vr)2 = −
2n
N
nm[(vr)n]2 ;
(72)
where the remaining parameters are left unchanged.
The reversion process implies the following energy changes:
(Trdm)rr → (Trdm)rr −
2n
N
nm[(vr)n]2 ;
Trdm → Trdm −
2n
N
nm[(vr)n]2 ;
Tsys → 0 +
2n
N
(Tsys)rr → 0 +
nm[(vr)n]2 ;
2n
N
nm[(vr)n]2 ;
(73)
(74)
(75)
(76)
while the contributions from random motions along the xs and xt principal
axis, (Trdm)ss and (Trdm)tt, remain unchanged.
With the system being relaxed, ζ = 1, in the case under discussion, the
generalized and effective anisotropy parameters, ζpp and eζpp, coincide with
their counterparts related to the usual tensor virial equations, and Eqs. (45a)
and (46a) take the explicit form:
ζℓℓ =
ζrr =
(1/3)Trdm
;
Trdm − (2n/N)nm[(vr)n]2
(1/3)Trdm − (2n/N)nm[(vr)n]2
Trdm − (2n/N)nm[(vr)n]2
ℓ = s, t ;
(77a)
;
(77b)
where the special case, n = 0, relates to the initial configuration, character-
ized by isotropic random velocity component distributions (ζpp = 1/3) and
no figure rotation.
16
In the extreme case where the reversion process is completed, n = N/2,
the changes expressed by Eqs. (73)-(76) take the form:
Trdm → Trdm −
N
2
m[(vr)N/2]2 ;
(Trdm)rr → (Trdm)rr −
N
2
m[(vr)N/2]2 ;
Tsys → 0 +
N
2
m[(vr)N/2]2 ;
(Tsys)rr → 0 +
N
2
m[(vr)N/2]2 ;
(78)
(79)
(80)
(81)
similarly, Eqs. (77) take the form:
ζℓℓ =
ζrr =
Trdm − (N/2)m[(vr)N/2]2
(1/3)Trdm − (N/2)m[(vr)N/2]2
Trdm − (N/2)m[(vr)N/2]2
(1/3)Trdm
;
ℓ = s, t ;
(82a)
;
(82b)
in any case, the anisotropy excess, ζℓℓ − ζrr > 0, is counterbalanced by (imag-
inary) figure rotation.
4.3 Change of reference frame and imaginary rotation
Performing the reversion process on a given fraction of particles, n/N, implies
the conversion of random motion (translation) kinetic energy into systematic
motion (translation) kinetic energy, with respect to the xr principal axis,
according to Eqs. (72)-(76). The related kinetic-energy tensor component of
the centre of inertia, by use of Eqs. (56) and (72) reads:
(TC)rr =
1
2
M(vr)2 =
2n
N
nm[(vr)n]2 = −∆(Trdm)rr
;
(83)
which, in the case under discussion, coincides with the kinetic energy of
the centre of inertia, TC, being (TC)ss = (TC)tt = 0. In the centre of inertia
reference frame after the reversion process, the random motion kinetic-energy
tensor component related to the xr axis, (T ′
rdm)rr, by use of Eqs. (56) and (83),
takes the expression:
(T ′
rdm)rr =
1
2
m
NXi=1hv(i)
r − vCri2
= (Trdm)rr −
1
2
M(vr)2 = (Trdm)rr − (TC)rr
;
(84)
17
where the kinetic energy, TC = (TC)rr, is hidden by the change of reference
frame (e.g., LL66, Chap. II, § 8).
Let the i-th particle be at the distance, w(i)
]2}1/2, from
the xr principal axis, with velocity component, v(i)
r . The imaginary angu-
lar velocity (Caimmi 1996; C06; C07), iΩ(i)
r , can be defined in such a way
the translational kinetic energy along the xr axis is counterbalanced by the
imaginary rotational kinetic energy around the xr axis, as:
r = {[x(i)
s ]2 + [x(i)
t
1
2
m[v(i)
r ]2 +
v(i)
r
w(i)
r
Ω(i)
r =
1
2
mhw(i)
r i2
r i2hiΩ(i)
= 0 ;
;
(85)
(86)
where the index, i, is related to the i-th particle, and the factor, i, is the
imaginary unit. In this view, the velocity components on the xr principal
axis may be translated into imaginary tangential velocity components on the
(O xs xt) principal plane, as:
(ivφr )2 = (iwrΩr)2 = −v2
r
;
(87)
according to Eqs. (85) and (86).
Let imaginary rotation around the xr axis be imparted to the particles
where the reversion process has been occurred, and their counterparts with
equal imaginary tangential velocity components, as prescribed by Eq. (87)
particularized to the mean axial velocity component, vr, expressed by Eq. (56).
The related increment in imaginary kinetic energy reads:
∆(Tsys)φrφr =
1
2
M(ivφr )2 =
2n
N
nm[(ivφr )n]2 ;
(88)
and the combination of Eqs. (72), (83), (87), and (88) yields:
∆(Tsys)φrφr = −∆(Tsys)rr = −(TC)rr = ∆(Trdm)rr
;
(89)
the above results may be reduced to a single statement.
Theorem 1. Given a R fluid with isotropic random velocity distribution
and no figure rotation, let the axial velocity component reversion process
be performed on a given fraction of particles, n/N, with respect to the xr
18
principal axis. Then turning to the centre of inertia reference frame with a
kinetic energy loss, ∆(Tsys)rr = (TC)rr, is equivalent to put the initial con-
figuration into imaginary rotation around the xr principal axis, with square
mean tangential velocity component, (ivφr )2 = −(vr)2, with a kinetic energy
gain, ∆(Tsys)φrφr = −(TC)rr.
Accordingly, Eqs. (73)-(76) are replaced by the following:
Tsys → 0 −
2n
N
nm[(vφr )n]2 ;
(Tsys)φrφr → 0 −
(Tsys)ℓℓ → 0 −
1
2
2n
N
2n
N
nm[(vφr )n]2 ;
nm[(vφr )n]2 ;
(90)
(91)
(92)
while the contributions from random motions remain unchanged, and the
random velocity distribution remains isotropic (ζ11 = ζ22 = ζ33 = 1/3).
In the extreme case where the reversion process is complete, n = N/2,
and the maximum amount of available imaginary rotation has been attained,
the changes expressed by Eq. (90)-(92) take the form:
Tsys → 0 −
N
2
m[(vφr )N/2]2 ;
(Tsys)φrφr → 0 −
(Tsys)ℓℓ → 0 −
N
4
m[(vφr )N/2]2 ;
N
2
m[(vφr )N/2]2 ;
(93)
(94)
(95)
the above results may be reduced to a single statement.
Theorem 2. Given a R fluid with isotropic random velocity distribution
and no figure rotation, let the axial velocity component reversion process be
performed on a given fraction of particles, n/N, with respect to the xr princi-
pal axis, and the reference frame changed into the centre of inertia reference
frame. Then the resulting configuration with anisotropy excess, ζℓℓ − ζrr > 0,
Eqs. (77), is equivalent to the initial configuration with null anisotropy ex-
cess, ζℓℓ − ζrr = 0, and imaginary rotation around the xr principal axis, with
square mean tangential velocity component, (ivφr )2 = −(vr)2.
19
4.4 Anisotropy excess and imaginary rotation
Given a nonrotating (fΩr = 0) isotropic (ζ11 = ζ22 = ζ33 = 1/3) configuration,
let the tangential velocity component on the (O xs xt) principal plane be
reversed in such a way Eqs. (65)-(71) hold. Let an equal amount of real and
imaginary tangential velocity component on the (O xs xt) principal plane, be
imparted to each particle for a total contribution equal to (2/3)frTrdm and
−(2/3)frTrdm, respectively, where fr is a positive real number, which leaves
the total energy unchanged. Let the reversion process be repeated for real
tangential velocity components, to attain a null (real) figure rotation. The
related changes, with respect to the initial configuration, are:
2
3
Trdm →(cid:18)1 +
(Trdm)φrφr →(cid:18)2
2(cid:18)2
(Trdm)ℓℓ →
3
3
fr(cid:19) Trdm ;
+
+
2
3
2
3
fr(cid:19) Trdm ;
fr(cid:19) Trdm ;
1
2
3
Tsys → 0 −
frTrdm ;
(Tsys)φrφr → 0 −
(Tsys)ℓℓ → 0 −
1
2
2
3
2
3
frTrdm ;
frTrdm ;
(96)
(97)
(98)
(99)
(100)
(101)
and the related anisotropy parameters read:
ζℓℓ =
ζrr =
[(1/3) + (1/3)fr]Trdm
[1 + (2/3)fr]Trdm
=
(1/3)Trdm
[1 + (2/3)fr]Trdm
=
3 + 2fr
1 + fr
3 + 2fr
1
;
;
ℓ = s, t ;
(102a)
(102b)
where the anisotropy excess, ζℓℓ − ζrr = fr/(3 + 2fr) > 0, is counterbalanced
by imaginary rotation, according to an initial configuration with isotropic
velocity component distribution and no figure rotation. As application of the
above results, two significative examples shall be taken into consideration.
First example. Nonrotating systems flattened by anisotropic velocity com-
ponent distribution (σ11 = σ22 > σ33).
20
Let tangential velocity components on the (O x1 x2) principal plane be re-
versed in a convenient fraction of particles, n/N, to yield a convenient amount
of figure rotation together with isotropic velocity component distribution
(σ′
22 = σ33), as sketched in Fig. 1.
11 = σ′
The combination of Eqs. (52), (53), and (55a) yields:
1
R2
Gr
=
4n2
N 2 [(vφr )n]2 ;
1
(σφrφr )2 =
R2
Gr
Gr ((v2
1
R2
φr ) −
4n2
N 2 [(vφr )n]2) ;
(103)
(104)
(fΩr)2 =
(cid:16)σfΩrfΩr(cid:17)2
while the mean square velocity components are left unchanged by the oc-
currence of the reversion process. The substitution of Eqs. (103) and (104)
into (37b) and (37c) shows the dependence of systematic and random motion
tangential kinetic-energy tensor components on the reversion process.
In the case under discussion (r = 3), the random velocity component
distribution has to be isotropic after the reversion process, which makes
Eqs. (60) and (63) reduce to:
(Trdm)ℓℓ =
(Tsys)ℓℓ =
M
2
M
2 hσ2
2 (cid:16)σ2
M
ℓℓ − σ2
33(cid:17)i =
ℓℓ −(cid:16)σ2
33(cid:17) ;
ℓℓ − σ2
4n2
N 2 M[(vφr )n]2 =
1
2
ℓ = 1, 2 ;
2n
N
nm[(vφr )n]2 =
σ2
33 ;
ℓ = 1, 2 ;
(105)
(106)
M
2
(vφr )2 =
M
2 (cid:16)σ2
ℓℓ − σ2
33(cid:17) ; (107)
which defines the configuration after the reversion process.
Second example. Systems elongated by anisotropic velocity component
distribution (σ11 = σ22 < σ33) with no figure rotation.
Let a convenient amount of real and imaginary figure rotation, vφ3 and ivφ3,
with respect to the x3 principal axis, be imparted to the system. Then the
kinetic energy remains unchanged. Concerning real rotation, let the reversion
process be performed on one half particles, leaving imaginary figure rotation
together with isotropic velocity component distribution (σ′
22 = σ33), as
sketched in Fig. 2. Accordingly, Eqs. (103) and (104) hold for imaginary tan-
gential velocity components on the (O x1 x2) principal plane and imaginary
figure rotation around the x3 principal axis.
11 = σ′
21
x3
✻
x3
✻
✵
O
r
x1
✠
✲
x2
O
r
✲
x2
x1
✠
σ11 = σ22 > σ33 ;
(vφ3 )2 = 0 ;
σ
′
11 = σ
′
22 = σ33 ;
(vφ3 )2 = σ
2
11 − σ
2
33 = σ
2
22 − σ
2
33 ;
Figure 1: After reversion of tangential velocity components in a convenient
fraction of particles, with respect to the (Ox1x2) principal plane, a configu-
ration flattened by anisotropic velocity component distribution (σ11 = σ22 >
σ33) with no figure rotation (vφ3 = 0), left picture, is turned into a configu-
ration flattened by figure rotation [vφ3 = (σ2
33)1/2] with
isotropic velocity component distribution (σ′
22 = σ33), right picture.
The symbol, ✵, denotes figure rotation around the related principal axis.
33)1/2 = (σ2
11 = σ′
11 − σ2
11 − σ2
22
x3
✻
x3
✻
✵
O
r
x1
✠
✲
x2
O
r
✲
x2
x1
✠
σ11 = σ22 < σ33 ;
(ivφ3 )2 = 0 ;
σ
′
11 = σ
′
22 = σ33 ;
(ivφ3 )2 = σ
2
11 − σ
2
33 = σ
2
22 − σ
2
33 ;
Figure 2: After imparting a convenient amount of real and imaginary fig-
ure rotation, vφ3 and ivφ3, with respect to the x3 principal axis, to the
system, and reversing real tangential velocity components on one half par-
ticles, a configuration elongated by anisotropic velocity component distri-
bution (σ11 = σ22 < σ33) with no figure rotation (vφ3 = 0),
left pic-
ture, is turned into a configuration elongated by (imaginary) figure rotation
[ivφ3 = (σ2
33)1/2] with isotropic velocity component
distribution (σ′
22 = σ33), right picture. The symbol, ✵, denotes figure
rotation around the related principal axis.
33)1/2 = (σ2
11 − σ2
11 − σ2
11 = σ′
23
In the case under discussion (r = 3), an isotropic velocity component
distribution after the reversion process implies the validity of Eqs. (105),
(106), and (107), where the tangential velocity components are imaginary
(σℓℓ < σ33), and the configuration after the reversion process is completely
defined.
4.5 Tangential velocity component reversion in the gen-
eral case
In the general case of anisotropic velocity component distribution (σ11 6=
0), with regard to the (O xs xt) principal plane, let the tangential velocity
component reversion process be applied to one half particles in such a way
σ22 6= σ33) and figure rotation around each principal axis (fΩ1 6= fΩ2 6= fΩ3 6=
no figure rotation around the xr principal axis occurs, fΩr = 0. Keeping in
mind Eqs. (50)-(53), the related energy changes read:
(Trdm)φrφr → (Trdm)φrφr +
(Tsys)φrφr → (Tsys)φrφr −
1
2
M(vφr )2 ;
1
2
M(vφr )2 = 0 ;
vφr = (vφr )N/2 =
1
N
NXi=1
v(i)
φr
;
∆(Tsys)φrφr = −∆(Trdm)φrφr = −
M(vφr )2 = −
MR2
(108)
(109)
(110)
(113)
1
2
1
Mσ2
ℓℓ
2
1
2
Gr(fΩr)2 ; (111)
(112)
;
ℓ = 1, 2 ;
[(Trdm)φrφr ]ℓℓ + [(Trdm)wrwr ]ℓℓ =
(Trdm)rr =
1
2
Mσ2
rr
;
where the rms velocity components, σ2
configuration.
ℓℓ and σ2
rr, are related to the initial
The changes in anisotropy paramenters, ζpp = σ2
pp/σ2, read:
ζℓℓ →
ζrr →
ζℓℓ + (1/2)∆ζℓℓ
1 + ∆ζℓℓ
ζrr
;
1 + ∆ζℓℓ
24
;
ℓ = s, t ;
(114a)
(114b)
∆ζℓℓ =
(vφr )2
σ2
;
ℓ = s, t ;
(114c)
where the rms velocity, σ2, is related to the initial configuration.
11, σ2
22, σ2
The application of the above procedure to the x1 and x2 principal axes,
makes the transition from an initial configuration with rms velocity compo-
nents, σ2
to a final configuration with rms velocity components, (σ′
33, and figure rotation around the principal axes, fΩ1, fΩ2, fΩ3,
and figure rotation around the principal axes, 0, 0, fΩ3. The related energy
changes read:
11)2, (σ′
22)2, (σ′
33)2,
M(vφ2)2 ;
M(vφ1)2 ;
Mh(vφ1)2 + (vφ2)2i
(Trdm)11 → (Trdm)11 +
(Trdm)22 → (Trdm)22 +
(Trdm)33 → (Trdm)33 +
1
4
1
4
1
4
Trdm → Trdm +
(Trdm)pp =
1
2
1
2
Mσ2
Mh(vφ1)2 + (vφ2)2i
;
pp ;
p = 1, 2, 3 ;
;
(115)
(116)
(117)
(118)
(119)
(120a)
(120b)
(120c)
(120d)
and the related changes in anisotropy parameters are:
ζ11 →
ζ22 →
ζ33 →
;
ζ11 + (1/2)∆ζ11
1 + ∆ζ11 + ∆ζ22
ζ22 + (1/2)∆ζ22
1 + ∆ζ11 + ∆ζ22
ζ33 + (1/2)[∆ζ11 + ∆ζ22]
;
1 + ∆ζ11 + ∆ζ22
;
∆ζ11 =
(vφ2)2
σ2
;
∆ζ22 =
(vφ1)2
σ2
;
;
the above results may be restricted to a single statement.
Theorem 3. Given a R fluid, a convenient application of the tangential
velocity component reversion process makes an adjoint configuration where
figure rotation occurs around a single principal axis, that is a R3 fluid.
Accordingly, the results valid for R3 fluids (C07) may be extended to the
general case of R fluids.
25
5 Discussion
As suggested in earlier attempts (Caimmi 1996; C06; C07), the equivalence
between a variation in figure rotation and in anisotropy excess, may provide
a useful tool for the description of collisionless fluids. The discussion here
shall be focused on the spin parameter (Peebles 1969, 1971):
λ2 = −
J 2E
G2M 5
;
(121)
where G is the gravitation constant, M the total mass, J the total angular
momentum, and E the total energy. The above formulation includes four
possibilities, namely (i) real rotation (J 2 ≥ 0) and bound system (E < 0),
which is the sole currently used in literature; (ii) imaginary rotation (J 2 < 0)
and bound system (E < 0); (iii)real rotation (J 2 ≥ 0) and unbound system
(E ≥ 0); (iv) imaginary rotation (J 2 < 0) and unbound system (E ≥ 0).
Accordingly, the spin parameter attains real values in cases (i) and (iv), and
imaginary values in cases (ii) and (iii).
In the light of the current model, oblate-like and prolate-like configura-
tions belong to cases (i) and (ii) outlined above, while cases (iii) and (iv) rep-
resent unbound structures for which the virial equations do not hold. Then
the comparison with observations and/or computations, must be restricted
to bound configurations.
The spin parameter distribution is usually fitted using a lognormal dis-
tribution (e.g., van den Bosh 1998; Gardner 2001; Ballin & Steinmetz 2005;
Hernandez et al. 2007) or, in general, dependent on log λ (e.g., Bett et al.
2007), in dealing with real rotation. The inclusion of imaginary rotation
would imply use of λ2 instead of λ as independent variable, allowing for both
positive (real rotation) and negative (imaginary rotation) values.
The following procedure should be followed for calculating λ2 from obser-
vations and/or computations: (1) determine the inertia tensor and the prin-
cipal axes of inertia for an assigned matter distribution; (2) determine the
potential-energy tensor; (3) determine the anisotropy parameters using the
generalized virial equations; (4) perform the reversion process with respect
to two principal axes of inertia to leave figure rotation around the third one
(R3 fluid); (5) convert the anisotropy excess into (real or imaginary) figure
rotation to obtain isotropic velocity component distribution (ζ11 = ζ22 = ζ33);
(6) evaluate the spin parameter; (7) act as already done for all the sample
26
objects; (8) determine the distribution of the spin parameter, P (λ2), with re-
spect to the sample of adjoints configurations, where the velocity component
distribution is isotropic.
The related results could provide further insight on the formation and
the evolution of large-scale collisionless fluids, such as galaxies and galaxy
clusters.
6 Conclusion
A theory of collisionless fluids has been developed in a unified picture, where
nonrotating (fΩ1 = fΩ2 = fΩ3 = 0) figures with isotropic (σ11 = σ22 = σ33) ran-
dom velocity component distributions and rotating (fΩ1 6= fΩ2 6= fΩ3) figures
with anisotropic (σ11 6= σ22 6= σ33) random velocity component distribu-
tions, make adjoints configurations to the same system. R fluids have been
defined as ideal, self-gravitating fluids satisfying the virial theorem assump-
tions (e.g., LL66, Chap. II, § 10; C07), in presence of figure rotation around
each principal axis of inertia.
To this aim, mean and rms angular velocities and mean and rms tangential
velocity components have been expressed, by weighting on the moment of
inertia and the mass, respectively. The figure rotation has been defined as
the mean angular velocity, weighted on the moment of inertia, with respect
to a selected axis.
The generalized tensor virial equations (Caimmi & Marmo 2005) have
been formulated for R fluids and further attention has been devoted to ax-
isymmetric configurations where, for selected coordinate axes, a variation in
figure rotation has to be counterbalanced by a variation in anisotropy excess
and vice versa.
A microscopical analysis of systematic and random motions has been
performed under a few general hypotheses, by reversing the sign of tangential
or axial velocity components of an assigned fraction of particles, leaving the
distribution function and other parameters unchanged (Meza 2002).
The application of the reversion process to tangential velocity compo-
nents, has been found to imply the conversion of random motion rotation
kinetic energy into systematic motion rotation kinetic energy. The applica-
tion of the reversion process to axial velocity components, has been found to
imply the conversion of random motion translation kinetic energy into sys-
27
tematic motion translation kinetic energy, and the loss related to a change of
reference frame has been expressed in terms of systematic motion (imaginary)
rotation kinetic energy.
A number of special situations have been investigated with further detail.
It has been found that a R fluid always admits an adjoint configuration
where figure rotation occurs around only one principal axis of inertia (R3
fluid), which implies that all the results related to R3 fluids (Caimmi 2007)
may be extended to R fluids.
Finally, a procedure has been sketched for deriving the spin parameter
distribution (including imaginary rotation) from a sample of observed or
simulated large-scale collisionless fluids i.e. galaxies and galaxy clusters.
References
[1] Ballin, J., Steinmetz, M.: 2005, ApJ, 627, 647.
[2] Bett, P., Eke, V., Frenk, C.S., et al.: 2007, MNRAS, 376, 215.
[3] Binney, J., & Tremaine, S.: 1987, Galactic Dynamics, Princeton Univer-
sity Press, Princeton. (BT87)
[4] Caimmi, R.: 1996, AN, 317, 401.
[5] Caimmi, R.: 2006, AN, 327, 925. (C06)
[6] Caimmi, R. 2007, SerAJ, 174, 13. (C07)
[7] Caimmi, R., Marmo, C.: 2005, AN, 326, 465.
[8] Chandrasekhar, S.: 1969, Ellipsoidal Figures of Equilibrium, Yale Uni-
versity Press, New Haven.
[9] Gardner, J.P.: 2001, ApJ, 557, 616.
[10] Hernandez, X., Park, C., Cervantes-Sodi, B., Choi, Y.: 2007, MNRAS,
375, 163.
[11] Landau, L., Lifchitz, E.: 1966, Mecanique, Mir, Moscow. (LL66)
[12] Lynden-Bell, D.: 1960, MNRAS, 120, 204.
28
[13] Lynden-Bell, D.: 1962, MNRAS, 124, 1.
[14] Meza, A.: 2002, A&A, 395, 25.
[15] Peebles, P.J.E.: 1969, ApJ, 155, 393.
[16] Peebles, P.J.E.: 1971, A&A, 11, 377.
[17] van den Bosh, F.C.: 1998, ApJ, 507, 601.
29
|
astro-ph/9905127 | 1 | 9905 | 1999-05-11T18:16:35 | Python I, II, and III CMB Anisotropy Measurement Constraints on Open and Flat-Lambda CDM Cosmogonies | [
"astro-ph"
] | We use Python I, II, and III cosmic microwave background anisotropy data to constrain cosmogonies. We account for the Python beamwidth and calibration uncertainties. We consider open and spatially-flat-Lambda cold dark matter cosmogonies, with nonrelativistic-mass density parameter Omega_0 in the range 0.1--1, baryonic-mass density parameter Omega_B in the range (0.005--0.029) h^{-2}, and age of the universe t_0 in the range (10--20) Gyr. Marginalizing over all parameters but Omega_0, the combined Python data favors an open (spatially-flat-Lambda) model with Omega_0 simeq 0.2 (0.1). At the 2 sigma confidence level model normalizations deduced from the combined Python data are mostly consistent with those drawn from the DMR, UCSB South Pole 1994, ARGO, MAX 4 and 5, White Dish, and SuZIE data sets. | astro-ph | astro-ph |
KSUPT-99/1, KUNS-1552
February 1999
Python I, II, and III CMB Anisotropy Measurement Constraints on Open and
Flat-Λ CDM Cosmogonies
Gra¸ca Rocha1,2, Rados law Stompor3,4, Ken Ganga5, Bharat Ratra1, Stephen R. Platt6,
Naoshi Sugiyama7, and Krzysztof M. G´orski8,9
ABSTRACT
We use Python I, II, and III cosmic microwave background anisotropy data
to constrain cosmogonies. We account for the Python beamwidth and calibration
uncertainties. We consider open and spatially-flat-Λ cold dark matter cosmogonies,
with nonrelativistic-mass density parameter Ω0 in the range 0.1 -- 1, baryonic-mass
density parameter ΩB in the range (0.005 -- 0.029)h−2 , and age of the universe t0 in
the range (10 -- 20) Gyr. Marginalizing over all parameters but Ω0, the combined
Python data favors an open (spatially-flat-Λ) model with Ω0 ≃ 0.2 (0.1). At the 2 σ
confidence level model normalizations deduced from the combined Python data are
mostly consistent with those drawn from the DMR, UCSB South Pole 1994, ARGO,
MAX 4 and 5, White Dish, and SuZIE data sets.
Subject headings: cosmic microwave background -- cosmology: observations -- large-scale
structure of the universe
1.
Introduction
Ganga et al. (1997a, hereafter GRGS) developed a technique to account for uncertainties,
such as those in the beamwidth and the calibration, in likelihood analyses of cosmic microwave
1Department of Physics, Kansas State University, Manhattan, KS 66506.
2Centro de Astrof´ısica da Universidade do Porto, Rua das Estrelas s/n, 4100 Porto, Portugal.
3Center for Particle Astrophysics, University of California, Berkeley, CA 94720.
4Copernicus Astronomical Center, Bartycka 18, 00-716 Warszawa, Poland.
5Infrared Processing and Analysis Center, MS 100 -- 22, California Institute of Technology, Pasadena, CA 91125.
6Snow and Ice Research Group, University of Nebraska, Lincoln, NE 68583-0850.
7Department of Physics, Kyoto University, Kitashirakawa-Oiwakecho, Sakyo-ku, Kyoto 606-8502, Japan.
8Theoretical Astrophysics Center, Juliane Maries Vej 30, 2100 Copenhagen Ø, Denmark.
9Warsaw University Observatory, Aleje Ujazdowskie 4, 00-478 Warszawa, Poland.
-- 2 --
background (CMB) anisotropy data. This technique has been used in conjunction with
theoretically-predicted CMB anisotropy spectra in analyses of the Gundersen et al. (1995) UCSB
South Pole 1994 data, the Church et al. (1997) SuZIE data, the MAX 4+5 data (Tanaka et al.
1996; Lim et al. 1996), the Tucker et al. (1993) White Dish data, and the de Bernardis et al.
(1994) ARGO data (GRGS; Ganga et al. 1997b, 1998; Ratra et al. 1998, 1999a, hereafter R99a).
A combined analysis of all these data sets is presented in Ratra et al. (1999b, hereafter R99b).
In this paper we present a similar analysis of CMB anisotropy data from the Python I, II,
and III observations performed at the South Pole (Dragovan et al. 1994, hereafter D94; Ruhl et
al. 1995b, hereafter R95; Platt et al. 1997, hereafter P97). The Python detectors and telescope
are described by Ruhl (1993) and D94; also see Ruhl et al. (1995a) and Alvarez (1996). In what
follows we review the information needed for our analysis.
Python I, II, and III CMB data were taken in a frequency band centered at 90 GHz with four
bolometric detectors centered at the corners of a 2.◦75 by 2.◦75 square on the sky. The beam profiles
are well-approximated by a Gaussian of FWHM 0.◦75 ± 0.◦05 (one standard deviation uncertainty).
Observations were centered at α = 23.h37, δ = −49.◦44 (J2000.0). Python I and II data were taken
at a single telescope elevation. Python III data were taken at this fiducial elevation as well as two
additional elevations offset 2.◦75/3 on the sky above and below the fiducial elevation. The reduced
Python data are shown in Figure 1.
All of the Python measurements were made by switching the four beams horizontally across
the sky in a three-point pattern by rotating a vertical flat mirror at 2.5 Hz. This chopping pattern
was then combined with slow (typically 0.1 Hz) azimuthal beam switching of the entire telescope
to produce a four-beam response to a sky signal.
The chopper throw and azimuthal telescope beam switching were both 2.◦75 on the sky for
the Python I and II observations. Python I (hereafter I) resulted in 16 data points, 8 each from
the lower and upper rows of detectors (D94). Python II observed two sets of points on the sky
(R95), although one of the detectors did not work during this season. The first set of observations,
hereafter IIA, overlapped the I points and yielded 7 measurements from the lower row of detectors
and 8 from the upper row. The second set, hereafter IIB, measured the same number of points on
the sky as the first set, but these were offset in azimuth relative to the I points by −2.◦75/2 on the
sky.
Two series of measurements were made at each of the three elevations observed during the
Python III season (P97). For each series, the physical throw of the chopper was the same at all
elevations. Because the actual throw on the sky depends on elevation, the Python III beam was
smeared to a Gaussian FWHM of 0.◦82 ± 0.◦05 (one standard deviation uncertainty) when forming
the four-beam pattern to account for the imperfect overlap of the beams caused by this effect.
This procedure also accounted for the relative pointing uncertainty and fluctuations in the chopper
throw.
The first series of Python III measurements, hereafter IIIL, used the same chopper and beam
-- 3 --
switch parameters as Python I and II. The IIIL data consists of two sets of points taken at the
I elevation but offset in right ascension by −2.◦75/3 and −2 × 2.◦75/3 on the sky. Two sets of
measurements were also made at each of the lower and upper elevations, but these were offset in
right ascension by 0.◦0 and +2 × 2.◦75/3 on the sky relative to I.
The second series of Python III measurements, hereafter IIIS, were made with both the
chopper throw and telescope beam switch reduced to 2.◦75/3. For each of the three telescope
elevations, the IIIS data consists of points separated horizontally by 2.◦75/3 on the sky for each of
the 2 rows of detectors. Figure 1 shows the points observed by IIIL and IIIS at each elevation.
Together, I, II, and III densely sample a 5.◦5 by 22◦ region of the sky.
The 1 σ absolute calibration uncertainty in the Python data is 20% (D94; R95; P97) and is
accounted for in our analysis. The absolute pointing uncertainty is 0.◦1 (D94; R95) and is not
accounted for in our analysis.
In §2 we summarize the computational techniques used in our analysis. See GRGS and R99a
for detailed discussions. Results are presented and discussed in §3. Conclusions are given in §4.
2. Summary of Computation
The zero-lag window function for the Python observations are shown in Figure 2 and the
zero-lag window function parameters are in Table 1.
In this paper we focus on a spatially open CDM model and a spatially flat CDM model
with a cosmological constant Λ. These low density models are largely consistent with current
observational constraints.10 For recent discussions see Park et al. (1998), Retzlaff et al. (1998),
Croft et al. (1999), and Peebles (1999).
The models have Gaussian, adiabatic primordial energy-density power spectra. The
flat-Λ model CMB anisotropy computations use a scale-invariant energy-density perturbation
power spectrum (Harrison 1970; Peebles & Yu 1970; Zel'dovich 1972), as predicted in the
simplest spatially-flat inflation models (Guth 1981; Kazanas 1980; Sato 1981). The open model
computations use the energy-density power spectrum (Ratra & Peebles 1994, 1995; Bucher,
Goldhaber, & Turok 1995; Yamamoto, Sasaki, & Tanaka 1995) predicted in the simplest
open-bubble inflation models (Gott 1982). The computation of the CMB anisotropy spectra is
described by Stompor (1994) and Sugiyama (1995).
As discussed in R99a, the spectra are parameterized by their quadrupole-moment amplitude
10 While not considered in this paper, a time-variable cosmological "constant" dominated spatially-flat model is
also largely consistent with current data (e.g., Peebles & Ratra 1988; Sugiyama & Sato 1992; Ratra & Quillen 1992;
Frieman & Waga 1998; Ferreira & Joyce 1998; Wang & Steinhardt 1998; Carroll 1998; Hu et al. 1999; Huterer &
Turner 1999; Liddle & Scherrer 1999; Starobinsky 1998).
-- 4 --
Qrms−PS, the nonrelativistic-mass density parameter Ω0, the baryonic-mass density parameter ΩB,
and the age of the universe t0. The spectra are computed for a range of Ω0 spanning the interval
0.1 to 1 in steps of 0.1, for a range of ΩBh2 [the Hubble parameter h = H0/(100 km s−1 Mpc−1)]
spanning the interval 0.005 to 0.029 in steps of 0.004, and for a range of t0 spanning the
interval 10 to 20 Gyr in steps of 2 Gyr.
In total 798 spectra were computed to cover the
cosmological-parameter spaces of the open and flat-Λ models. Figure 2 shows examples of the
CMB anisotropy spectra used in our analysis. Other examples are in Figure 2 of R99a and Figure
1 of R99b.
While it is of interest to also consider other cosmological parameters, such as tilt or gravity
wave fraction or a time-variable cosmological "constant" (instead of a constant Λ), to make the
problem tractable we have focussed on the four parameters mentioned above. We emphasize
however that the results of the analysis are model dependent. For instance, a time-variable Λ
model would likely lead to a different constraint on Ω0 than that derived below in the constant Λ
model.
Following GRGS, for each of the 798 spectra considered the "bare" likelihood function is
computed at the nominal beamwidth and calibration, as well as at a number of other values of
the beamwidth and calibration determined from the measurement uncertainties. The likelihood
function used in the derivation of the central values and limits is determined by integrating
(marginalizing) the bare likelihood function over the beamwidth and calibration uncertainties
with weights determined by the measured probability distribution functions of the beamwidth
and the calibration. See GRGS for a more detailed discussion. When marginalizing over the
beamwidth uncertainty we have checked in a few selected cases that the five-point Gauss-Hermite
quadrature summation approximation to the integral agrees extremely well with the three-point
Gauss-Hermite approximation used by GRGS (and for most of the analysis in this paper). The
likelihoods are a function of four parameters mentioned above: Qrms−PS, Ω0, ΩBh2, and t0. We
also compute marginalized likelihood functions by integrating over one or more of these parameters
after assuming a uniform prior in the relevant parameters. The prior is set to zero outside the
ranges considered for the parameters. GRGS and R99a describe the prescription used to determine
central values and limits from the likelihood functions. In what follows we consider 1, 2, and 3 σ
highest posterior density limits which include 68.3, 95.4, and 99.7% of the area.
3. Results and Discussion
Table 2 lists the derived values of Qrms−PS and bandtemperature δTl for the flat bandpower
spectrum, for various combinations of the I, II, and III data. These numerical values account
for the beamwidth and calibration uncertainties. The last two δTl entries in Table 2 are quite
consistent with those derived by P97; the small differences reflect the different methods used to
account for beamwidth and calibration uncertainties here and in P97.
-- 5 --
For the flat bandpower spectrum the combined I, II, and III data average 1 σ δTl error bar
is ∼ 25%11 : Python data results in a very significant detection of CMB anisotropy, even after
accounting for beamwidth and calibration uncertainties. Note that the calibration uncertainty,
20%, is the most important contributor to this error bar.
A number of other interesting conclusions follow from the entries in Table 2. Comparing the
I+II and IIIL results, which are from experiments which probe almost identical angular scales, we
see that the IIIL amplitude is ∼ 1 σ higher than the I+II amplitude. Comparing the result from
the analysis of the coadded I and IIA data (which are from experiments with identical window
functions) and the result from the full analysis of the I and IIA data (which takes into account all
the spatial correlations), we see that the deduced amplitudes are almost identical. This is probably
mostly a reflection of the fact that the individual I and IIA amplitudes are almost identical (see
Table 2).
As discussed in R99a and R99b, the four-dimensional posterior probability density distribution
function L(Qrms−PS, Ω0, ΩBh2, t0) is nicely peaked in the Qrms−PS direction but fairly flat in
the other three directions. Marginalizing over Qrms−PS results in a three-dimensional posterior
distribution L(Ω0, ΩBh2, t0) which is steeper, but still relatively flat. As a consequence, limits
derived from the four- and three-dimensional posterior distributions are generally not highly
statistically significant. We therefore do not show contour plots of these functions here.
Marginalizing over Qrms−PS and one other parameter results in two-dimensional posterior
probability distributions which are more peaked. See Figures 3 and 4. As in the ARGO (R99a)
and combination (R99b) data set analyses, in some cases these peaks are at an edge of the
parameter range considered.
Figure 3 shows that the two-dimensional posterior distributions allow one to distinguish
between different regions of parameter space at a fairly high formal level of confidence. For
instance, the open model near Ω0 ∼ 0.75, ΩBh2 ∼ 0.03, and t0 ∼ 20 Gyr, and the flat-Λ model
near Ω0 ∼ 0.6, ΩBh2 ∼ 0.03, and t0 ∼ 20 Gyr, are both formally ruled out at ∼ 3 σ confidence.
However, we emphasize, as discussed in R99a and R99b, care must be exercised when interpreting
the discriminative power of these formal limits, since they depend sensitively on the fact that the
uniform prior has been set to zero outside the range of the parameter space we have considered.
Figure 4 shows the contours of the two-dimensional posterior distribution for Qrms−PS and
Ω0, derived by marginalizing the four-dimensional distribution over ΩBh2 and t0. These are shown
for the combined Python I, II, and III data, the DMR data, and three combinations of data from
the SP94, ARGO, MAX 4+5, White Dish, and SuZIE experiments (R99b), for both the open and
flat-Λ models. Constraints on these parameters from the combined Python data are consistent
with those from the DMR data for the flat-Λ models, panel a), while for the open model, panel b),
11 For comparison, the corresponding DMR 1 σ δTl error bar is ∼ 10 − 12% (depending on model, G´orski et al.
1998).
-- 6 --
>
∼ 0.2 (0.35). The combined Python data amplitudes are a
consistency at 2 σ (1 σ) requires Ω0
little higher than those derived from the other small-scale data combinations, panels c) − h), but
at 2 σ confidence the various amplitudes are mostly consistent.
Figure 5 shows the one-dimensional posterior distribution functions for Ω0, ΩBh2, t0, and
Qrms−PS, derived by marginalizing the four-dimensional posterior distribution over the other three
parameters. From these one-dimensional distributions, the combined I, II, and III data favors an
open (flat-Λ) model with Ω0 = 0.19 (0.10), or ΩBh2 = 0.005 (0.005), or t0 = 10 (13) Gyr, amongst
the models considered. At 2 σ confidence the combined Python data formally rule out only small
regions of parameter space. From the one-dimensional distributions of Figure 5, the data requires
Ω0 < 0.71 or > 0.8 (Ω0 < 0.55 or > 0.63), or ΩBh2 < 0.028 (ΩBh2 < 0.028), or t0 < 20 Gyr (t0
< 20 Gyr) for the open (flat-Λ) model at 2 σ. As discussed in R99a and R99b, care is needed
when interpreting the discriminative power of these formal limits. These papers also discuss a
more conservative Gaussian posterior distribution limit prescription. Using this more conservative
prescription, we find only an upper 1 σ limit on Ω0 ( <
∼ 0.5) in the open model.
While the statistical significance of the constraints on cosmological parameters is not high,
it is reassuring that the combined Python data favor low-density, low ΩBh2, young models,
consistent with some of the indications from the combinations of CMB anisotropy data considered
by R99b, and the indications from most recent non-CMB observations (see discussion in R99b).
The peak values of the one-dimensional posterior distributions shown in Figure 5 are listed
in the figure caption for the case when the four-dimensional posterior distributions are normalized
such that L(Qrms−PS = 0 µK) = 1. With this normalization, marginalizing over the remaining
parameter the fully marginalized posterior distributions are 2 × 10106(1 × 10106) for the open
(flat-Λ) model and the combined Python data. This is qualitatively consistent with the indication
from panels a) and b) of Figure 5 that the most-favored open model is somewhat more favored
than the most-favored flat-Λ one.
4. Conclusion
The combined Python I, II, and III data results derived here are mostly consistent with
those derived from the DMR, SP94, ARGO, MAX 4+5, White Dish and SuZIE data. The
combined Python data significantly constrains Qrms−PS (for the flat bandpower spectrum
Qrms−PS = 40 +12
−8 µK at 1 σ) and weakly favors low-density, low ΩBh2, young models.
We acknowledge helpful discussions with D. Alvarez, M. Dragovan, G. Griffin, J. Kovac, and
J. Ruhl. This work was partially carried out at the Infrared Processing and Analysis Center
and the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the
National Aeronautics and Space Administration. KG also acknowledges support from NASA ADP
grant NASA-1260. BR and GR acknowledge support from NSF grant EPS-9550487 with matching
-- 7 --
support from the state of Kansas and from a K∗STAR First award. GR also acknowledges support
from a PRAXIS XXI program of FCT (Portugal) grant. RS acknowledges support from NASA
AISRP grant NAG-3941 and help from Polish Scientific Committee (KBN) grant 2P03D00813.
-- 8 --
Table 1: Numerical Values for the Zero-Lag Window Function Parametersa
Python I/II
Python IIIL
Python IIIS
le−0.5
53
52
128
le
91.7
87.7
171
lm
73
72
176
le−0.5 pI(Wl)
99
98
230
1.34
1.30
0.623
aThe value of l where Wl is largest, lm, the two values of l where Wl
le = I(lWl)/I(Wl), and I(Wl) = P∞
l=2(l + 0.5)Wl/{l(l + 1)}.
e−0.5 = e−0.5Wlm, le−0.5 , the effective multipole,
-- 9 --
Table 2: Numerical Values for Qrms−PS and δTl
Bandpower Spectrum
from Likelihood Analyses Assuming a Flat
Data Seta
Qrms−PS
b Ave. Abs. Err.c Ave. Frac. Err.d
I
IIA
C(I+IIA)
I+IIA
IIBf
II
C(I+IIA)+IIB
I+II
IIIL
IIIS
III
I+II+IIIL
I+II+III
(µK)
39 56
28
39 57
27
40 57
29
40 56
29
9.3 20
0
31 44
23
34 47
26
34 46
25
41 55
32
42 56
33
41 54
33
39 51
31
40 52
32
(µK)
14
15
14
14
10
10
11
11
11
11
10
10
10
36%
38%
35%
35%
110%
34%
31%
31%
28%
27%
25%
27%
25%
b
δTl
(µK)
61 87
43
60 88
43
62 89
45
61 87
45
14 31
0
48 68
35
53 73
40
52 72
39
64 84
49
66 87
51
64 83
51
60 80
47
63 81
50
LRe
2 × 1014
9 × 1012
3 × 1031
2 × 1031
1
1 × 1012
1 × 1030
1 × 1030
3 × 1033
2 × 1029
6 × 1068
3 × 1067
1 × 10105
aC(...+...) means that the data from the two sets of observations have been coadded prior to analysis, II refers to the
combined IIA and IIB data, and III refers to the combined IIIL and IIIS data.
bFor each data set, the first of the three entries is where the posterior probability density distribution function peaks
and the vertical pair of numbers are the ±1 σ (68.3% highest posterior density) values.
cAverage absolute error on Qrms−PS in µK.
dAverage fractional error, as a fraction of the central value.
eLikelihood ratio.
f IIB does not have a 2 σ highest posterior density detection; the appropriate equal tail 2 σ upper limits are 50 µK
(Qrms−PS) and 77 µK (δTl).
-- 10 --
REFERENCES
Alvarez, D.L. 1996, Princeton University Ph.D. Thesis
Bucher, M., Goldhaber, A.S., & Turok, N. 1995, Phys. Rev. D, 52, 3314
Carroll, S.M. 1998, Phys. Rev. Lett., 81, 3067
Church, S.E., Ganga, K.M., Ade, P.A.R., Holzapfel, W.L., Mauskopf, P.D., Wilbanks, T.M., &
Lange, A.E. 1997, ApJ, 484, 523
Croft, R.A.C., Weinberg, D.H., Pettini, M., Hernquist, L., & Katz, N. 1999, ApJ, in press
de Bernardis, P., et al. 1994, ApJ, 422, L33
Dragovan, M., Ruhl, J.E., Novak, G., Platt, S.R., Crone, B., Pernic, R., & Peterson, J.B. 1994,
ApJ, 427, L67 (D94)
Ferreira, P.G., & Joyce, M. 1998, Phys. Rev. D, 58, 023503
Frieman, J.A., & Waga, I. 1998, Phys. Rev. D, 57, 4642
Ganga, K., Ratra, B., Church, S.E., Sugiyama, N., Ade, P.A.R., Holzapfel, W.L., Mauskopf, P.D.,
& Lange, A.E. 1997b, ApJ, 484, 517
Ganga, K., Ratra, B., Gundersen, J.O., & Sugiyama, N. 1997a, ApJ, 484, 7 (GRGS)
Ganga, K., Ratra, B., Lim, M.A., Sugiyama, N., & Tanaka, S.T. 1998, ApJS, 114, 165
G´orski, K.M., Ratra, B., Stompor, R., Sugiyama, N., & Banday, A.J. 1998, ApJS, 114, 1
Gott, J.R. 1982, Nature, 295, 304
Gundersen, J.O., et al. 1995, ApJ, 443, L57
Guth, A. 1981, Phys. Rev. D, 23, 347
Harrison, E.R. 1970, Phys. Rev. D, 1, 2726
Hu, W., Eisenstein, D.J., Tegmark, M., & White, M. 1999, Phys. Rev. D, 59, 023512
Huterer, D., & Turner, M.S. 1999, Phys. Rev. Lett., submitted
Kazanas, D. 1980, ApJ, 241, L59
Liddle, A.R., & Scherrer, R.J. 1999, Phys. Rev. D, 59, 023509
Lim, M.A., et al. 1996, ApJ, 469, L69
Park, C., Colley, W.N., Gott, J.R., Ratra, B., Spergel, D.N., & Sugiyama, N. 1998, ApJ, 506, 473
Peebles, P.J.E. 1999, PASP, 111, 274
Peebles, P.J.E., & Ratra, B. 1988, ApJ, 325, L17
Peebles, P.J.E., & Yu, J.T. 1970, ApJ, 162, 815
Platt, S.R., Kovac, J., Dragovan, M., Peterson, J.B., & Ruhl, J.E. 1997, ApJ, 475, L1 (P97)
-- 11 --
Ratra, B., Ganga, K., Stompor, R., Sugiyama, N., de Bernardis, P., & G´orski, K.M. 1999a, ApJ,
510, 11 (R99a)
Ratra, B., Ganga, K., Sugiyama, N., Tucker, G.S., Griffin, G.S., Nguyen, H.T., & Peterson, J.B.
1998, ApJ, 505, 8
Ratra, B., & Peebles, P.J.E. 1994, ApJ, 432, L5
Ratra, B., & Peebles, P.J.E. 1995, Phys. Rev. D, 52, 1837
Ratra, B., & Quillen, A. 1992, MNRAS, 259, 738
Ratra, B., Stompor, R., Ganga, K., Rocha, G., Sugiyama, N., & G´orski, K.M. 1999b, ApJ, 517, in
press (R99b)
Retzlaff, J., Borgani, S., Gottlober, S., Klypin, A., & Muller, V. 1998, New Astron., 3, 631
Ruhl, J. 1993, Princeton University Ph.D. Thesis
Ruhl, J.E., Dragovan, M., Novak, G., Platt, S.R., Crone, B.K., & Pernic, R.W. 1995a, Astro.
Lett. and Comm., 32, 249
Ruhl, J.E., Dragovan, M., Platt, S.R., Kovac, J., & Novak, G. 1995b, ApJ, 453, L1 (R95)
Sato, K. 1981, Phys. Lett. B, 99, 66
Starobinsky, A.A. 1998, JETP Lett., 68, 757
Stompor, R. 1994, A&A, 287, 693
Stompor, R. 1997, in Microwave Background Anisotropies, ed. F.R. Bouchet, R. Gispert, B.
Guiderdoni, & J. Tran Thanh Van (Gif-sur-Yvette: Editions Frontieres), 91
Sugiyama, N. 1995, ApJS, 100, 281
Sugiyama, N., & Sato, K. 1992, ApJ, 387, 439
Tanaka, S.T., et al. 1996, ApJ, 468, L81
Tucker, G.S., Griffin, G.S., Nguyen, H.T., & Peterson, J.B. 1993, ApJ, 419, L45
Wang, L., & Steinhardt, P.J. 1998, ApJ, 508, 483
Yamamoto, K., Sasaki, M., & Tanaka, T. 1995, ApJ, 455, 412
Zel'dovich, Ya. B. 1972, MNRAS, 160, 1P
This preprint was prepared with the AAS LATEX macros v4.0.
-- 12 --
Figure Captions
Fig. 1. -- Measured thermodynamic temperature differences (with ±1-σ error bars). Panel a) shows
Python I (open circles), IIA (filled squares, offset horizontally from true positions for clarity), IIB
(open squares), and IIIL (crosses) data, while panel b) shows Python IIIS data (crosses).
Fig. 2. -- CMB anisotropy multipole moments l(l + 1)Cl/(2π) × 1010 (solid lines, scale on left
axis, note that these are fractional anisotropy moments and thus dimensionless) as a function of
multipole l, for selected models normalized to the DMR maps (G´orski et al. 1998; Stompor 1997).
Panels a) − c) show selected flat-Λ models. The heavy lines are the Ω0 = 0.1, ΩBh2 = 0.005, and
t0 = 12 Gyr case, which is close to where the combined Python data likelihoods (marginalized over
all but one parameter at a time) are at a maximum. Panel a) shows five ΩBh2 = 0.005, t0 = 12
Gyr models with Ω0 = 0.1, 0.3, 0.5, 0.7, and 0.9 in descending order at the l ∼ 200 peaks. Panel
b) shows seven Ω0 = 0.1, t0 = 12 Gyr models with ΩBh2 = 0.029, 0.025, 0.021, 0.017, 0.013, 0.009,
and 0.005 in descending order at the l ∼ 200 peaks. Panel c) shows six Ω0 = 0.1, ΩBh2 = 0.005
models with t0 = 20, 18, 16, 14, 12, and 10 Gyr in descending order at the l ∼ 200 peaks. Panels
d) − f ) show selected open models. The heavy lines are the Ω0 = 0.2, ΩBh2 = 0.005, and t0 = 10
Gyr case, which is close to where the combined Python data likelihoods (marginalized over all but
one parameter at a time) are at a maximum. Panel d) shows five ΩBh2 = 0.005, t0 = 10 Gyr
models with Ω0 = 1, 0.8, 0.6, 0.4, and 0.2 from left to right at the peaks (the peak of the Ω0 = 0.2
model is off scale). Panel e) shows seven Ω0 = 0.2, t0 = 10 Gyr models with ΩBh2 = 0.029, 0.025,
0.021, 0.017, 0.013, 0.009, and 0.005 in descending order at l ∼ 400. Panel f ) shows six Ω0 = 0.2,
ΩBh2 = 0.005 models with t0 = 20, 18, 16, 14, 12, and 10 Gyr in descending order at l ∼ 400.
Also shown are the Python zero-lag window functions Wl (scale on right axis): I/II (long-dashed
lines), IIIL (short-dashed lines), and IIIS (dotted lines). See Table 1 for Wl-parameter values.
Fig. 3. -- Confidence contours and maxima of the combined Python data two-dimensional posterior
probability density distribution functions, as a function of the two parameters on the axes of each
panel (derived by marginalizing the four-dimensional posterior distributions over the other two
parameters). Dashed lines (crosses) show the contours (maxima) of the open case and solid lines
(solid circles) show those of the flat-Λ model. Panel a) shows the (ΩBh2, Ω0) plane, and panel b)
shows the (t0, Ω0) plane.
Fig. 4. -- Confidence contours and maxima of the two-dimensional (Qrms−PS, Ω0) posterior
probability density distribution functions. Panels a), c), e), & g) in the left column show the flat-Λ
model and panels b), d), f ), & h) in the right column show the open model. Note the different scale
on the vertical (Qrms−PS) axes of pairs of panels in each row. Heavy lines show the ±1 and ±2 σ
confidence limits and solid circles show the maxima of the two-dimensional posterior distributions
derived from the combined Python I, II, and III data. Shaded regions show the two-dimensional
posterior distribution 1 σ (denser shading) and 2 σ (less dense shading) confidence regions for the
DMR data (G´orski et al. 1998; Stompor 1997) in panels a) & b); for the SP94, ARGO, MAX 4
-- 13 --
and 5, White Dish and SuZIE data combination (R99b) in panels c) & d); for the previous data
combination excluding SuZIE (R99b) in panels e) & f ); and for the SP94Ka, MAX 4 ID, and
MAX 5 HR data combination (R99b) in panels g) & h). The DMR results are a composite of
those from analyses of the two extreme data sets: i) galactic frame with quadrupole included and
correcting for faint high-latitude galactic emission; and ii) ecliptic frame with quadrupole excluded
and no other galactic emission correction (G´orski et al. 1998). In panels c) − h) crosses show the
maxima of the appropriate non-Python data two-dimensional posterior distributions.
Fig. 5. -- One-dimensional posterior probability density distribution functions for Ω0, ΩBh2, t0,
and Qrms−PS (derived by marginalizing the four-dimensional one over the other three parameters)
in the open and flat-Λ models. These have been renormalized to unity at the peaks. Dotted
vertical lines show the confidence limits derived from these one-dimensional posterior distributions
and solid vertical lines in panels g) and h) show the ±1 and ±2 σ confidence limits derived by
projecting the combined Python I, II, and III data four-dimensional posterior distributions. The
2 σ DMR (marginalized and projected) confidence limits in panels g) and h) are a composite of
those from the two extreme DMR data sets (see caption of Figure 4). When the four-dimensional
posterior distributions are normalized such that L(Qrms−PS = 0 µK) = 1, the peak values of the
one-dimensional distributions shown in panels a) − h) are 2 × 10106, 3 × 10106, 6 × 10107, 7 × 10107,
1 × 10105, 2 × 10105, 9 × 10104, and 6 × 10104, respectively.
Figure 1a)
Figure 1b)
Figure 2
Figure 3
Figure 4
Figure 5.1
Figure 5.2
|
astro-ph/9602005 | 1 | 9602 | 1996-02-01T20:28:07 | Analysis of the Type Ia Supernova SN1994D | [
"astro-ph"
] | We present an analysis of the observed light curves and spectra of the Type Ia supernova SN1994D in the galaxy NGC 4526. The sensitivity of theoretical light curves and spectra to the underlying hydrodynamical model is discussed.
The calculations are consistent with respect to the explosion mechanism, the optical and infrared light curves, and the spectral evolution, leaving the description of the nuclear burning front and the structure of the white dwarf as the only free parameters.
The explosions are calculated using a one-dimensional Lagrangian code including a nuclear network. Subsequently, the light curves are constructed. Detailed NLTE-spectra are computed for several instants of time using the density, chemical, and luminosity structure resulting from the light curve code.
Different scenarios have been compared to the observations including detonations, deflagrations, delayed detonations, tamped detonations and Helium detonations. Only one of our delayed detonations models shows good agreement with the observations of SN1994D both for the B, V, R and I colors and the spectral evolution. The classical deflagration model W7 does not reproduce the high Si velocities as observed early on. The deflagration velocity is close to the laminar deflagration $(v =0.03 c_s)$, and the transition from the deflagration to the detonation occurs at $\rho_{tr} = 2.~10^7 g/cm^{3}$. The initial central density of the white dwarf (WD) is $2.7~10^9 g/cm^{3}$.
The distance to SN1994D is determined to $16.2 \pm 2 Mpc$. The explosion took | astro-ph | astro-ph |
ANALYSIS OF THE TYPE IA SUPERNOVA SN1994D
P. H OFLICH
Harvard University, Center for Astrophysics
60 Garden Str., Cambridge, MA 02138, USA
1. Introduction
On Mar. 7, 1994, Treffers, Filippenko and Van Dyk (1994) discovered the
Type Ia supernova (SN Ia) 1994D in the SO galaxy NGC 4526, a member
of the Virgo cluster which was was about 12 days before maximum light.
Subsequently, both the light curve in Johnson's UBVRI colors and spectra
a have been monitored by several groups (e.g. Filippenko et al. private
communication, Smith et al., 1995) making this supernova one of the best
observed until now. In addition, starting at Mar 12, IR-spectra are reported
to be taken at the Royal Greenwich Observatory showing a P-Cygni line at
about 1.0 µm which, likely, is due to He I (Meikle 1995).
It is generally accepted that SN Ia are thermonuclear explosions of
carbon-oxygen white dwarfs (WD) (Hoyle & Fowler 1960). However, details
of the scenario are still under debate. For discussions of various theoretical
aspects see e.g. Wheeler & Harkness 1990, Canal 1995, Nomoto et al. 1995,
Hoflich & Khokhlov 1995, Woosley & Weaver 1995). What we observe as
a supernova event is not the explosion itself but the light emitted from
a rapidly expanding envelope produced by the stellar explosion. As the
photosphere recedes, deeper layers of the ejecta become visible. A detailed
analysis of the light curves (LC) and spectra gives us the opportunity to
determine the density, velocity and composition structure of the ejecta.
2. Comparison of light curves and spectra
Our analysis is based on observations of SN1994D by the supernova group
at the Center for Astrophysics. For the first time, a detailed analysis of a
Type Ia supernova is presented which is consistent both with respect to the
explosion mechanism, the optical and infrared light curves, and the spectral
evolution. The only free parameters are the initial structure of the WD and
2
P. H OFLICH
Figure 1. Observed V and B LCs for SN1994D in comparison to the theoretical LCs
of the deflagration W7, the delayed detonation N21, the helium detonation HeD10, the
pulsating delayed detonation PDD3, and the envelope model DET2env4.
the description of the burning front. Our approach provides a direct link
between observational properties and both the explosion mechanism and,
maybe, the progenitor evolution.
The explosions are calculated using one-dimensional Lagrangian hy-
dro with artificial viscosity (Khokhlov, 1991) and radiation-hydro codes
(Hoflich & Khokhlov 1995) including a nuclear network. Subsequently, the
LC are constructed. Spectra are computed for several instants of time us-
ing the density, chemical, and luminosity structure resulting from the light
curve code which includes the solution of the radiation transport implic-
itly via the moment equations, expansion opacities, a detailed Monte-Carlo
γ−ray transport, and a detailed equation of state. Our NLTE code for calcu-
lating synthetic spectra solves the relativistic radiation transport equations
in comoving frame consistently with the statistical equations and ioniza-
tion due to γ radiation for the most important elements (He, C, O, Ne, Na,
Mg, Si, S, Ca, Fe). About 300,000 additional lines are included assuming
LTE-level populations and an equivalent-two-level approach for the source
functions. For more details on technical aspects and a discussion of the
relation between observational properties and the underlying model, see
Khokhlov et al. (1993), Hoflich (1995), Hoflich & Khokhlov (1995), Hoflich
et al. (1995, this volume) and references therein.
Although not presented in detail for all models available (see Nomoto
et al., 1984, Khokhlov et al. 1993, Hoflich et al. 1994, Hoflich, 1995, Hoflich
& Khokhlov, 1995), we have compared the observed optical and infrared
light curves of SN1994D for deflagration (W7, DF1), pulsating delayed
detonation (PDD1a-c,3-9), delayed detonation (N21, N32, M35-39), helium
detonations (HeD2-12), and envelope models (DET2env2-4) being a crude
representation of a merger scenario.
The pulsating delayed detonation models, helium detonation and enve-
lope models can be ruled out because of the shape of their V and B light
ANALYSIS OF THE TYPE IA SUPERNOVA SN1994D
3
Figure 2.
detonation model M36.
Some Abundances as a function of the expansion velocity for the delayed
Figure 3. Observed LCs of SN1994D in comparison to the theoretical LCs of M36.
curves (Fig.1). W7 and N32 that fit observations of several SNe Ia (Muller &
Hoflich, 1994) provide better fits to SN1994D, but they fail to reproduce the
IR light curves (Hoflich 1995) and, less critical, the post-maximum decline
in V. One may conclude that SN1994D is no 'standard' Type Ia supernova,
but it is close to standard. We find best agreement between theory and ob-
servation for a the delayed detonation model M36 with ρtr = 2.4 107g/cm3
and a deflagration velocity of 0.03 times the sound velocity (Figs. 2 & 3).
The initial central density of the WD is 2.7 109g/cm3, i.e. about 20 %
lower than in our delayed detonation models previously considered. The
lower density may be understood in terms of a higher accretion rate on
the progenitor. During the explosion, 0.6M⊙ of 56N i are produced. Similar
4
P. H OFLICH
Figure 4. Synthetic spectrum at day 15 for M36 compared to observations at Mar. 16th.
models with ρtr of 2.0 107 and 3.0 107g/cm3 and Ni productions of 0.51
and 0.67 M⊙, respectively, are not able to reproduce the observations.
From the vertical shift of the light curves, the time of explosion can be
determined to be between JD 244 9414.5 and 244 9415.5, meaning SN1994D
was discovered just about 3 to 4 days after the explosion! Our models
are consistent with no interstellar reddening. Taking the uncertainties in
our models into account, we determine the distance of NGC 4526 to be
16 ± 2 M pc. The explosion took place between March 3rd and 4th, 1994.
The NLTE-analysis of the spectra confirmed the results from the LC
analysis and give an insight to several new aspects. Taking into account
that we do not allow for any artificial adjustment to provide better fits,
M36 reproduces the observed spectra reasonably well (Fig. 4), including the
evolution with time (Hoflich, 1995). The agreement of the velocity shifts
both of the elements of partial burning such as Si, S, O and also the iron
rich elements indicate that the chemical and density profiles of M36 seem
to resemble SN1994D. In the spectrum from March 11, the Doppler shift
of the absorption minimum of Si II indicates photospheric expansion ve-
locities of ≈ 15000km/sec with wings reaching out to more than 22000
km/sec. This strongly suggest the presence of a significant amount of Si
at high velocities. Our delayed detonation model M36 provide a Si mass
fraction of 15 % up to the outer layers ( ≈ 25000km/sec, Fig. 1). For com-
parison, the classical deflagration model W7 does not show Si at velocities
ANALYSIS OF THE TYPE IA SUPERNOVA SN1994D
5
≤ 16400km/sec because the smaller flame speed in the outer layers and,
consequently, the larger pre-expansion of the WD envelope during the ex-
plosion. In our previous analysis, we had to report the serious problem that
the Ti features are too strong and, artificially, we had to reduce Ti in the
outer layers. In recent calculations for delayed detonation models (Hoflich
et al. 1995), we find that the Ti problem vanishes if we use a larger nuclear
network and start with a metallicity according to a population II star.
Finally, we want to mention some of the limitations of our study. Despite
similarities in both the spectra and light curves, the photospheric expansion
velocities inferred from observed spectra show strong individual variations
(e.g. Branch 1987 and Barbon et al. 1990). Different models are needed to
fit the light curves (Muller & Hoflich 1994, Hoflich et al. 1994). Therefore,
the results of our study must not be generalized. Although we treat several
elements in full NLTE, detailed atomic models shall be implemented for the
iron group elements, namely Ti, Co, Ni. Our approach to the line scattering
by using an equivalent level approach for the 'LTE-lines' seems to work, but
it must be regarded only as a first step towards a more consistent treatment.
Finally, M36 produces only 0.003 M⊙ of He in the region of α rich freeze-out
whereas, a much larger amount is needed to produce a strong He feature at
1.µ. For a more detailed discussion of the latter problem, see Meikle 1995.
3. References
Barbon R.,Bennetti S.,Cappalaro E.,Rosino L.,Turatto M. (1990), A&A, 237, 79
Branch D. (1987), ApJ, 316, L81
Canal R. (1995),Les Houches Lectures, eds. J. Audouze et al., Elsevier, in press
Hoflich P. (1995), ApJ , 443, 533
Hoflich P., Khokhlov A. (1995), ApJ , in press
Hoflich P., Khokhlov A., Wheeler C. (1994), ApJ , 444, 831
Hoflich P., Khokhlov A., Nomoto K., Thielemann F.K, Wheeler C.J. (1995), this
volume
Hoyle P., Fowler , (1960) ApJ , 132, 565
Khokhlov A., Muller E., Hoflich P. (1993), A&A, 270, 23
Khokhlov A., (1991), A&A, 245, 114
Meikle P. (1995), this volume
Muller E., Hoflich P., (1994), A&A, 281, 51
Nomoto K., Yamaoka H., Shigeyama T., Iwamoto K., (1995), in: Supernovae, ed.
R.A. McCray, Cambridge University Press, in press
Nomoto K., Thielemann F.-K., Yokoi K. (1984), ApJ, 286, 644
Smith et al., (1994), ApJ, in preparation combined observations of CTIO and CfA
Treffers, Filippenko, Van Dyke (1994), IAU-circular 5946
Wheeler J. C., Harkness R .P., (1990), Rep. Prog. Phys., 531467
Woosley S. E., Weaver T. A. (1995),Les Houches Lectures, eds. J. Audouze et al.,
Elsevier, in press
|
astro-ph/0511124 | 1 | 0511 | 2005-11-04T09:52:00 | Comments on Long-Wavelength Backreaction and Dark Energy | [
"astro-ph",
"gr-qc",
"hep-ph",
"hep-th"
] | In this brief note we comment on a recent attempt by Martineau and Brandenberger (astro-ph/0510523) to explain the acceleration of the Universe using the back-reaction of long-wavelength perturbations associated with isocurvature perturbation modes. | astro-ph | astro-ph | Comments on Long-Wavelength Backreaction and Dark Energy
DFPD-05/A/21
Edward W. Kolb∗
Particle Astrophysics Center, Fermi National Accelerator Laboratory,
Batavia, Illinois 60510-0500, USA
and Department of Astronomy and Astrophysics, Enrico Fermi Institute,
University of Chicago, Chicago, Illinois 60637-1433 USA
Sabino Matarrese†
Dipartimento di Fisica "G. Galilei" Universit`a di Padova,
INFN Sezione di Padova, via Marzolo 8, Padova I-35131, Italy
Antonio Riotto‡
CERN, Theory Division, CH-1211 Geneva 23, Switzerland
(Dated: October 20, 2019)
Abstract
In this brief note we comment on a recent attempt by Martineau and Brandenberger [1] to
explain the acceleration of the Universe using the back-reaction of long-wavelength perturbations
associated with isocurvature perturbation modes.
PACS numbers: 98.80.Cq, 95.35.+d, 4.62.+v
5
0
0
2
v
o
N
4
1
v
4
2
1
1
1
5
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
∗Electronic address: [email protected]
†Electronic address: [email protected]
‡Electronic address: [email protected]
1
In this brief note we would like to comment on a work by Martineau and Brandenberger
(MB) [1], who have recently presented a solution to the dark energy problem making use of
the gravitational backreaction of long wavelength (super-Hubble) fluctuation modes on the
background metric.
The issue of the physical influence of the super-Hubble modes on cosmological observables,
such as the local Hubble expansion rate, has already been posed in the literature [2, 3, 4,
5, 6, 7]. Sometime ago, some of us [8] have shown that if adiabaticity holds, then at any
order in perturbation theory super-Hubble perturbations do not have any impact on local
physical observables (if gradients are neglected). Let us briefly summarize here the line of
reasoning. Our starting point is the Arnowitt-Deser-Misner (ADM) formalism, with metric
ds2 = −N 2 dt2 + Ni dt dxi + γij dxi dxj.
(1)
The three-metric γij, and the lapse and the shift functions N and Ni, describe the evo-
lution of the spacelike hypersurfaces. In the ADM formalism the equations simplify consid-
erably if we set N i = 0. We may then express the spatial metric in the form
γij = exph2α(t, xi)i hij(xi),
(2)
where the conformal factor exp[α(t, xi)] may be interpreted as the spatially dependent scale
factor. The time-independent three-metric, hij(xi), (with unit determinant) describes the
three geometry of the conformally transformed space. Within linear perturbation theory,
α(t, xi) would contribute only to scalar perturbations, whereas hij contains another scalar,
as well as vector and tensor perturbations. Vector and tensor perturbations are necessary
to satisfy Einstein equations beyond linear order, however in the long wavelength limit they
do not affect the equations for the scalar perturbations. Since exp[α(t, xi)] is interpreted as
a scale factor, we can define the local Hubble parameter H(t, xi) = α(t, xi)/N(t, xi).
Now consider a Universe filled with ideal fluid(s) described by an energy-momentum
tensor of the form Tµν = (ρ + P ) uµuν + P gµν, where ρ and P are the energy density and
pressure, respectively. The four-velocity vector can be chosen locally to be uµ = (1,~0). In
the case of a single fluid, this amounts to saying that a volume measured by a local observer
is comoving with the energy flow of the fluid. In the multi-fluid case, the volume comoving
with the total energy density ρ is not the same as the individual volumes comoving with the
individual energy density components. However, one can show that on super-Hubble scales
2
all comoving volumes become equivalent, and therefore our choice of the four velocity is well
justified.
We consider the effect on the local expansion rate of the Universe and set
exphα(t, xi)i = a(t) exph−ψ(t, xi)i
(3)
and choose to work in the synchronous gauge for which N = 1 (together with N i = 0). In
this gauge the field equations look just like the familiar homogeneous Friedmann-Robertson-
Walker equations point by point. As mentioned, for a generic set of fluids we may safely take
the four velocity to be uµ = (1,~0) on super-Hubble scales, from which we define the local
expansion rate of tangential surfaces orthogonal to the fluid flow, i.e., the local expansion
rate, to be Θ ≡ Dµuµ. In the synchronous gauge this quantity coincides with 3H.
H(xi, t) = 1
Using Eq. (3), we find that at any order in perturbation theory the local expansion rate is
3Θ(xi, t) = a/a − ψ. We infer that the local physical expansion rate is influenced
by long-wavelength perturbations only if the gravitational potential is time dependent (if we
ignore spatial gradients in ψ). Let us analyze the case of an adiabatic fluid. From Einstein's
equations for a globally flat space we deduce (on super-Hubble scales) that locally
3H 2 = 8πGρ = −8πG P + 2 H.
(4)
If the pressure P is a unique function of the energy density, we may expand ρ = ρ0 + ∆ρ,
where ρ0 is the energy density entering the homogeneous Einstein equations from which
the homogeneous scale factor a(t) is computed. Inserting this expansion into Eq. (4) and
eliminating the pressure in favor of the energy density ρ ∝ H 2, we find a differential equation
of the generic form G(cid:16) ψ, ψ(cid:17) = 0, which does not contain any term proportional to ψ.
Therefore, ψ = ψ(xi) is a solution of Einstein equations at any order in perturbation theory
and ψ = 0 holds on super-Hubble scales. From this generic argument we conclude that there
is no influence of very long wavelength modes on the physical local expansion rate of the
Universe if adiabaticity holds.
In the adiabatic case, the influence of infrared modes is not locally measurable. There is
a simple explanation of this result. When adiabaticity holds, the pressure is a well defined
function of the energy density. This means that the Hubble rate is only a function of the
unique available physical clock, the energy density. Indeed, suppose that the local expansion
3
rate is H(t, xi) = H(ρ(t, xi), t). This leads to
∂H
∂t !xi
∂t!xi ∂H
∂ρ !t
∂t!xi ∂H
+ ∂t
∂t !ρ
= ∂ρ
= −4πG(ρ + P ) + ∂H
∂t !ρ
,
(5)
where we have made use of the equation 3H = − ρ/(ρ + P ). Comparing Eq. (5) with
H = −4πG(ρ + P ), we see that (∂H/∂t)ρ = 0, and hence H(t, xi) = H(ρ(t, xi)). The
dependence of the local expansion rate of the Universe on the clock time takes the same
form as in the unperturbed Universe when evaluated at a fixed value of the only clock
available, the energy density ρ. Infrared modes do not have any locally measurable effect
on the expansion rate of the Universe. This result applies, in particular, during inflation if
the energy density is dominated by a single inflaton field since in such a case adiabaticity
is guaranteed. One implication of this result is that long-wavelength modes may not be
responsible for stopping inflation.
This conclusion does not hold when adiabaticity is not attained. Indeed, the right-hand
side of Eq. (5) would contain an extra factor
∂t !xi ∂H
∂P !t
∂P
,
(6)
which makes it impossible for the local expansion rate to be a unique function of the energy
density. The same conclusion was reached using a second-order approach [4, 6, 7].
MB have very recently argued that one can make use of the fluctuations of a scalar field
ϕ and so long as these fluctuations are associated with an isocurvature mode, the effect on
local observables should be measurable, so that at late times the backreaction could mimic
dark energy. This conclusion is reached by inspecting the second-order effective energy
momentum tensor. Their approach is not fully consistent as the metric fluctuations and the
scalar field fluctuations are not properly expanded at second-order to include intrinsically
second-order quantities. However, for the sake of comparison, we will perform the same
approximation.
Following Ref. [1], let us consider a scalar field ϕ with potential V (ϕ) = λϕ4 . As
it was shown in Ref. [9], the equation of state of the zero mode ϕ0 is that of radiation,
implying ϕ0 ∝ a−1, a being the background scale factor. The scalar field may be split
as ϕ = ϕ0 + δϕ and one may consider the long-wavelength limit of the spatially averaged
4
energy density of the scalar field. MB argue that it is dominated at large times by a piece
δρϕ = 1
2V ′′hδϕ2i = −δPϕ. It dominates over the matter contribution, thus leading to an
effective cosmological constant behavior. The way this conclusion is reached is by making
use of Einstein equations which relate the scalar fluctuations to the metric fluctuations
( H + 3H 2)Φ = −4πG V ′δϕ +
δρm
2 ! ,
(7)
where Φ is the metric (scalar) perturbation in the longitudinal gauge, and dropping the piece
proportional to δρm. Since during a matter-dominated epoch H 2 ∼ a−3 ∼ V ′ and assuming
that the gravitational potential Φ is constant, MB conclude that δϕ does not evolve with
time, implying that δρϕ ∼ −δPϕ ∼ hΦ2i/a2. This term will eventually dominate over matter
thus leading to acceleration.
This result is, however suspicious because MB employ the "adiabatic" part of the fluctu-
ation of the scalar field, that is the part of the fluctuation proportional to the gravitational
potential. Let us work in the longitudinal gauge with metric ds2 = −(1 + 2Φ)dt2 + a2(t)(1 −
2Φ)δijdxidxj. As we stressed above, one needs a purely isocurvature component to hope
that super-Hubble modes have any influence on cosmological observables. The first-order
Klein-Gordon equation for the fluctuation of the scalar field reads
δ ϕ + 3Hδ ϕ + V ′′δϕ = 4 ϕ0 Φ − 2V ′ϕ0Φ .
The exact solution of this equation is (we assume a constant Φ as in Ref. [1])
δϕ = q
δϕ∗
ϕ0∗
ϕ0 + t ϕ0Φ ,
(8)
(9)
where the label "∗" indicates that the initial conditions have been fixed at some time t∗.
The first piece represents the pure isocurvature component of the fluctuation, while the
piece proportional to the gravitational potential is the adiabatic component. Its origin can
be understood via the separate Universe approach: when the fluctuation starts having some
nontrivial evolution after inflation, initial conditions are different in different Hubble patches
if adiabatic perturbations set by the gravitational potential Φ are present. The parameter
q ∼ 1 depends upon the polynomial form of the potential. From Eq. (9) we see that
δϕ ∼ ϕ0 ∼ a−1 throughout the all evolution of the Universe. This means that δρϕ ∼ a−4
during all epochs and there is no way the long-wavelength fluctuations of the scalar field
can dominate over nonrelativistic matter.
5
The reason why MB reach a different conclusion is that they solve Eq. (7) to infer the time
behavior of the scalar fluctuation. However, the piece proportional to δρm is dropped. This is
somewhat inconsistent since there are extra components in that equation to account for the
other fluids (and indeed more than one fluid is necessary to have isocurvature perturbation
without which no super-Hubble influence on observables is possible). On the contrary,
solving directly the Klein-Gordon equation shows that super-Hubble fluctuations behave
like a relativistic gas as the zero mode.
To understand this result in another way, we consider a nonlinear generalization of the
metric in the longitudinal gauge, ds2 = −e2Φdt2 + e−2Ψa2(t)δijdxidxj. Again, following
Ref. [1], we take the two gravitational potentials equal, Φ = Ψ (even though this not fully
consistent), and constant in time (there is a mild logarithmic dependence upon time at
second-order, which is, however, not relevant in the considerations of Ref.
[1]). As long
as we are interested in long-wavelength perturbations, we can recast this perturbed metric
in a FRW form by putting dt = eΦdt and a = e−Φa. The Klein-Gordon equation for the
quantum scalar field containing the zero mode and the super-Hubble perturbations reads
ϕ′′ + 3Hϕ′ +
dV
dϕ
= 0,
(10)
where the primes indicate now differentiation with respect to t and H = (da/dt) =
e−2Φ(H − Φ). Eq. (10) can then be rewritten as
ϕ + 3H ϕ + e2Φ dV
dϕ
= 4 ϕ Φ ,
(11)
which expanded to first-order gives exactly the perturbed Klein-Gordon equation, Eq. (8).
For a constant gravitational potential, one concludes that the scalar field (including the
long-wavelength modes) behaves exactly like the zero mode (upon rescaling: λ → e2Φλ).
Therefore, the energy density associated with the long-wavelength "adiabatic" scalar per-
turbations scales like a relativistic gas (as the zero mode) and may never dominate over
matter. Since super-Hubble perturbations may have an effect on observables only if they
are associated with isocurvature perturbations, we conclude that the effect, if any, must
depend upon the isocurvature component of δϕ, i.e., on δϕ∗/ϕ0∗. Indeed, it is easy to see
that the isocurvature perturbation Smϕ = 3 (ζm − ζϕ), where the ζm and ζϕ are the comov-
ing curvature perturbations for matter and the scalar field respectively, is proportional to
δϕ∗/ϕ0∗. However, isocurvature perturbations induce a sizable effect only when the fluids
6
have comparable energy densities, ζ ∼ H ( ρϕ ρm/ ρ2) Smϕ, and it is not clear to us if it suffices
to have a super-Hubble isocurvature component to mimic dark energy.
Acknowledgments
We thank D. H. Chung for useful correspondence. A.R. is on leave of absence from
INFN, Padova (Italy). E.W.K. is supported in part by NASA grant NAG5-10842 and by
the Department of Energy. S.M. acknowledges partial financial support by INAF.
[1] P. Martineau and R. Brandenberger, arXiv:astro-ph/0510523.
[2] V. F. Mukhanov, L. R. W. Abramo and R. H. Brandenberger, Phys. Rev. Lett. 78, 1624 (1997).
[3] W. Unruh, astro-ph/9802323.
[4] G. Geshnizjani and R. H. Brandenberger, Phys. Rev. D 66, 123507 (2002).
[5] F. Finelli, G. Marozzi, G. P. Vacca and G. Venturi, Phys. Rev. D 69, 123508 (2004).
[6] G. Geshnizjani and R. Brandenberger, JCAP 0504, 006 (2005).
[7] R. H. Brandenberger and C. S. Lam, hep-th/0407048.
[8] E. W. Kolb, S. Matarrese, A. Notari and A. Riotto, Mod. Phys. Lett. A 20, 2705
(2005) (e-print arXiv:astro-ph/0410541) See also, E. W. Kolb, S. Matarrese and A. Riotto,
arXiv:astro-ph/0506534, section IIIA.
[9] M. S. Turner, Phys. Rev. D 28, 1243 (1983); Y. Shtanov, J. H. Traschen and R. Brandenberger,
Phys. Rev. D 51, 5438 (1995).
7
|
astro-ph/0010338 | 3 | 0010 | 2001-01-18T14:31:42 | Upper Limits On Periodic, Pulsed Radio Emission from the X-Ray Point Source in Cassiopeia A | [
"astro-ph"
] | The Chandra X-ray Observatory recently discovered an X-ray point source near the center of Cassiopeia A, the youngest known Galactic supernova remnant. We have conducted a sensitive search for radio pulsations from this source with the Very Large Array, taking advantage of the high angular resolution of the array to resolve out the emission from the remnant itself. No convincing signatures of a dispersed, periodic source or of isolated dispersed pulses were found, whether for an isolated or a binary source. We derive upper limits of 30 and 1.3 mJy at 327 and 1435 MHz for the phase-averaged pulsed flux density from this source. The corresponding luminosity limits are lower than those for any pulsar with age less than 10^4 years. The sensitivities of our search to single pulses were 25 and 1.0 Jy at 327 and 1435 MHz. For comparison, the Crab pulsar emits roughly 80 pulses per minute with flux densities greater than 100 Jy at 327 MHz and 8 pulses per minute with flux densities greater than 50 Jy at 1435 MHz. These limits are consistent with the suggestion that the X-ray point source in Cas A adds to the growing number of neutron stars which are not radio pulsars. | astro-ph | astro-ph | Draft version October 26, 2018
Preprint typeset using LATEX style emulateapj v. 04/03/99
1
0
0
2
n
a
J
8
1
3
v
8
3
3
0
1
0
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
UPPER LIMITS ON PERIODIC, PULSED RADIO EMISSION FROM THE X-RAY POINT
SOURCE IN CASSIOPEIA A
M. A. McLaughlin1, J. M. Cordes1, A. A. Deshpande2
,
3, B. M. Gaensler4
,
5,
T. H. Hankins2, V. M. Kaspi6
,
7, & J. S. Kern2
Draft version October 26, 2018
ABSTRACT
The Chandra X-ray Observatory recently discovered an X-ray point source near the center of Cassiopeia
A, the youngest known Galactic supernova remnant. We have conducted a sensitive search for radio
pulsations from this source with the Very Large Array, taking advantage of the high angular resolution
of the array to resolve out the emission from the remnant itself. No convincing signatures of a dispersed,
periodic source or of isolated dispersed pulses were found, whether for an isolated or a binary source.
We derive upper limits of 30 and 1.3 mJy at 327 and 1435 MHz for its phase-averaged pulsed flux
density. The corresponding luminosity limits are lower than those for any known radio pulsar with age
less than 104 years. Our search sensitivities to single pulses were 25 and 1.0 Jy at 327 and 1435 MHz.
For comparison, the Crab pulsar emits roughly 80 pulses per minute with flux densities greater than 100
Jy at 327 MHz and 8 pulses per minute with flux densities greater than 50 Jy at 1435 MHz. These limits
suggest that Cas A belongs to the growing population of young neutron stars that are radio quiet
1.
INTRODUCTION
Cassiopeia A, with an age of only 320 years,
is the
youngest known Galactic supernova remnant. The super-
nova that gave rise to this remnant was probably discov-
ered in 1680, when Flamsteed (1725) observed a 6th mag-
nitude star within 13′ of Cas A's location (Ashworth 1980).
Because its abundances of heavy elements are consistent
with those expected from the explosion of a massive star
(Hughes et al. 2000), Cas A is thought to be the remnant
of a Type II supernova. Given the stellar initial mass func-
tion, a large fraction of Type II supernovae are expected to
leave behind neutron stars. While many searches at vari-
ous wavelengths for the compact remnant in Cas A have
been undertaken, only upper limits had been established
until the recent Chandra discovery of an X-ray point source
(hereafter XPS) near the center of the remnant (Tanan-
baum 1999). This XPS was later confirmed in archival
ROSAT (Aschenbach 1999) and Einstein (Pavlov & Za-
vlin 1999) images. The XPS is within 5′′ of the expansion
center of the remnant (van den Bergh & Kamper 1983).
For a remnant distance of 3.4+0.3
−0.1 kpc, calculated through
radial velocity, proper motion, and age data (Reed et al.
1995), this offset corresponds to a maximum transverse ve-
locity of roughly 200 km s−1, a typical velocity for young
pulsars (Cordes & Chernoff 1998).
Pavlov et al. (2000) find that the X-ray spectrum of the
XPS is fit well by a power law with index 2.6 − 4.1, with
luminosity (2−60)×1034 ergs s−1. Similarly, Chakrabarty
et al. (2001) estimate a power law index of 2.8 − 3.6 and
luminosity of (2 − 16) × 1034 ergs s−1. These indices are
much steeper than the X-ray spectra of Active Galactic
Nuclei (AGN). Furthermore, given the measured surface
density of AGN, the probability of finding an AGN that
close to the center of the SNR is negligible. For these
reasons, Pavlov et al.
(2000) conclude that the XPS is
almost certainly associated with the Cas A remnant. The
fitted power law indices for the XPS are also much steeper
than those measured for the 6 young X-ray pulsars hav-
ing ages less than 104 years, whose spectral indices range
from 1.1 − 1.7 (Chakrabarty et al. 2001).
In addition,
the estimated XPS luminosities are only marginally consis-
tent with measured young pulsar X-ray luminosities, which
range from 3 × 1035 to 4 × 1036 ergs s−1. It is also signif-
icant that, unlike 5 of the 6 young pulsars with ages less
than 104 years, Cas A shows no evidence for a synchrotron
nebula surrounding the point source.
While there is much evidence for and against the neu-
tron star nature of the XPS, the definitive proof would
be the detection of pulsations. No X-ray pulsations have
been found in data taken with the Chandra High Resolu-
tion Camera. However, a time-tagging problem caused the
timing accuracy to be degraded from 16 µs to ≈ 4 ms, al-
lowing searches for only broad pulses with periods greater
than 20 ms (Chakrabarty et al. 2001). The detection of
radio pulsations, however, would be just as conclusive. For
this reason, we have conducted a search for radio pulsa-
tions from the XPS using the Very Large Array (VLA) at
frequencies of 327 and 1435 MHz. There have been several
previous searches for a radio pulsar in Cas A (Davies &
Large 1970, Seiradakis & Graham 1980, Woan & Duffett-
Smith 1993, Lorimer et al. 1998). Our VLA search is
many times more sensitive, as having an accurate position
for the XPS allows us to resolve out the strong emission
1Astronomy Department, Cornell University, Ithaca, NY 14853
2Physics Department, New Mexico Institute of Mining and Technology, Socorro, NM 87801
3Raman Research Institute, Bangalore 560080, India
4Hubble Fellow
5Center for Space Research, Massachusetts Institute of Technology, Cambridge, MA 02139
6Physics Department, McGill University, Montreal, Canada H3A2T8
7on leave from Physics Department and Center for Space Research, Massachusetts Institute of Technology, Cambridge, MA 02139
1
2
from the remnant, which with 327- and 1435-MHz flux
densities of 6400 and 2000 Jy is one of the strongest radio
sources known.
2. DATA
We observed the XPS on three days (23 October
1999, 31 October 1999, and 28 November 1999) with
the phased-array VLA in B array, where the maximum
baseline is 11.4 km. The J2000 coordinates of the XPS
are RA=23h23m27s.94, DEC=+58o48′42′′, with positional
uncertainty of 1′′ (Tananbaum 1999). As the 327- and
1435-MHz synthesized beamwidths are approximately 20
and 4 arcseconds respectively, one beam at each frequency
was sufficient for this search. During each session, the XPS
was observed at center frequencies of 327 and 1435 MHz,
resulting in three pointings of length 20, 40, and 50 min-
utes at 327 MHz, and four pointings of length 40, 40, 45,
and 50 minutes at 1435 MHz. During each observing ses-
sion, we observed the known pulsar PSR B0355+54, which
has a period of 154 ms and a dispersion measure (DM) of
57 pc cm−3. A blank-sky pointing and the flux and phase
calibrator J2350+646 (with flux densities of 27 and 5 Jy
at 327 and 1435 MHz) were also observed.
We used the VLA's High Time Resolution Processor
(Moffett 1997) to record dual circular polarization data at
a sampling frequency of 9.22 kHz. Each scan was started
on a 10-second tick, tied to the VLA's hydrogen maser,
compared to Universal Time through the GPS. At 327
MHz, 14 × 0.125-MHz frequency channels covered a total
bandwidth of 1.75 MHz. These channels were spaced non-
contiguously to avoid known interference frequencies and
covered a total bandwidth range of 3 MHz. At 1435 MHz,
14×4-MHz channels covered a total contiguous bandwidth
of 48 MHz. (As only 50 MHz of bandwidth is available,
some channels were duplicated). The gain in each of the
28 channels (2 polarizations and 14 frequencies) was cal-
culated from observations of the calibrator source.
3. ANALYSIS
For each of the seven XPS data sets, we first summed
right and left circular polarizations and then dedispersed
the data over a range of trial DMs. The Taylor and Cordes
(1993) model predicts a maximum DM of approximately
75 pc cm−3 in the direction of the XPS. We dedispersed
the data with 60 trial DMs from 0 to 300 pc cm−3 to allow
for an underestimate or a contribution from Cas A. These
60 DMs were spaced nonuniformly, with greater spacing
at higher DMs, where the temporal smearing due to dis-
persion across an individual frequency channel is larger.
Each dedispersed time series was Fourier transformed with
32 Mpoint and/or 16 Mpoint Fast Fourier Transforms
(FFTs), depending on the length of the data set. Shorter
FFTs of 8 and 4 Mpoints were done to increase our sensi-
tivity to some binary pulsars. The resulting power spectra
were harmonically summed (up to a maximum of 16 har-
monics) to increase our sensitivity to narrow pulses (Lyne
& Smith 1998). All features in the resulting power spec-
tra with signal-to-noise above a threshold of seven were
saved as pulsar candidates. For each candidate, the corre-
sponding dedispersed time series was folded at the period
corresponding to the fundamental harmonic. These pro-
files were inspected and candidate lists from different days
and at different frequencies were compared. While many
periodic signals were detected, they were predominantly
at low DM, had noiselike pulse profiles, and did not ap-
pear on multiple epochs or at multiple frequencies, and
were therefore attributed to radio frequency interference
(RFI). Figure 1 shows the folded pulse profiles resulting
from applying the search algorithm to PSR B0355+54.
Fig. 1. -- Folded pulse profiles for PSR B0355+54, which has a
period of 156 ms and DM of 57 pc cm−3, at 327 MHz (left) and
1435 MHz (right). Note that these profiles are for 5 and 2 minutes
of data at 327 and 1435 MHz, whereas our longest XPS search scans
were 50 minutes at these same frequencies.
An independent Fourier analysis of the same data, car-
ried out at the New Mexico Institute of Mining and Tech-
nology, was also unsuccessful. This analysis was sensitive
to periods ≥10 ms and covered a smaller DM range, but
with a finer DM spacing. Three filter-bank channels with
high levels of RFI were rejected, decreasing the number of
spurious candidates significantly with only a 12% sensitiv-
ity loss. Power spectra were harmonically summed up to a
maximum of 32 harmonics. Final pulsar candidates were
evaluated by comparing average profiles from two halves of
the bandpass and two halves of profiles folded over a two
period span. These were compared with the corresponding
profiles for non-dedispersed data.
The Crab pulsar, currently the youngest known radio
pulsar, was initially detected through its 'giant' pulses,
not through its time-averaged pulsed emission (Staelin &
Reifenstein 1968). Giant pulses are individual pulses with
amplitudes much greater than the mean pulse amplitude
(Lundgren et al. 1995). As we might expect other young
pulsars to behave similarly to the Crab, we also searched
for aperiodic, dispersed pulses from the XPS. We recorded
all pulses with amplitudes above a signal-to-noise thresh-
old of 4 for each dedispersed time series, enchancing our
sensitivity to broadened pulses by repeating this thresh-
olding with different levels of time-series smoothing. Given
our 4σ threshold, we were sensitive to all single pulses with
flux densities greater than 25/√wms Jy, where wms is the
pulse width in milliseconds, at 327 MHz and 1.0/√wms
Jy at 1435 MHz. Our smoothing algorithm optimized the
search for pulse widths up to 15 ms.
(For comparison,
the widths of the Crab pulsar's giant pulses range from
roughly 0.2 to 0.4 ms.) We found no evidence for isolated
pulses from the XPS. The result of this analysis, along
with the result for PSR B0355+54, is shown in Figure 2.
For comparison, the Crab pulsar emits roughly 80 pulses
per minute with flux densities greater than 100 Jy at 327
MHz and 8 pulses per minute with flux densities greater
than 50 Jy at 1435 MHz.
3
Figure 3 shows the dependence of our search sensitiv-
ity on DM and period at the two observing frequencies,
accounting for the effects of scattering and dispersion on
the intrinsic duty cycle of the pulsar. Assuming values for
period and DM of 30 ms and 75 pc cm−3, we calculate
7σ upper limits of 30 and 1.3 mJy at 327 and 1435 MHz.
As shown in Figure 3, these limits will decrease for longer
period, smaller duty cycle, and lower DM. There are no
hardware limitations to our maximum detectable period,
determined solely by the length of the FFT.
While there is no infrared or optical evidence for a bi-
nary companion (van den Bergh & Pritchet 1986), accel-
eration effects from such a companion would cause pulse
smearing, decreasing our sensitivity. We estimate the
signal-to-noise degradation introduced for various orbital
systems by taking a Gaussian pulse and applying orbital
smearing by iteratively calculating emission times across
the orbit. Smearing times due to time constant, DM,
and scattering are calculated and applied to the orbital-
smeared pulse. We calculate the best harmonic sum for
the resultant waveform and compare this to the best har-
monic sum for a waveform with no orbital smearing to
calculate the loss in signal-to-noise. For a 30 ms period
and duty cycle of 0.2, a 1 solar mass companion in an 8
hour circular orbit with an inclination angle (i.e. angle
between the orbital plane and our line of sight) of 30◦, our
sensitivity decreases by roughly a factor of 2, averaging
over all orbital phases, for a 16 Mpoint FFT. In the case
of our 8 and 4 Mpoint FFTs, these same binary param-
eters would decrease our search sensitivity by factors of
2.4 and 2.8 over the original 16 Mpoint sensitivity to an
unaccelerated pulsar. Although the nominal sensitivities
are less for these shorter FFT lengths, the shorter FFTs
would allow us to detect pulsars in fast binaries where the
binary smearing during a longer FFT is comparable to or
greater than the pulse period. The sensitivity would be
further decreased for smaller spin periods, faster orbits,
and more massive companions. If the orbit was eccentric,
the net smearing, averaged over all orbital phases, would
be less, and the resultant sensitivity would be higher. Ac-
cretion from a binary companion could completely quench
the radio emission.
Fig. 3. -- Contours of pulsed search 7σ sensitivity vs. P and DM
for searches at both 327 and 1435 MHz. The thickest (rightmost)
line corresponds to a sensitivity of 16 mJy at 327 MHz and of 0.8
mJy at 1435 MHz. Minimum detectable flux densities increase by
a factor of 2 with eac h subsequently thinner line. The contours are
not smooth because of the finite spacing of our trial DMs.
Interstellar scintillation modulations of the flux of
the XPS are mostly negligible during our observations.
Fig. 2. -- The number of isolated, dispersed pulses at 327 MHz
above a 4σ threshold is plotted vs. DM for a 5-minute observation
of PSR B0355+54 (left) and a 50-minute observation of the XPS
(right). While a peak at the 57 pc cm−3 DM of PSR B0355+54 is
obvious, there is no evidence for such an excess from the XPS. The
peak at low DM in the right plot is consistent with interference, and
is similar to the number of counts observed in the consecutive blank
sky pointing. I n the absence of radio frequency interference and/or
a pulsar, we would expect a flat distribution.
4. RESULTS
To calculate an upper limit for pulsed radio emission
from the XPS, we must account for the excess system tem-
perature due to the emission from Cas A itself. The Cas A
remnant is 5′ in size (Green 2000). Hence, the ratios of
our beam area to remnant area are 0.003 and 0.0002 at
327 and 1435 MHz. Given Cas A's measured flux density
and spectral index (Green 2000), we calculate remnant
flux densities of 6432 and 2060 Janskys at 327 and 1435
MHz, contributing to an increase in system temperature
of a modest 40 K at 327 MHz and a mere 1 K at 1435 MHz
for the phased array. However, while the phased array re-
solves out emission from the nebula, each single dish will
see the whole remnant, leading to an increase in system
temperature of 475 K at 327 MHz and 230 K at 1435 MHz.
By comparison, a single dish with the same collecting area
as the VLA would suffer increases of system temperature
of 13000 and 6200 K at 327 and 1435 MHz.
We calculate the minimum detectable flux density of our
search using the expression (Cordes & Chernoff 1997)
Smin =
ηTsys
GpNpol∆ν∆tNFFT
√Nh
PNh
l=1 Rl
,
(1)
where η is the signal-to-noise threshold used, Tsys is the
system temperature, G is the telescope gain, Npol is the
number of polarization channels, ∆ν is the total band-
width, ∆t is the sample interval, NFFT is the number of
points in the FFT, Nh is the optimal number of harmon-
ics with which a pulsar can be detected, and Rl ≤ 1 is the
ratio of the lth harmonic to the fundamental. Nh will de-
pend on the pulsar's duty cycle, or ratio of pulse width to
period, with narrower pulses producing more harmonics.
For this calculation, we take the intrinsic duty cycle to be
the minimum of 0.3 and 0.03(Pms/1000.)−0.5, where Pms is
the pulse period in milliseconds (Biggs 1990). Pulses may
then be additionally broadened by propagation effects, in-
cluding scattering and dispersion. The Taylor & Cordes
(1993) model predicts 1.7 ms and 3 µs of pulse broadening
due to scattering in the direction of the XPS and for a
distance of 3.4 kpc at 327 and 1435 MHz, respectively.
XPS, gamma-ray searches with future, more sensitive in-
struments, could be successful. Further measurements of
the spectrum and time variability of the XPS will also help
determine which of the above scenarios is the case.
This search adds to the growing body of evidence that
there are manifestations of young neutron stars other
than the standard Crab-like radio pulsar. Although it
is widely accepted that radio pulsars are born in super-
novae, fewer than 10% of searches targeted at supernova
remnants have found young (age < 25 kyr) pulsars (Kaspi
et a. 1996, Gorham et al. 1996, Lorimer et al. 1998).
Some of this may be due to scattering at low frequencies
in the Galactic plane and the decreased sensitivity due
to the bright remnants, but it also hints at the unrec-
ognized diversity of the neutron star population. While
some neutron stars are detected as radio pulsars, others
may be born as anomalous X-ray pulsars, soft gamma-ray
repeaters, quiescent neutron stars, or other classes of ob-
jects not yet discovered (Kaspi 2000, Gotthelf & Vasisht
2000). The pathways to and percentages of these different
classes of objects are not yet clear. However, future Chan-
dra detections of compact X-ray sources, and large-scale
radio pulsar surveys, like the Parkes Multibeam Survey
(Lyne et al. 2000), will help to answer these questions.
We wish to thank Miller Goss for expediting our ob-
servations and Duncan Lorimer for helpful comments on
the manuscript. We thank Larry Rudnick and Barron Ko-
ralesky for kindly providing their radio images of Cas A
for our analysis. This work was partially supported by
NSF grant AST-9618408. The work was also supported
by the National Astronomy and Ionosphere Center, which
is operated by Cornell University under cooperative agree-
ment with the National Science Foundation (NSF). The
National Radio Astronomy Observatory is a facility of
the National Science Foundation operated under cooper-
ative agreement by Associated Universities, Inc. BMG
acknowledges the support of NASA through Hubble Fel-
lowship grant HST-HF-01107.01-A awarded by the Space
Telescope Science Institute, which is operated by the Asso-
ciation of Universities for Research in Astronomy, Inc., for
NASA under contract NAS 5 -- 26555. VMK acknowledges
support from an Alfred P. Sloan research fellowship, from
an NSF CAREER award (AST-9875897) and an NSERC
Research Grant (RGPIN228738-00).
4
The predicted scintillation bandwidth and timescale (see
Cordes & Rickett 1998), assuming a uniform medium and
a velocity of 50 km s−1 are approximately 0.1 kHz and
30 seconds at 327 MHz and 0.6 MHz and 150 seconds at
1435 MHz, in both cases too small to significantly affect
our quoted upper limits. If this velocity is higher than 50
km s−1, as a young pulsar's velocity is likely to be (Cordes
& Chernoff 1998), the predicted scintillation timescale will
be smaller and will therefore be even less of a factor in our
analysis.
We have also used VLA images of Cas A to search for a
radio continuum point-source at the position of the XPS.
At 1.3 GHz and 4.4 GHz, we find 5σ upper limits on the
flux density from the XPS of 40 mJy and 6 mJy respec-
tively. While not as constraining as those derived from our
VLA and Arecibo pulsed searches described above, these
continuum upper limits are completely independent of the
period, dispersion and scattering of any radio pulsations
from this source.
5. CONCLUSIONS
We have calculated upper limits of 30 and 1.3 mJy at
327 and 1435 MHz for the phase-averaged pulsed flux den-
sity from the XPS in Cas A. We did not detect any single,
dispersed pulses above our sensitivity of 25 and 1.0 Jy (as-
suming a pulse width of 1 ms) at 327 and 1435 MHz. We
now compare our upper limit to those from previous Cas A
searches. Davies & Large (1970) carried out an unsuccess-
ful 408-MHz search for single pulses from Cas A. Seiradakis
& Graham (1980) quoted a 1420-MHz upper limit of 1.6
mJy for pulsed emission from Cas A. However, their limit
does not account for the additional system temperature
contributed by the nebula, significant for their beamwidth
of 9′. More recently, Woan & Duffett-Smith (1993) pub-
lished an upper limit of 80 mJy at 408 MHz and Lorimer et
al. (1998) published an upper limit of 46 mJy at 606 MHz.
For a typical pulsar spectrum, our 1435-MHz upper limit
is an order of magnitude better than these previously pub-
lished limits, due to the known position of the XPS and
the power of the VLA to resolve out the excess emission
from the nebula.
Our upper limit of 1.3 mJy at 1435 MHz translates to a
luminosity of 15 mJy kpc2 for a neutron star at a distance
of 3.4 kpc. This is lower than the mean 1400-MHz lumi-
nosity of 30 mJy kpc2 estimated for young pulsars (Frail
& Moffett 1993) and is lower than the luminosity of any
known radio pulsar with characteristic age less than 104
years (see Figures 1 and 2 of Brazier & Johnston 1999).
This suggests that the XPS in Cas A is either not a pul-
sar, is a radio-quiet pulsar, or is a radio pulsar beamed
away from us. While evidence suggests that pulsar beams
are wider at short periods (see e.g. Biggs (1990)) and that
they are therefore likely to intersect the Earth's line of
sight, individual pulsars show mixed properties. For in-
stance, the Crab pulsar shows narrow radio components
that may be part of a large beam, but the profile of another
young pulsar in the Large Magellanic Cloud, PSR B0540-
69, is very broad (Manchester et al. 1993). Furthermore,
it is difficult to disentangle true beam width from viewing
angle for these pulsars. It is certainly possible that, like
the Geminga pulsar, the XPS shows no radio pulses but
is an X-ray, and perhaps gamma-ray, pulsar. Although
there is no EGRET point source near the position of the
REFERENCES
5
Aschenback, B., 1999, IAU Circ. 7249
Ashworth, W. B., 1980, J. Hist. Astron., 11, 1
Biggs, J. D. 1990, MNRAS, 245, 514
Brazier, K. T. S. & Johnston, S. 1999, MNRAS, 305, 671
Camilo, F. et al. ASP Conf. Ser. 202: IAU Colloq. 177: Pulsar
Astronomy - 2000 and Beyond, 3
Cordes, J. M., & Chernoff, D. F., 1997, ApJ, 482, 971
Cordes, J. M. & Chernoff, D. F. 1998, ApJ, 505, 31
Cordes, J. M., & Rickett, B. J. 1998, ApJ, 507, 846
Davies, J. G., & Large, M. I., 1970, MNRAS, 149, 301
Chakrabarty, D., Pivovaroff, M. J., Hernquist, L. E., Heyl, J. S., &
Narayan, R., 2001, ApJ, 546, in press
Flamsteed, J. 1725, London : H. Meere; in folio; DCC.f.9, DCC.f.10,
DCC.f.11
Frail, D. A, & Moffett, D. A., 1993, ApJ, 408, 67
Gorham, P., Ray, P., Anderson, S., Kulkarni, S., Prince, T. 1996,
ApJ, 458, 257
Gotthelf, E. V., & Vasisht, G., 2000, ASP Conf. Ser. 202: IAU Colloq.
177: Pulsar Astronomy - 2000 and Beyond, 699
Green D. A., 2000, 'A Catalogue of Galactic Supernova Remnants
(2000 August version)', Mullard Radio Astronomy Observatory,
Cavendish
Laboratory,
Cambridge, United Kingdom (available on the World-Wide-Web
at "http://www.mrao.cam.ac.uk/surveys/snrs/")
Hughes, J. P., Rakowski, C. E., Burrows, D. N., & Slane, P. O.
2000,ApJ, 528, L109
Kaspi, V., Manchester, R., Johnston, S., Lyne, A., D'Amico, N. 1996,
AJ, 111, 2028
Kaspi, V. M. 2000, ASP Conf. Ser. 202: IAU Colloq. 177: Pulsar
Astronomy - 2000 and Beyond, 485
Lorimer, D. R, Lyne, A. G., & Camilo, F., 1998, A&A, 331, 1002
Lundgren, S. C. et al. 1995, ApJ, 453, 433
Lyne, A. G., & Smith, F. G., 1998, Pulsar Astronomy, 2nd ed.,
Cambridge University Press, Cambridge
Lyne, A. G. et al. 2000, MNRAS, 312, 69
Manchester, R. N., Mar, D. P., Lyne, A. G., Kaspi, V. M. and
Johnston, S. 1993, 403, L29
Moffett, D. A. 1997, PhD. Thesis
Pavlov, G. G. & Zavlin, V. E., 1999, IAU Circ. 7270
Reed, J. E., Hester, J. J., Fabian, A. C., & Winkler, P. F. 1995, ApJ,
440, 706
Seiradakis, J. H. & Graham, D. A., 1980, A&A, 85, 353
Staelin, D. H. & Reifenstein, E. C., 1968, Science, 162, 1481
Tananbaum, H. 1999, IAU Circ. 7246
Taylor, J. H., Manchester, R. N., & Lyne, A. G. 1993, ApJS, 88, 529
Taylor, J. H. & Cordes, J. M. 1993,ApJ, 411, 674
van den Bergh, S., & Kamper, K. W. 1983, ApJ, 268, 129
van den Bergh, S. & Pritchet, C. J. 1986, ApJ, 307, 723
Woan, G. & Duffett-Smith, P. J., 1993, MNRAS, 260, 693
|
astro-ph/0611374 | 1 | 0611 | 2006-11-13T05:01:19 | Spectropolarimetry of the 3.4 micron absorption feature in NGC 1068 | [
"astro-ph"
] | In order to test the silicate-core/organic-mantle model of galactic interstellar dust, we have performed spectropolarimetry of the 3.4 micron C-H bond stretch that is characteristic of aliphatic hydrocarbons, using the nucleus of the Seyfert 2 galaxy, NGC 1068, as a bright, dusty background source. Polarization calculations show that, if the grains in NGC 1068 had the properties assigned by the core-mantle model to dust in the galactic diffuse ISM, they would cause a detectable rise in polarization over the 3.4 micron feature. No such increase is observed. We discuss modifications to the basic core-mantle model, such as changes in grain size or the existence of additional non-hydrocarbon aligned grain populations, which could better fit the observational evidence. However, we emphasize that the absence of polarization over the 3.4 micron band in NGC 1068 - and, indeed, in every line of sight examined to date - can be readily explained by a population of small, unaligned carbonaceous grains with no physical connection to the silicates. | astro-ph | astro-ph |
Spectropolarimetry of the 3.4 µm absorption feature in NGC 1068
Gemini Observatory Northern Operations Centre, 670 N. A'ohoku Place, Hilo, HI 96720, USA
R. E. Mason
[email protected]
G. S. Wright
UK Astronomy Technology center, Royal Observatory Edinburgh, Blackford Hill, Edinburgh EH9 3HJ, UK
[email protected]
A. Adamson
Joint Astronomy Centre, 660 North A'ohoku Place, Hilo, HI 96720, USA
[email protected]
and
Y. Pendleton
NASA Ames Research Center, Mail Stop 245-3, Moffet Field CA 94035, USA
[email protected]
ABSTRACT
In order to test the silicate-core/organic-mantle model of galactic interstellar dust, we have
performed spectropolarimetry of the 3.4 µm C -- H bond stretch that is characteristic of aliphatic
hydrocarbons, using the nucleus of the Seyfert 2 galaxy, NGC1068, as a bright, dusty background
source. Polarization calculations show that, if the grains in NGC1068 had the properties assigned
by the core-mantle model to dust in the galactic diffuse ISM, they would cause a detectable rise
in polarization over the 3.4 µm feature. No such increase is observed. We discuss modifications
to the basic core-mantle model, such as changes in grain size or the existence of additional
non-hydrocarbon aligned grain populations, which could better fit the observational evidence.
However, we emphasize that the absence of polarization over the 3.4 µm band in NGC1068 --
and, indeed, in every line of sight examined to date -- can be readily explained by a population
of small, unaligned carbonaceous grains with no physical connection to the silicates.
Subject headings: galaxies: individual: NGC1068 -- galaxies: ISM -- galaxies: Seyfert -- dust, extinction
-- polarization
1.
Introduction
In our galaxy and no doubt others, stars and
planets form from components available in the lo-
cal interstellar medium. The organic compounds
observed in interstellar space may therefore be the
first step towards the complex materials that help
make planets habitable. A greater understanding
of the origin and evolution of the organic materi-
als in the ISM, both in our galaxy and in others,
is thus of great interest.
Particularly puzzling are the aliphatic hy-
drocarbons, whose presence in the diffuse ISM
(DISM) of the Milky Way has been established
through observations of C-H bond stretches near
1
3.4 µm, seen in absorption throughout the DISM
of our own galaxy and others (Sandford et al.
1991; Pendleton et al. 1994; Sandford et al. 1995;
Imanishi 2000; Rawlings et al. 2003; Risaliti et al.
2003; Dartois et al. 2004; Mason et al. 2004). Two
very different models exist to explain the pro-
duction of this material: photoprocessing of the
icy dust grain mantles found in dense molecular
clouds (e.g. Li & Greenberg 1997), and hydro-
genation in the DISM of small, amorphous carbon
particles of the kind produced in outflows from
evolved stars (e.g. Mennella et al. 2002). As the
former process would result in the 3.4 µm band of
DISM hydrocarbons arising in organic mantles on
silicate grain cores, it is commonly known as the
"core-mantle" dust model.
Both of these models are to some extent con-
sistent with the observational data. For instance,
laboratory experiments mimicking both dust for-
mation pathways have succeeded in producing ma-
terials with 3.4 µm bands which are a good match
to the astronomical
feature (Greenberg et al.
1995; Mennella et al. 1999, 2002), although the
critically diagnostic 5-10 µm region reveals a
much closer match between observations and hy-
drogenated amorphous carbon (HAC) materi-
als (Chiar et al. 2000; Pendleton & Allamandola
2002). The presence of the 3.4 µm band in a pro-
toplanetary nebula (Chiar et al. 1998) shows that
hydrocarbon solids can be produced in circumstel-
lar regions, while the attribution of the 6 µm ex-
cess absorption in protostars to organic refractory
material (Gibb & Whittet 2002) would,
if con-
firmed, indicate that ice processing can also pro-
duce some fraction of the hydrocarbons. The or-
ganic OCN− band at 4.62 µm provides evidence of
ice processing in such environments (Demyk et al.
1998; Pendleton et al. 1999).
Spectropolarimetry provides opportunities for
clearly discriminating between these scenarios. In
the core-mantle model, the silicate core/organic
mantle grains are responsible for the bulk of the
visual and IR extinction and all of the polariza-
tion in the DISM (Li & Greenberg 1997). As the
polarization is caused by dichroic absorption, we
would therefore expect to observe polarization of
the continuum, with an increase in polarization
over absorption features associated with the grains
(e.g. Hough et al. 1988; Pendleton et al. 1990;
Smith et al. 2000). Specifically, increases in polar-
2
ization must occur simultaneously over the 3.4 µm
"mantle" feature, and the 9.7 µm absorption char-
acteristic of the silicate core, with roughly com-
parable efficiency (Li & Greenberg 2002). This
was first investigated by Adamson et al. (1999),
who compared the polarization over the 3.4 µm
feature towards the Galactic center source, IRS 7,
with that of the 9.7 µm band towards the nearby
Galactic center source, IRS 3. The increase in po-
larization over the 3.4 µm absorption was found
to be .0.08%, much less than the 0.4% expected
on the basis of the silicate feature polarization, in
apparent conflict with the core-mantle model.
However, as pointed out by Li & Greenberg
(2002), the silicate feature polarization used in
that work, while arising toward an object at a
projected distance of about 0.25pc from that to-
wards which the 3.4 µm polarization was mea-
sured (Geballe et al. 1989), does not refer to ex-
actly the same line of sight, admitting the possi-
bility that the silicate feature in the diffuse ISM
towards IRS 7 may be less polarized than ex-
pected. Studies of additional lines of sight (at
lower signal-to-noise ratio and/or spectral resolu-
tion) while reaching similar conclusions to those of
Adamson et al. (1999) , suffer from similar limita-
tions (Nagata et al. 1994; Ishii et al. 2002). More
recently, Chiar et al. (2006) measured the polar-
ization of both the hydrocarbon and silicate fea-
tures towards the Quintuplet Cluster in the Galac-
tic Center, finding the hydrocarbons to be signif-
icantly less polarized than would be expected if
they existed as mantles on the polarized silicate
grains.
To provide another data point against which
the predictions of the core-mantle grain model can
be compared, and to extend this study to a new
and different environment, we have used the bright
nucleus of the Seyfert 2 galaxy NGC1068 as a
source against which to measure the polarization
of the 3.4 µm band and the continuum around it.
Rather than the 10 µm silicate feature, we have
used the degree of continuum polarization to cal-
culate the enhancement in polarization that would
be expected over the 3.4 µm feature from a screen
of elongated, coated silicate grains which polarize
by the selective absorption of one plane of the in-
cident light. In §2, we describe the line of sight to-
wards the nucleus of NGC1068 and its suitability
for this work. The spectropolarimetric observa-
tions and their treatment are outlined in §3. The
calculations that we have carried out are discussed
in §4, and the results presented in §5. This work
is then summarized in §6.
2. Dust and dichroic polarization in NGC 1068
As the archetypal Seyfert 2 galaxy, NGC 1068
(d=16 Mpc; H0 = 70km s−1 Mpc−1; 1′′ = 72
pc) has been the center of much attention. The
unified model of active galactic nuclei (AGN) im-
plies that the galaxy harbors a Seyfert 1 nu-
cleus, obscured from our point of view by a torus
of dust and molecular gas, and 10 µm interfer-
ometry (Jaffe et al. 2004) as well as the detec-
tion of broad emission lines in polarized light
(Antonucci & Miller 1985) have provided strong
evidence for the existence of such a torus. Co-
pious other mid-IR and X-ray data also show
that the nucleus of NGC 1068 is obscured by
large amounts of warm dust (e.g. Bock et al. 2000;
Matt et al. 2000; Tomono et al. 2001; Alloin et al.
2000; Mason et al. 2006). While the classification
of NGC 1068 as a type 2 object suggests that the
obscuring torus is oriented roughly edge-on to our
line of sight, the disk of the galaxy has quite a low
inclination (i ≈ 29 ◦; Garc´ıa-G´omez et al. 2002).
This means that the line of sight to the center of
NGC 1068 samples dust local to the active nu-
cleus, with a negligible contribution from dust in
the disk of the galaxy.
The 3.4 µm absorption band has been observed
in NGC 1068 (Bridger et al. 1994; Wright et al.
1996; Imanishi et al. 1997; Marco & Brooks 2003;
Mason et al. 2004), as has silicate absorption at
9.7 µm (Kleinmann et al. 1976; Roche et al. 1984;
Jaffe et al. 2004; Mason et al. 2006; Rhee & Larkin
2006). The 3.4 µm band has a very similar pro-
file to that observed in Galactic lines of sight;
although the band carrier exists in the central re-
gion of an active galaxy, it apparently undergoes
little extra processing compared to the Galactic
DISM (Dartois et al. 2004; Mason et al. 2004).
NGC 1068 has a nuclear dust spiral which ap-
pears physically connected to the galaxy disk
(Pogge & Martini 2002), so it is quite possible
that the hydrocarbons in the nucleus of NGC 1068
formed in the disk of the galaxy and then mi-
grated to the center. We therefore proceed on the
assumption that the hydrocarbon-containing dust
3
in the nucleus of NGC 1068 initially formed by one
of the mechanisms that has been suggested to be
responsible for Galactic aliphatic hydrocarbons.
The nucleus of NGC 1068 is known to be po-
larized from the UV to the mid-IR, and the mech-
anisms responsible for this polarization have been
the subject of intense study. Antonucci & Miller
(1985) were among the first to observe the UV
and visible polarization of NGC 1068, and they
interpreted the high, wavelength-independent po-
larization as being due to scattering off very small
particles, probably free electrons. Bailey et al.
(1988) later discussed the polarization properties
of NGC 1068 from the UV out to 10µm, and found
a twist in position angle from the optical into the
near-IR. They interpreted this as dust or electron
scattering being replaced by dichroic absorption,
the preferential absorption of one component of
the electric vector, as the most important polariz-
ing mechanism. Their data also show a change
of ∼ 70◦ in position angle between the L and
M bands, consistent with dichroic emission from
aligned dust grains becoming the dominant polar-
izing mechanism at longer wavelengths.
Further evidence for the importance of dichroic
absorption as a polarizing mechanism in the IR
comes from imaging polarimetry. After analyzing
the deviations from the centrosymmetric patterns
expected from scattering alone, Lumsden et al.
(1999) concluded that their data require another
mechanism of constant position angle contribut-
ing to the polarization. The effect of this mecha-
nism grows with increasing wavelength, as would
be expected if absorption were beginning to domi-
nate over scattering, and was estimated to account
for > 90% of the K band polarization. Lums-
den at el. were successful in fitting their observed
JHK polarized flux points with greybody emission
from hot (T ∼ 1200K) dust, reddened by a λ−1.75
extinction curve and scaled by a Serkowski law
appropriate for moderately extinguished Galactic
sources. Furthermore, Packham et al. (1997) in-
terpreted the different aperture dependence of the
J-, H- and K- band polarization as consistent with
a dichroic contribution in the H- and K-bands.
With higher-resolution HST imaging polarimetry,
Simpson et al. (2002) also found that the K-band
polarization has a contribution from dichroism,
but they were able to set tighter limits on the lo-
cation of the dichroic component, to within 1′′ of
the nucleus.
This agrees with the more detailed modeling of
Young et al. (1995), who find that neither electron
nor dust scattering can account for the near-IR
polarization of NGC 1068. However, if a dichroic
component is added, the near-IR polarized flux
spectrum can be reproduced. In this model there
is still some contribution from electron scattering
to the near-IR flux, but this decreases with wave-
length relative to the dichroic component. By the
K band, electron scattering contributes perhaps
10% of the total flux, with dust making up almost
all of the remainder. At L, the scattering contri-
bution in their model would be negligible.
Watanabe et al. (2003) are also successful in
modeling the near-IR polarization of NGC 1068
with polarization from aligned dust grains. In ad-
dition, they raise the possibility that the IR po-
larization could be caused by scattering off large
grains in the torus, and point out that neither
the 70 ◦change in position angle nor the absence
of a centrosymmetric scattering pattern at longer
wavelengths necessarily rules out scattering as the
polarizing mechanism. This has yet to be tested in
any detail, but the implications that it may have
for the conclusions of this study are discussed in
§5.
While the near-IR data point to dust absorption
being the dominant polarizing mechanism in the
L band, spectropolarimetry of the 9.7 µm silicate
feature does not show the pronounced polarization
excess that might be expected if emission or ab-
sorption from aligned silicates were causing the po-
larization (Aitken et al. 1984). This observation
was interpreted as evidence that the mid-IR polar-
ization arises in emission from non-silicate grains
or has a nonthermal origin. However, the lack of
polarization in the silicate feature does not neces-
sarily rule out the presence of aligned silicate-core
grains in the nucleus of NGC 1068. In radiative
transfer calculations of dichroic polarization from
silicate-containing grain mixtures in dusty disks,
Efstathiou et al. (1997) and Aitken et al. (2002)
find numerous configurations in which the polar-
ization over the feature is in fact quite flat, consis-
tent with the observations of Aitken et al. (1984).
Although it appears possible for radiative trans-
fer effects to suppress the silicate feature polariza-
tion, the shorter-wavelength 3.4 µm band should
be much less affected by such interplay between
absorption and emission. The ratio of the optical
depths of the 3.4 and 9.7µm bands in NGC 1068
supports this suggestion. In the five Galactic lines
of sight examined by Sandford et al. (1995) there
is a fair degree of correlation between the depths
of the two features (τ9.7/τ3.4 = 13 − 19, with the
higher values towards the Galactic Centre), but
in NGC 1068 τ9.7/τ3.4 ≈ 5, suggesting that the
3.4 µm band is indeed less affected by underly-
ing emission than is the silicate feature. This fur-
ther suggests that treating the dust producing the
3.4 µm band as a uniform absorbing screen (an
assumption implicit in the calculations outlined in
§ 4), while undoubtedly a major simplification of
the dust geometry and temperature structure in
this AGN, is still a useful approximation. Overall,
imaging polarimetry, spectral modeling and the
near-90 ◦ rotation in position angle all imply that
a model of the L-band polarization of NGC 1068
based on selective dust absorption is a reasonable
representation of the true situation.
3. Observations and Data Reduction
L-band spectropolarimetry of the nucleus of
NGC1068 was obtained on the nights of 2000
September 19 and 2000 October 6 using the IR-
POL2 spectropolarimetry module and CGS4 on
the 3.8m UK Infrared Telescope on Mauna Kea,
Hawaii. The 40 l/mm grating and 0.6′′-wide slit
were used, providing R=1360 at 3.4 µm. The
weather conditions were good during the second
night, but some thin cirrus was present on the
first1.
In order to obtain Stokes q and u parameter
spectra a frame was taken with IRPOL2's half-
wave plate at each of four positions: 0◦, 45◦, 22.5◦,
and 67.5◦. This cycle was repeated 64 times in all.
At each of these waveplate angles, ordinary and
extraordinary beams were extracted from the sky-
subtracted frames, stacked together, and the final,
total spectra combined using the "ratio" method
(see e.g. Tinbergen 1996) to produce Stokes q and
u parameter spectra. This method has the advan-
tage of minimizing the effects of variations in sky
transmission during the observations.
Prior to calculation of the polarization, further
1Spectropolarimetry of the 9.7 µm silicate feature was later
also attempted, but poor weather meant that none of the
data obtained were useful.
4
treatment of the raw q and u spectra was nec-
essary. Firstly, data points between 3.3 and 3.4
µm (observed) were rejected. This part of the
spectrum contains deep absorptions from the hy-
drocarbon cement in IRPOL2's Wollaston prism
(at 3.35-3.4 µm) and from atmospheric methane
(3.31-3.33 µm), and these lines do not ratio out at
all satisfactorily.
A second effect which must be dealt with is
a large-amplitude ripple in the spectrum which
is thought to be caused by multiple reflections
in the waveplate. The ripple was removed using
the FFT technique described by Adamson et al.
(1999). Briefly, the ripple causes peaks to ap-
pear around 0.5 and 1.0 times the Nyquist fre-
quency in the Fourier transform of the spectrum.
These peaks were set to zero, then the transform
was inverted. The instrumental zero point of po-
larization was determined and removed from the
NGC1068 q and u spectra using observations of
the unpolarized star HD18803 (Clayton & Martin
1981), and the resulting polarization spectrum
corrected for the imperfect efficiency of the wave-
plate2. The position angle of polarization (PA)
was calibrated using observations of the young
stellar object, AFGL2519, whose PA at 3-4 µm has
previously been measured by Hough et al. (1989).
An intensity spectrum was also constructed using
only a few frames taken near in time to HD18803,
which was used as a telluric standard.
The polarization, position angle and total flux
spectra of NGC1068 are shown in Figure 1, to-
gether with the polarimetric results of Bailey et al.
(1988) and Lebofsky et al. (1978).
4. Polarization Calculations
To calculate the L-band polarization produced
by coated dust grains, we have taken advantage
of the discrete dipole approximation (DDA) code,
DDSCAT6.1 (Draine 1988; Draine & Flatau 1994,
2004). In the DDA the dust grain is represented
as an array of dipole oscillators on a cubic lattice,
and absorption, scattering and extinction cross-
sections are then calculated from the amplitudes
of the dipoles as they interact with the incident
electric field.
2 as given by
www.jach.hawaii.edu/UKIRT/instruments/irpol/CGS4/cgs4pol.html#5
The optical constants adopted for the grain
cores and mantles are those of "astronomical sili-
cate" (Draine & Lee 1984; Draine 1985; Weingartner & Draine
2001) and the organic refractory material of
Li & Greenberg (1997). The refractive index of
the organic component is derived from the Murchi-
son meteorite and laboratory residues exposed to
the solar UV field aboard the EURECA satel-
lite (Greenberg et al. 1995; Greenberg & Li 1996).
While both of those materials show absorption
bands at 5 - 9 µm that are not seen in the dif-
fuse ISM (Pendleton & Allamandola 2002), the
3.4 µm bands of both bear a good resemblance to
the observed feature.
The existence of polarization caused by dust
absorption indicates that the dust grains must
be aspherical, but beyond that their shapes are
not well-established. We have treated the dust
grains as oblate spheroids with an axial ratio of
2:1, the cores and mantles having equal eccen-
tricities. Based on fits to the Trapezium 9.7µm
silicate feature, Draine & Lee (1984) find that
2:1 oblate spheroids are a reasonable representa-
tion of dust grains, and Hildebrand & Dragovan
(1995) also favor oblate over prolate spheroids.
They conclude that the axial ratio must be <3:1,
with a best-fitting value of 1.5:1. There has
been some dispute over whether oblate or prolate
spheroids give a better fit to the observational data
(Greenberg & Li 1996), but we note that the main
discrepancy in the polarization predicted by these
two shapes comes over the 9.7µm silicate feature.
The differences are small at the wavelengths of in-
terest in this work (see Fig. 6 of Draine & Lee
1984). The elongation of the grains will affect the
efficiency with which they polarize, but has little
effect on the wavelength dependence of the polar-
ization (Greenberg & Li 1996).
We have adopted simple picket-fence alignment
for this work (which is equivalent to the per-
fect spinning alignment approximation for oblate
grains), with the grains' short axes orientated per-
pendicular to the direction of propagation of the
incident light. This has been shown to be quite ad-
equate if only the wavelength dependence of the
polarization is required, rather than its absolute
value (Chlewicki & Greenberg 1990). The func-
tional form used for the grain size distribution is
the MRN a−3.5 power law with limits of 0.005 and
0.25 µm (Mathis et al. 1977).
5
Given this information about the size, shape,
composition and orientation of the grains, DDSCAT
calculates a number of quantities. Those relevant
to this work are
Qabs = Cabs/πa2
Qsca = Csca/πa2
Qext = Qabs + Qsca
Qpol = Qext, k −Qext, ⊥
(1)
(2)
(3)
(4)
in which Qs are efficiencies and Cs cross-sections
for absorption, scattering and extinction, and
Qext, k and Qext, ⊥ refer to extinction efficien-
cies for the two orthogonal incident polarization
states (Draine 1988).
From Qext and Qpol, extinction and polariza-
tion spectra for ensembles of grains can be ob-
tained in the following manner:
τ (λ) = Ngrain Z amax
p(λ) = Ngrain Z amax
amin
amin
n(a)CgeoQext(a, λ)da (5)
n(a)Cgeo
Qpol
2
(a, λ)da (6)
where Cgeo is the geometrical cross-section of the
grain in question, Ngrain the column density of
grains (with n(a), the number of grains in the
interval a, a + da, appropriately normalized) and
amin and amax the limits of the size distribution
to be considered.
5. Results and discussion
We are interested in the magnitude of any in-
crease in polarization that would be expected over
the 3.4 µm absorption band, given a certain con-
tinuum polarization caused by dichroic absorption
by core-mantle grains. To obtain a crude esti-
mate of the polarization that might be expected
over the 3.4 µm band, we can consider the polar-
ization observed over the 9.7 µm silicate band in
galactic lines of sight. For a given grain shape,
size and composition, changes in abundance and
degree of alignment will alter the level of con-
tinuum polarization without affecting the ratio
of feature to continuum polarization, Pf eat/Pcont.
In galactic lines of sight that can be fitted with
pure absorptive polarization, this ratio is typically
about 3-4, with a couple of objects reaching val-
ues of ∼7 (OMC1 BN, SgrA GCS3 II; Smith et al.
6
(2000)). Alternatively, the optical constants of
astronomical silicate give rise to Pf eat/Pcont ≈
12 (Draine & Lee 1984; Hildebrand & Dragovan
1995). The "excess" polarization over an absorp-
tion feature is determined by the strength of the
band, and Li & Greenberg (2002) have shown that
silicates and hydrocarbons polarize to a similar de-
gree per unit optical depth. Observed values of
τ9.7/τ3.4 range from about 13-19 (Sandford et al.
1995), implying Pf eat/Pcont ∼1.15 - 2.0 in the 3.4
µm band. For NGC1068, where the L-band con-
tinuum polarization is about 2.5%, this translates
to peak polarizations of 2.9 - 5% across the 3.4
µm feature. This suggests that a significant ex-
cess polarization should be detectable over the 3.4
µm feature, although the spread in silicate feature
polarizations and τ9.7/τ3.4, and the fact that most
of the sources in Smith et al. (2000) are dominated
by molecular cloud material mean that these num-
bers should be treated with some caution.
To estimate the likely maximum observed
polarization change over the 3.4 µm band in
NGC 1068, the best-fitting linear continuum was
first found for the polarization data. A model
polarization spectrum was then created from the
optical depth spectrum derived from the spec-
tropolarimetry data, assuming that p ∝ τ . The
model spectrum was normalized to the continuum
around 3.55 µm, then scaled by various factors and
the reduced χ2 between it and the binned polar-
ization spectrum calculated for each scaling. The
best fit corresponds to a deviation of -0.12±0.25%
(95% confidence) from the linear continuum; the
data are in fact consistent with a small decrease in
polarization over the 3.4 µm feature. The model
fit giving the upper limit on the change in po-
larization over the 3.4 µm band (i.e., peaking at
0.13% above the continuum) is shown in Figure 2.
The polarization calculated as described in
§ 4 for a population of core-mantle grains with
the MRN size distribution is also shown in Fig.
2. As grain size may affect the magnitude of
the 3.4 µm polarization excess (§5.1), the po-
larization from a 0.2 µm grain, close to the
largest size in the MRN distribution and some-
what larger than the average < a >= 0.087µm
of the finite cylinders in the Li & Greenberg
(1997) model,
The different
curves in the figure result from different values
of fcarb, the fraction of the grain volume that
is also shown.
is contained in the mantle. Various figures have
been suggested for this quantity; Li & Greenberg
(2002) consider fcarb = 0.2, 0.5, 0.75 while ac-
cording to Chlewicki & Greenberg (1990), fcarb ∼
0.9. Greenberg & Li (1996) and Li & Greenberg
(1997) estimate mantle/core volume ratios of
about 2:1. As expected, the polarization at
the peak of the 3.4 µm band increases with the
amount of carbonaceous material in the grains,
from approximately 3.0% for the 25% -mantle
MRN grains, to about 3.6% for the 75%-mantle
particles.
For the core/mantle ratios closest to those most
commonly quoted in the literature, the 3.4 µm po-
larization predicted by the model is clearly well in
excess of the observed polarization. In the case of
the grains with only 25% of the volume contained
in the mantle, the polarization at the peak of the
3.4 µm absorption is 0.29% over the linear contin-
uum fit mentioned above. This is still somewhat
above the limit we derive on the observed polar-
ization; the expected and observed polarizations
are not reconciled even with only a small fraction
of each grain being composed of organic refractory
material. The core-mantle grain model, at least in
the form proposed for the Galactic diffuse ISM,
is unlikely to explain the L-band polarization of
NGC 1068.
5.1. Variations on the basic core-mantle
model
Are there likely to be significant differences in
the grain population(s) in NGC1068 that might di-
minish the amount of polarization expected over
the 3.4 µm band, even while core-mantle grains
are present? The similarity of the band profile
in NGC 1068 and the Galactic diffuse ISM sug-
gests that the composition of the hydrocarbon
material is similar in both galaxies (Dartois et al.
2004; Mason et al. 2004), but other differences
may arise. For instance, as the Li & Greenberg
(1997) model proposes that the entire infrared
polarization arises in the core-mantle grains, we
have so far also assumed this, but if another, non-
hydrocarbon-containing aligned grain component
also contributed to the polarization in NGC1068 it
would diminish the size of the rise in polarization
expected over the 3.4 µm band while contributing
to AV but not τ3.4. In NGC1068, such a second
polarizing grain population could conceivably be,
7
for example, bare silicates arising from destruction
of the mantles on some fraction of the grains.
Values of extinction to the complex, infrared-
emitting regions of the nucleus of NGC1068 may
not be as straightforward to interpret as the ex-
tinction through the large column of cold dust to-
wards the Galactic Center, but estimates range
from ∼15 (Lumsden et al. 1999; Watanabe et al.
2003) to ∼40 (Young et al. 1995), disregarding
values based on τ3.4. Given that τ3.4 ≈ 0.1, this
implies AV /τ3.4 ∼ 150−400, compared with galac-
tic values of ∼150 (towards the galactic center;
Pendleton et al. 1994) or ∼several hundred (to-
wards field stars at various galactic longitudes;
Rawlings et al. 2003). There is therefore no com-
pelling reason to think that grain components
other than those proposed for the galactic diffuse
ISM must exist in NGC1068, but the wide range
of estimates of AV /τ3.4 for both the galactic and
extragalactic lines of sight means that this cannot
be ruled out either.
Another issue that could affect the 3.4 µm band
in both extinction and polarization is that of grain
size.
Increasing grain size tends to increase the
efficiency of continuum extinction while decreas-
ing the strength of absorption features, so large
grains might be able to polarize the L-band con-
tinuum effectively without producing much excess
polarization through the 3.4 µm band. For grains
much larger than about 0.2 µm in radius, scatter-
ing starts to become important (Qabs/Qsca ≈ 5
for a 0.2 µm core-mantle 2:1 oblate spheroid at
3.3 µm) and calculations of the polarization from
such particles in a complex system like NGC1068,
where the inclination of the dusty torus and mul-
tiple scattering effects may be critical, are beyond
the scope of this paper.
It has been suggested that dust grains in AGN
may be biased towards larger sizes than in the dif-
fuse ISM, but the evidence remains contradictory.
Laor & Draine (1993) pointed out that large (∼10
µm) dust grains are likely to survive longer than
small grains close to an AGN and will not produce
a silicate emission feature. The weakness or ab-
sence of silicate emission in type 1 AGN prompted
the inclusion of large grains in some torus models
(van Bemmel & Dullemond 2003), but its recent
discovery in several quasars may argue against
large grains (Hao et al. 2005; Siebenmorgen et al.
2005). Flat L-M′ colors, peculiar EB−V /NH and
these observations
AV /NH ratios in Seyfert 2 nuclei and the absence
of the 2175A feature in reddened Seyfert 1s have
all been interpreted as evidence for large grains
(Imanishi 2001; Maiolino et al. 2001a,b), but geo-
metrical effects may also be able to explain many
of
(Weingartner & Murray
2002).
In the specific case of NGC1068, both
Young et al. (1995) and Watanabe et al. (2003)
were able to fit the observed J- to K-band and
optical polarization with grain sizes no different
from those thought to exist in the galactic DISM,
although Watanabe et al.
suggest that scatter-
ing from large grains in the torus might be a vi-
able alternative. Detailed extinction and polariza-
tion calculations such as those of Watanabe et al.
(2003),
if extended to the 3.4 µm band, could
provide valuable constraints on both the size and
structure of the grains in AGN tori.
Finally, we note that if the grain size distribu-
tion is biased towards larger sizes in NGC1068,
this could be through preferential destruction of
small grains, or large grains may have grown by
coagulation in this warm, dense environment. If
the latter, then they may have lost some of their
former core-mantle nature.
It seems likely that
small hydrocarbon inclusions in large, coagulated
grains would leave their polarization signature on
the 3.4 µm band, but again, further calculations
would be needed to test this.
5.2. Hydrocarbon formation in the diffuse
ISM
The absence of excess polarization over the 3.4
µm band is naturally explained if the feature arises
in a population of grains that does not contribute
significantly to the continuum polarization. Such
a grain population must be small and/or optically
isotropic (small grains being harder to align than
large ones; Lazarian 2003, and references therein),
and have no physical connection to the silicate
grains. Recent laboratory work has provided per-
suasive evidence that small carbon grains like
those thought to be ejected from AGB stars can
be hydrogenated during the later stages of stellar
evolution (Schnaiter et al. 1999) and in the dif-
fuse ISM (Mennella et al. 1999; Munoz Caro et al.
2001; Mennella et al. 2002). Calculations indi-
cate that hydrogenation proceeds at a fast enough
rate in the diffuse ISM to balance the dehydro-
genation caused by UV photons. Conversely,
8
in dense molecular clouds, the reduced abun-
dance of atomic H and the presence of icy man-
tles act to prevent rehydrogenation of the grains,
which can still be efficiently dehydrogenated by
photons and cosmic rays (Mennella et al. 2001,
2003). These and other lines of evidence (e.g.
Shenoy et al. 2003) imply that most hydrocarbon-
containing grains are formed in the diffuse ISM
and that re-formation dominates over the dehy-
drogenation that subsequently occurs, in agree-
ment with the non-detection of the 3.4 µm feature
in dense cloud material. This evolutionary sce-
nario for the aliphatic hydrocarbons is entirely
consistent with the lack of a 3.4 µm polarization
excess in NGC 1068 and several Galactic lines
of sight (Adamson et al. 1999; Ishii et al. 2002;
Chiar et al. 2006).
6. Summary
We have performed L-band spectropolarimetry
of NGC 1068 and shown that the excess polariza-
tion over the 3.4 µm feature is below that which
would be expected on the basis of the silicate-
core/organic-mantle grain model as applied to the
galactic diffuse ISM, consistent with a growing
body of evidence suggesting that the aliphatic
hydrocarbons in the general diffuse ISM are not
formed by processing of the ice mantles that form
on silicate grains in molecular clouds. The coated
grain model could still be valid in NGC 1068
if there also exists an extra, non-hydrocarbon
aligned grain population, or possibly if the grain
size distribution is biased to larger sizes than in the
diffuse ISM of our galaxy (detailed calculations of
both the continuum and 3.4 µm feature polariza-
tion from micron-sized dust grains may be a useful
way of constraining the size and/or composition of
the carbonaceous grain population in AGN). Al-
ternatively, reaction of small carbon grains with
atomic hydrogen in the diffuse ISM would be ex-
pected to produce a population of small, nonpolar-
izing hydrocarbon-containing grains which would
naturally explain the lack of polarization of the
3.4 µm feature. Such a model is also successful
in accounting for the non-detection of the 3.4 µm
band in molecular clouds, which is otherwise diffi-
cult to explain.
7. Acknowledgments
We would like to thank R. Antonucci, P. Hirst,
M. Kishimoto, T. Roush, M. Smith and A. Tie-
lens for taking the time to comment, and A. Li for
providing the optical constants in tabular form.
We are grateful to the anonymous referee for com-
ments that strengthened the paper. We thank
the Department of Physical Sciences, University
of Hertfordshire for providing IRPOL2 for the
UKIRT. The United Kingdom Infrared Telescope
is operated by the Joint Astronomy center on be-
half of the U.K. Particle Physics and Astronomy
Research Council. Supported by the Gemini Ob-
servatory, which is operated by the Association of
Universities for Research in Astronomy, Inc., on
behalf of the international Gemini partnership of
Argentina, Australia, Brazil, Canada, Chile, the
United Kingdom, and the United States of Amer-
ica.
REFERENCES
Adamson, A. J., Whittet, D. C. B., Chrysosto-
mou, A., Hough, J. H., Aitken, D. K., Wright,
G. S., & Roche, P. F. 1999, ApJ, 512, 224
Aitken, D. K., Briggs, G., Bailey, J. A., Roche,
P. F., & Hough, J. H. 1984, Natur, 310, 660
Aitken, D. K., Efstathiou, A., McCall, A., &
Hough, J. H. 2002, MNRAS, 329, 647
Alloin, D., Pantin, E., Lagage, P. O., & Granato,
G. L. 2000, A&AP, 363, 926
Antonucci, R. R. J., & Miller, J. S. 1985, ApJ,
297, 621
Bailey, J., Axon, D. J., Hough, J. H., Ward, M. J.,
McLean, I., & Heathcote, S. R. 1988, MNRAS,
234, 899
Bock, J. J., Neugebauer, G., Matthews, K., Soifer,
B. T., Becklin, E. E., Ressler, M., Marsh, K.,
Werner, M. W., Egami, E., & Blandford, R.
2000, AJ, 120, 2904
Bridger, A., Wright, G. S., & Geballe, T. R. 1994,
in ASSL Vol. 190: Astronomy with Arrays, The
Next Generation, 537
Chiar, J. E., Adamson, A. J., Whittet, D. C. B.,
Chrysostomou, A., Hough, J. H., Kerr, T. H.,
9
Mason, R. E., Poche, P. F., & Wright, G. 2006,
ArXiv Astrophysics e-prints
Chiar, J. E., Pendleton, Y. J., Geballe, T. R., &
Tielens, A. G. G. M. 1998, ApJ, 507, 281
Chiar, J. E., Tielens, A. G. G. M., Whittet,
D. C. B., Schutte, W. A., Boogert, A. C. A.,
Lutz, D., van Dishoeck, E. F., & Bernstein,
M. P. 2000, ApJ, 537, 749
Chlewicki, G., & Greenberg, J. M. 1990, ApJ, 365,
230
Clayton, G. C., & Martin, P. G. 1981, AJ, 86, 1518
Dartois, E., Marco, O., Munoz-Caro, G. M.,
Brooks, K., Deboffle, D., & d'Hendecourt, L.
2004, A&A, 423, 549
Demyk, K., Dartois, E., D'Hendecourt, L., Jour-
dain de Muizon, M., Heras, A. M., & Breitfell-
ner, M. 1998, A&A, 339, 553
Draine, B. T. 1985, ApJS, 57, 587
-- . 1988, ApJ, 333, 848
Draine, B. T., & Flatau, J. 1994, J. Opt. Soc Am.,
11, 1491
Draine, B. T., & Flatau, P. J. 2004, User Guide
for the Discrete Dipole Approximation Code
DDSCAT (Version 6.1), astro-ph/0309069
Draine, B. T., & Lee, H. M. 1984, ApJ, 285, 89
Efstathiou, A., McCall, A., & Hough, J. H. 1997,
MNRAS, 285, 102
Garc´ıa-G´omez, C., Athanassoula, E., & Barber`a,
C. 2002, A&A, 389, 68
Geballe, T. R., Baas, F., & Wade, R. 1989, A&A,
208, 255
Gibb, E. L., & Whittet, D. C. B. 2002, ApJL, 566,
L113
Greenberg, J. M., & Li, A. 1996, A&A, 309, 258
Greenberg, J. M., Li, A., Mendoza-Gomez, C. X.,
Schutte, W. A., Gerakines, P. A., & de Groot,
M. 1995, ApJL, 455, L177
Hao, L., Spoon, H. W. W., Sloan, G. C., Mar-
shall, J. A., Armus, L., Tielens, A. G. G. M.,
Sargent, B., van Bemmel, I. M., Charmandaris,
V., Weedman, D. W., & Houck, J. R. 2005,
ApJ, 625, L75
Hildebrand, R. H., & Dragovan, M. 1995, ApJ,
450, 663
Hough, J. H., Sato, S., Tamura, M., Yamashita,
T., McFadean, A. D., Rouse, M. F., Kaifu, N.,
Suzuki, H., Nagata, T., Gatley, I., & Bailey, J.
1988, MNRAS, 230, 107
Hough, J. H., Whittet, D. C. B., Sato, S., Ya-
mashita, T., Tamura, M., Nagata, T., Aitken,
D. K., & Roche, P. F. 1989, MNRAS, 241, 71
Imanishi, M. 2000, MNRAS, 319, 331
-- . 2001, AJ, 121, 1927
Imanishi, M., Terada, H., Sugiyama, K., Moto-
hara, K., Goto, M., & Maihara, T. 1997, PASJ,
49, 69
Ishii, M., Nagata, T., Chrysostomou, A., &
Hough, J. H. 2002, AJ, 124, 2790
Jaffe, W., Meisenheimer, K., Rottgering, H. J. A.,
Leinert, C., Richichi, A., Chesneau, O., Fraix-
Burnet, D., Glazenborg-Kluttig, A., Granato,
G.-L., Graser, U., Heijligers, B., Kohler, R.,
Malbet, F., Miley, G. K., Paresce, F., Pel, J.-
W., Perrin, G., Przygodda, F., Schoeller, M.,
Sol, H., Waters, L. B. F. M., Weigelt, G.,
Woillez, J., & de Zeeuw, P. T. 2004, Nature,
429, 47
Maiolino, R., Marconi, A., & Oliva, E. 2001a,
A&A, 365, 37
Maiolino, R., Marconi, A., Salvati, M., Risaliti,
G., Severgnini, P., Oliva, E., La Franca, F., &
Vanzi, L. 2001b, A&A, 365, 28
Marco, O., & Brooks, K. J. 2003, A&A, 398, 101
Mason, R. E., Geballe, T. R., Packham, C., Lev-
enson, N. A., Elitzur, M., Fisher, R. S., & Perl-
man, E. 2006, ApJ, 640, 612
Mason, R. E., Wright, G., Pendleton, Y., & Adam-
son, A. 2004, ApJ, 613, 770
Mathis, J. S., Rumpl, W., & Nordsieck, K. H.
1977, ApJ, 217, 425
Matt, G., Fabian, A. C., Guainazzi, M., Iwasawa,
K., Bassani, L., & Malaguti, G. 2000, MNRAS,
318, 173
Mennella, V., Baratta, G. A., Esposito, A., Ferini,
G., & Pendleton, Y. J. 2003, ApJ, 587, 727
Mennella, V., Brucato, J. R., Colangeli, L., &
Palumbo, P. 1999, ApJL, 524, L71
-- . 2002, ApJ, 569, 531
Mennella, V., Munoz Caro, G. M., Ruiterkamp,
R., Schutte, W. A., Greenberg, J. M., Brucato,
J. R., & Colangeli, L. 2001, A&A, 367, 355
Munoz Caro, G. M., Ruiterkamp, R., Schutte,
W. A., Greenberg, J. M., & Mennella, V. 2001,
A&A, 367, 347
Nagata, T., Kobayashi, N., & Sato, S. 1994, ApJL,
Kleinmann, D. E., Gillett, F. C., & Wright, E. L.
423, L113+
1976, ApJ, 208, 42
Laor, A., & Draine, B. T. 1993, ApJ, 402, 441
Lazarian, A. 2003, JQSRT, 79, 881
Lebofsky, M. J., Kemp, J. C., & Rieke, G. H. 1978,
ApJ, 222, 95
Li, A., & Greenberg, J. M. 1997, A&A, 323, 566
-- . 2002, ApJ, 577, 789
Lumsden, S. L., Moore, T. J. T., Smith, C., Fu-
jiyoshi, T., Bland-Hawthorn, J., & Ward, M. J.
1999, MNRAS, 303, 209
Packham, C., Young, S., Hough, J. H., Axon,
D. J., & Bailey, J. A. 1997, MNRAS, 288, 375
Pendleton, Y. J., & Allamandola, L. J. 2002,
ApJS, 138, 75
Pendleton, Y. J., Sandford, S. A., Allamandola,
L. J., Tielens, A. G. G. M., & Sellgren, K. 1994,
ApJ, 437, 683
Pendleton, Y. J., Tielens, A. G. G. M., Tokunaga,
A. T., & Bernstein, M. P. 1999, ApJ, 513, 294
Pendleton, Y. J., Tielens, A. G. G. M., & Werner,
M. W. 1990, ApJ, 349, 107
10
Pogge, R. W., & Martini, P. 2002, ApJ, 569, 624
Weingartner, J. C., & Murray, N. 2002, ApJ, 580,
Rawlings, M. G., Adamson, A. J., & Whittet,
D. C. B. 2003, MNRAS, 341, 1121
Rhee, J. H., & Larkin, J. E. 2006, ApJ, 640, 625
Risaliti, G., Maiolino, R., Marconi, A., Bas-
sani, L., Berta, S., Braito, V., Della Ceca, R.,
Franceschini, A., & Salvati, M. 2003, ApJL,
595, L17
Roche, P. F., Whitmore, B., Aitken, D. K., &
Phillips, M. M. 1984, MNRAS, 207, 35
Sandford, S. A., Allamandola, L. J., Tielens,
A. G. G. M., Sellgren, K., Tapia, M., & Pendle-
ton, Y. 1991, ApJ, 371, 607
Sandford, S. A., Pendleton, Y. J., & Allamandola,
L. J. 1995, ApJ, 440, 697
Schnaiter, M., Henning, T., Mutschke, H., Kohn,
B., Ehbrecht, M., & Huisken, F. 1999, ApJ, 519,
687
Shenoy, S. S., Whittet, D. C. B., Chiar, J. E.,
Adamson, A. J., Roberge, W. G., & Hassel,
G. E. 2003, ApJ, 591, 962
Siebenmorgen, R., Haas, M., Krugel, E., & Schulz,
B. 2005, A&A, 436, L5
Simpson, J. P., Colgan, S. W. J., Erickson, E. F.,
Hines, D. C., Schultz, A. S. B., & Trammell,
S. R. 2002, ApJ, 574, 95
Smith, C. H., Wright, C. M., Aitken, D. K., Roche,
P. F., & Hough, J. H. 2000, MNRAS, 312, 327
Tinbergen, J. 1996, Astronomical polarimetry
(Cambridge, New York: Cambridge University
Press)
Tomono, D., Doi, Y., Usuda, T., & Nishimura, T.
2001, ApJ, 557, 637
van Bemmel, I. M., & Dullemond, C. P. 2003,
A&A, 404, 1
Watanabe, M., Nagata, T., Sato, S., Nakaya, H.,
& Hough, J. H. 2003, ApJ, 591, 714
Weingartner, J. C., & Draine, B. T. 2001, ApJ,
548, 296
88
Wright, G. S., Bridger, A., Geballe, T. R., &
Pendleton, Y. 1996, in ASSL Vol. 209: New
Extragalactic Perspectives in the New South
Africa
Young, S., Hough, J. H., Axon, D. J., Bailey, J. A.,
& Ward, M. J. 1995, MNRAS, 272, 513
This 2-column preprint was prepared with the AAS LATEX
macros v5.2.
11
Fig. 2. -- Observed and predicted polarization over
the 3.4 µm feature in NGC1068 for an MRN distri-
bution of coated silicate particles with different thick-
nesses of mantle. Also shown are the polarization of
a 0.2 µm grain (fcarb= 0.25), close to the largest size
in the MRN distribution, and the limit on the change
in polarization over the feature based on a model fit
to the observed polarization (see text). The column
density of grains (eq. 6) was chosen to reproduce the
observed polarization around 3.55 µm, a region ex-
pected to be unaffected by the 3.4 µm feature.
Fig.
1. -- Top: L-band polarization spectrum of
NGC 1068. The binned spectrum was obtained by
calculating the standard error of the q and u spec-
tra in 0.010 µm bins and is offset by 3% for clar-
ity. Also shown are the measurements of Bailey et al.
(1988,
square,
offset by 0.01 µm in wavelength). Middle: continuum-
subtracted flux spectrum, showing the peak of the
3.4 µm absorption feature. Bottom: position angle of
polarization, with the measurements of Bailey et al.
(1988) and Lebofsky et al. (1978), as above.
triangles) and Lebofsky et al. (1978,
12
|
0801.0740 | 2 | 0801 | 2008-05-14T19:39:24 | Non-thermal X-rays from the Ophiuchus galaxy cluster and dark matter annihilation | [
"astro-ph",
"hep-ph"
] | We investigate a scenario where the recently discovered non-thermal hard X-ray emission from the Ophiuchus cluster originates from inverse Compton scattering of energetic electrons and positrons produced in weakly interacting dark matter pair annihilations. We show that this scenario can account for both the X-ray and the radio emission, provided the average magnetic field is of the order of 0.1 microGauss. We demonstrate that GLAST will conclusively test the dark matter annihilation hypothesis. Depending on the particle dark matter model, GLAST might even detect the monochromatic line produced by dark matter pair annihilation into two photons. | astro-ph | astro-ph |
Non-thermal X-rays from the Ophiuchus galaxy cluster and dark matter annihilation
Stefano Profumo1, ∗
1Santa Cruz Institute for Particle Physics and Department of Physics,
University of California, Santa Cruz CA 95064
We investigate a scenario where the recently discovered non-thermal hard X-ray emission from the
Ophiuchus cluster originates from inverse Compton scattering of energetic electrons and positrons
produced in weakly interacting dark matter pair annihilations. We show that this scenario can ac-
count for both the X-ray and the radio emission, provided the average magnetic field is of the order
of 0.1 µG. We demonstrate that GLAST will conclusively test the dark matter annihilation hypoth-
esis. Depending on the particle dark matter model, GLAST might even detect the monochromatic
line produced by dark matter pair annihilation into two photons.
PACS numbers: 95.35.+d,, 98.80.Cq, 95.85.Nv, 95.85.Pw
Clusters of galaxies are the largest bound dark matter
(DM) structures in the universe. As such, they are nat-
ural targets for the search for observational signatures
of particle DM [1]. If DM is in the form of weakly in-
teracting massive particles (WIMPs) [2, 3], DM pair an-
nihilations generically produce γ-rays as well as a non-
thermal energetic electron-positron (e±) population. The
latter, in turn, is expected to yield secondary emissions at
soft γ-ray, X-ray and radio frequencies via inverse Comp-
ton scattering, bremsstrahlung and synchrotron radia-
tion, opening up the possibility of a multi-wavelength
approach to particle DM detection [1, 4]. A generic fea-
ture of the broad-band DM annihilation spectrum is a
significant hard X-ray component [1, 5].
Interestingly, the discovery of a non-thermal hard X-
ray emission from the Ophiuchus cluster, detected with
relatively robust statistical significance in a 3 Ms obser-
vation with the IBIS/ISGRI and JEM-X instruments on
board INTEGRAL, was recently reported in Ref. [6]. The
Ophiuchus cluster is a nearby (z ≃ 0.028 [7]) rich cluster
with a high temperature plasma (kT ∼ 10 keV), featur-
ing the second brightest emission in the 2-10 keV band.
In addition to X-ray observations, the steep-spectrum ra-
dio source MSH 17-203 [8] was associated to the Ophi-
uchus cluster [7], indicating the presence of relativistic
electrons. However, the cluster was not detected at γ-
ray frequencies by EGRET [9].
Several clusters are known to host extended radio
emissions [10], suggesting the existence of energetic
non-thermal electrons that radiate at radio frequencies
through synchrotron emission. The same electron pop-
ulation should also produce hard X-rays via inverse-
Compton (IC) scattering off cosmic microwave back-
ground photons. Up to now, however, firm evidence
for extended non-thermal hard X-ray emission in clus-
ters was still missing, with a few controversial [11, 12]
exceptions, including BeppoSAX [13] and INTEGRAL
[14] observations of the Coma cluster, BeppoSAX obser-
vations of Abell 2256 [15] and Chandra observations of
the Perseus cluster [16]. In addition, under the assump-
tion of negligible contamination from obscured AGNs,
[17] reports ∼ 2σ detections, using the BeppoSAX PDS
instrument, of hard X-ray non-thermal components from
a few more merging clusters, including Abell 2142, 2199,
3376, Virgo and the Ophiuchus cluster itself.
While ordinary astrophysical mechanisms,
including
merger shocks, can be invoked to explain the non-thermal
electrons presumably responsible for the observed non-
thermal activity in galaxy clusters, in the present anal-
ysis we propose and investigate a novel scenario where
WIMP annihilations produce, or significantly contribute
to, said non-thermal population responsible for the hard
X-ray detection in the Ophiuchus galaxy cluster.
In a
model independent approach, we determine the param-
eters of the particle DM setups that provide the best
fits to the INTEGRAL X-ray data, and we compute the
resulting multi-wavelength spectra. We then compare
these spectra with the radio data and with the γ-ray
limits and future prospects for the soon-to-be-launched
Gamma-Ray Large Area Telescope (GLAST). The high-
lights of our analysis are: (1) the DM hypothesis will
conclusively be probed with GLAST; (2) the radio emis-
sion can in principle also be fitted with the synchrotron
emission from DM-annihilation-produced e±, as long as
the average magnetic field in the cluster is of the order of
0.1 µG; (3) GLAST might be able detect the monochro-
matic γ-rays produced in direct DM pair annihilation
into two photons.
The flux of e± produced by WIMP pair annihilations
depends on the particle DM setup and on the DM den-
sity distribution. We define a source function Qe(Ee, ~x),
which gives the number of e± per unit time, energy and
volume element produced locally in space, as
Qe(Ee, ~x) = hσvi0 Xf
dN f
e
dEe
(Ee) Bf Npairs(~x).
(1)
In the equation above, hσvi0 is the WIMP annihilation
rate at zero temperature, the sum is over all kinemat-
ically allowed Standard Model annihilation final states
f , each with a branching ratio Bf and an e± distri-
bution dN f
e /dEe, and Npairs(~x) is the number density
i.e.
DM(~x)/(2m2
the number
of WIMP pairs at a given point ~x,
of WIMP particle pairs per volume element squared:
Npairs(~x) = ρ2
DM), where ρDM stands for the
DM density. The particle physics framework sets the
quantity hσvi0, the list of Bf and the mass of the WIMP,
mDM. The latter also determines the energy scale of the
pair annihilation event, and, together with the specific
final state f , the dN f
e /dEe spectral functions, which we
numerically compute with the Monte Carlo code Pythia
[18]. In addition, mDM enters in the determination of the
local number density of WIMP pairs.
Once the source function Qe(Ee, ~x) is determined, the
e± spectrum and density are affected by spatial diffusion
and energy loss processes, usually described -- under the
assumptions of negligible convection and re-acceleration
effects -- by a diffusion-loss equation of the form
]
1
-
s
2
-
m
c
s
n
o
t
o
h
p
[
x
u
l
F
10-1
10-2
10-3
10-4
10-5
10-6
10-7
10-8
10-9
10-10
101
2
]
1
-
s
2
-
10-3
m
c
s
n
o
t
o
h
p
[
x
u
l
F
10-4
10-5
20
30
70
Photon Energy [keV]
50
100
b b, mχ=46 GeV
W+W-, mχ=82 GeV
τ+τ−
, mχ=10 GeV
102
103
104
Photon Energy [keV]
105
EGRET limit
106
∂
∂t
dne
dEe
(Ee, ~x) = ~∇ ·hD(Ee, ~x)~∇
∂Eehb(Ee, ~x)
dne(Ee, ~x)
dEe
∂
dEe
dne(Ee, ~x)
i
i + Qe(Ee, ~x),
+
FIG. 1: The hard X-ray and γ-ray spectrum for three DM par-
ticle models, plus a single-temperature MEKAL model [19] for
the thermal X-ray emission, compared with the INTEGRAL
data [6] and with the EGRET upper limit [9].
(2)
where dne/dEe is the number density of electrons per
unit energy, D is the diffusion coefficient, and
b(Ee, ~x) = bIC + bsyn + bCoul + bbrem
(3)
encodes the various energy loss mechanisms [1].
Knowledge of the distribution of the DM-induced
e± population dne/dEe, of the magnetic field struc-
ture and strength, as well as of the electron, gas and
starlight densities allows one to compute the WIMP-
induced secondary emissions. Specifically, at radio fre-
quencies the DM-induced emission is dominated by the
synchrotron radiation of the relativistic secondary elec-
trons and positrons.
IC scattering of the non-thermal
e± on target CMB and starlight photons gives rise to a
spectrum of photons stretching from below the extreme
ultra-violet up to the soft γ-ray band, peaking in the
X-ray energy band. Non-thermal bremsstrahlung, i.e.
the emission of γ-ray photons in the deflection of the
charged particles by the electrostatic potential of ion-
ized gas, contributes in the soft γ-ray band. Finally, a
hard γ-ray component arises from prompt emission in
WIMP pair annihilations, mostly originating from the
two photon decay of neutral pions, and, at the high en-
ergy end of the spectrum, from internal bremsstrahlung
from charged particle final states. The γ-ray spectrum
extends up to energies equal to the kinematic limit set by
the WIMP mass, Eγ ≤ mDM, and might feature one or
more monochromatic lines associated to two-body anni-
hilation final states where one (or both) of the particles
is a photon. We refer the reader to Ref. [1] for details on
the computation of the multi-wavelength emission from
DM annihilation.
We show in Fig. 1 the photon flux in the hard X-ray
and γ-ray bands for three benchmark DM particle model
accounting for the INTEGRAL data. We also include
a thermal component from the bremsstrahlung emis-
sion of the intra-cluster medium, obtained with a single-
temperature MEKAL model [19] with the abundance
fixed to 0.49 compared to the solar value and kT = 8.5
keV [6]. The thermal component was normalized to pro-
duce the best fit for the INTEGRAL data below 20 keV.
The DM models were instead normalized to obtain the
best global fit to the data above 20 keV. We chose DM
models with Bf = 1 for f = b¯b, W +W −, τ +τ −, i.e. each
model pair annihilating into a single Standard Model fi-
nal state. The three particular final states were selected
for two reasons: (1) the resulting e± spectra dN f
e /dEe
range from the softest (b¯b) to the hardest (τ +τ −) possible
case [1]; (2) the three final states correspond to common
well-defined cases found in supersymmetric DM models
[2]. For instance, in the minimal supergravity scenario
[20] f ≃ b¯b corresponds to the so-called bulk and funnel
regions where the neutralino has a relic abundance com-
patible with the cold DM density, τ +τ − is found in the
coannihilation region (where neutralino pair-annihilation
proceeds predominantly through scalar tau exchange)
and W +W − in the focus point region [21]. This choice of
benchmark models follows here and generalizes the ap-
proach of [1] -- linear combinations of the considered mod-
els produce almost any WIMP multi-wavelength emission
spectrum.
For each final state, we selected the DM particle mass
giving the lowest χ2 in the fit to the INTEGRAL data:
mDM (f = b¯b) = 46 GeV, mDM (f = W +W −) = 82
GeV and mDM (f = τ +τ −) = 10 GeV. Following [22] we
assumed, for the diffusion coefficient, the form
D(Ee) = D0
d2/3
B
B1/3
µ (cid:18) Ee
1 GeV(cid:19) , D0 = 3.1 × 1028 cm2s−1,
where dB ≃ 20 is the minimum scale of uniformity of
the magnetic field in kpc and Bµ is the average mag-
netic field in µG. Notice that, while we take into account
spatial diffusion in the numerical computation, the effect
of changing parameters such as D0 or dB is minimal on
the resulting WIMP multi-wavelength spectrum. In the
computation of the photon flux we neglected the IC from
starlight and assumed an average thermal gas density
nth = 10−3 cm3, relevant for the computation of bCoul
and bbrem in Eq. (3).
In the portion of the spectrum
shown in Fig. 1 the value of the magnetic field is not cru-
cial, and was fixed here, for reference, to B = 0.1 µG.
As shown in the inset, the fit to the INTEGRAL data
improves dramatically with the contribution from DM
annihilation. Also, the emission in the γ-ray band is
compatible with the EGRET limit [9], shown for refer-
ence with a horizontal orange line.
What is the DM pair annihilation cross section re-
quired to reproduce the spectra shown in Fig. 1? To
answer this question we need to integrate the number
density of DM pairs over the line of sight. In turn, this re-
quires knowledge of the DM density profile for the Ophi-
uchus cluster. We follow here the analysis of Ref. [1],
and assume, for reference, the DM density profile ob-
tained in the numerical simulations of Ref. [25] (namely
ρ(r) = ρ0g(r/a), with g(x) = exp[−2(xα − 1)/α] and
α ≃ 0.17) and the DM substructure setup outlined in
[26]. We verified that using other DM profiles changes
our predictions by less than one order of magnitude. We
derived from Ref. [27] a virial mass of ∼ 1.5×1015M⊙h−1
and a virial concentration of ∼ 10. We parametrize the
contribution of substructures via the fraction fs of total
mass in subhalos, and assume a ratio Rs = 5 between
the concentration parameter in subhalos and that in iso-
lated halos with equal mass, Rs ≡ hcsi/hcviri. In Fig. 2
(a) we compute the cross section as a function of the
DM particle mass, for the three benchmark final states,
giving the best fit to the INTEGRAL data. We assume
fs = 0.5 for the lower lines, as suggested by numerical
simulations [28], while we neglect the contribution from
substructures for the upper lines.
While we find rather large values for the pair annihi-
lation cross section compared to the naive expectation
hσvi0 ≃ 3 × 10−26 cm3s−1 motivated by requiring a ther-
mal WIMP relic abundance compatible with the CDM
density through simple scaling arguments, the range we
get is consistent with several examples of supersymmet-
ric DM models (see e.g. Fig. 15 in [1]). Also, the values
we obtain for both hσvi0 and mDM are consistent with
all available particle physics constraints on DM, and are
compatible with WIMPs being in the right density to-
100
1000
3
]
1
-
s
3
m
c
[
0
>
v
σ
<
10-7
10-8
)
γ
γ
>
-
M
D
M
D
(
R
B
10
10-19
10-20
10-21
10-22
10-23
10-24
10-25
(a)
(b)
]
1
-
s
2
-
m
c
.
h
p
[
10-2
10-3
10-4
10
Cross Section
Integrated γ ray flux
(Eγ>100 MeV)
Best Fit Masses
(c)
Monoch. γ ray line
b b, mχ=46 GeV
W+W-, mχ=82 GeV
τ+τ−
, mχ=10 GeV
100
mDM [GeV]
1000
FIG. 2: The preferred pair annihilation cross section (a), the
integrated γ-ray flux above 100 MeV (b) and the minimal
branching ratio for the detection of the monochromatic γ-
ray line at Eγ = mDM (c), as a function of the DM particle
mass.
In panel (a) the upper lines refer to the case of no
substructures, the lower lines refer to the substructure setup
described in the text, and the gray shaded region is ruled out
by EGRET [23] and H.E.S.S. [24] data on the gamma-ray flux
from the galactic center region.
day provided, for instance, non-thermal production or a
modified cosmological expansion rate is assumed at the
time of WIMP freeze-out [29]. In addition, uncertainties
on (1) the dark matter density distribution and (2) the
galactic and extra-galactic gamma-ray background un-
dermine the possibility of ruling out WIMP models with
large pair-annihilation cross sections via gamma-ray data
from the center of the Galaxy [23, 24, 30, 31], nearby
galaxies [32, 33, 34], and from the galactic halo [35]. For
instance, in the upper panel of fig. 2 we shade in gray
the region ruled out by EGRET and H.E.S.S. data from
the galactic center region, assuming a cored dark mat-
ter profile following the analysis of Ref. [30, 31]. Clearly,
gamma-ray data from the galactic center do not rule out
the range of cross sections we find.
Similarly, antimatter constraints not only depend on
the dark matter density across the hole galactic halo, but
are also affected by sizable uncertainties in the diffusion
and propagation of cosmic rays in the Galaxy [35]. Even
for cross sections as large as to account for the EGRET
log10 ( ν [Hz] )
15
10
20
25
b b, mχ=46 GeV
W+W-, mχ=82 GeV
τ+τ−
, mχ=10 GeV
INTEGRAL data
Radio data
T
S
A
L
G
T
h
e
r
m
a
l
B
r
e
m
s
.
10-10 10-8
10-6
10-2
100
10-4
102
Photon Energy [keV]
104
106
108
]
1
-
s
2
-
m
c
g
r
e
[
)
ν
(
S
ν
10-10
10-11
10-12
10-13
10-14
10-15
FIG. 3: The spectral energy distribution for the multi-
wavelength emission of three DM particle models and for the
thermal X-ray emission. We also show the INTEGRAL [6]
and radio [8] data.
data on the galactic gamma-ray emission [36] (hence at
the level of the cross sections in the shaded gray region in
fig. 2), antimatter fluxes from WIMP annihilations are in
general compatible with available data [37]. Finally, both
the search for energetic neutrinos from WIMP annihila-
tion in the Sun or the Earth, and direct WIMP detection,
depend on the WIMP-nucleon scattering cross section, a
quantity which is unrelated to the WIMP pair annihila-
tion. Even in special models, such as supersymmetry, the
range of predictions is so wide to make it impossible to
constrain hσvi0 with neutrino fluxes or direct detection
searches [38, 39].
In conclusion, the WIMP pair anni-
hilation rates we consider here, while larger than what
naively expected, are compatible with WIMP cosmolog-
ical DM production and DM searches.
Panel (b) in Fig. 2 shows the integrated γ-ray flux
above 0.1 GeV for the best fit models as a function of
mDM. In all cases we find that the expected γ-ray flux is
well above the anticipated GLAST LAT integral flux sen-
sitivity, estimated to be around a few ×10−10 cm−2s−1
[40]. Fig. 2 (c) shows the branching ratio hσvitot/hσviγγ
for the monochromatic DM DM → γγ channel needed to
obtain, for the best fit models, the detection of at least
10 photons1 with Eγ = mDM. Notice that the values
shown are independent of the assumed DM profile. While
hσvitot/hσviγγ is entirely model dependent, the range we
obtain is generically consistent with what is expected e.g.
in supersymmetry [41], and especially in next-to-minimal
1 The branching ratio corresponding to a larger photon flux can
be obtained by linearly rescaling the lines in Fig. 2 (c).
10
0.25
0.2
0.15
0.1
0.05
]
G
µ
[
B
]
G
µ
[
B
0.25
0.2
0.15
0.1
0.05
10
4
100
1000
disfavored by radio data
b b
best fit to radio data
0.25
0.2
]
0.15
G
µ
[
B
0.1
0.05
W+ W-
τ+τ-
100
mDM [GeV]
1000
FIG. 4: Preferred values of the cluster magnetic field as a
function of the DM particle mass, for three representative
pair annihilation final states.
supersymmetric extensions of the Standard Model [42].
GLAST can therefore easily detect a sizable number of
monochromatic energetic γ rays, depending on the spe-
cific DM particle model.
We illustrate the whole broad-band spectrum for the
DM annihilation interpretation of the INTEGRAL hard
X-ray emission in Fig. 3. Particularly crucial here is the
question of whether the radio data on the Ophiuchus clus-
ter are compatible with our predictions. This question
depends on the assumed value of the average cluster mag-
netic field B. In the Figure we use values of B giving the
best fit to the radio data, namely B(f = b¯b) = 0.15 µG,
B(f = W +W −) = 0.1 µG and B(f = τ +τ −) = 0.18 µG.
Clearly, a linear superposition of the various final states
can yield a remarkably good fit to the available radio
data. We also indicate the anticipated GLAST sensitiv-
ity, which shows that a sizable flux of γ-rays is expected
if the DM annihilation scenario is indeed correct.
As pointed out in the analysis of Ref. [6], simple IC in-
terpretations of the INTEGRAL data force the estimate
for the magnetic field in the range B ∼ 0.1 − 0.2 µG,
which agrees with the present analysis. While this range
is compatible with the value obtained from the hard X-
ray emission in Coma [13], it is below the estimates ob-
tained, for other clusters, through Faraday rotation mea-
sures [43] (see however the discussion in [12]). We show
in Fig. 4 with black and red lines the values of B respec-
tively giving the best fit to radio data and exceeding by
5-σ the measured radio flux in at least one bin. While
values of B above the red lines are disfavored by radio
data, they are not strictly ruled out if one accounts, e.g.,
for a radial dependence of the magnetic field [1].
The detection of γ-rays from cosmic rays in clusters
of galaxies is generically expected to be possible with
GLAST [12, 44]. Specifically, the analysis of Ref. [45] in-
dicates that the Ophiuchus cluster will be detectable by
GLAST provided the ratio of the energy density of cos-
mic rays to that of the thermal gas is larger than 0.3%
to 8.8%, depending on assumptions on the cosmic ray
spectral index and radial distribution. This warrants a
systematic comparison between the expected γ-ray flux
from cosmic rays and from DM in clusters, which is cur-
rently under way [46].
In general, though, the γ-ray
spectra expected from cosmic rays (a power law above
∼ 1 GeV, with a spectral index depending on the pri-
mary cosmic-ray protons spectral index) are significantly
different from what expected from DM annihilation, see
Fig. 1 and 3. If DM annihilation is responsible for most
of the hard X-ray emission in the Ophiuchus cluster,
GLAST will collect such large statistics that discrimina-
tion from a cosmic ray γ-ray emission appears reasonably
feasible.
In summary, we showed that the origin of the non-
thermal particles plausibly responsibly for the recently
firmly discovered non-thermal hard X-ray emission from
the Ophiuchus cluster might be generated by electrons
and positrons produced in WIMP pair annihilations, pro-
vided the rate for the latter is large enough. This sce-
nario is compatible with all observational information on
the cluster, with particle DM production and searches,
and, more importantly, will be thoroughly tested by
GLAST. The future γ-ray telescope might even detect
the monochromatic two-photon emission provided the
particle model has a large enough branching ratio in that
channel. Other galaxy clusters exhibiting non-thermal
activity [17] will also be outstanding sites for GLAST to
look for signatures of WIMP DM pair annihilation [46].
Finally, radio data from the Ophiuchus galaxy cluster can
also be accounted for in the DM annihilation scenario, as
long as the average magnetic field in the Ophiuchus clus-
ter is below ≈ 0.2 µG.
The author wishes to thank Tesla Jeltema for several
insightful comments and for help with the computation
of the thermal X-ray emission.
∗ Electronic address: [email protected]
[1] S. Colafrancesco, S. Profumo and P. Ullio, Astron. As-
trophys. 455 (2006) 21 [arXiv:astro-ph/0507575].
[2] G. Jungman, M. Kamionkowski, and K. Griest, Phys.
Rept. 267, 195 (1996); G. Bertone, D. Hooper, and
J. Silk, Phys. Rept. 405, 279 (2005).
[3] D. Hooper and S. Profumo, Phys. Rept. 453 (2007) 29
[arXiv:hep-ph/0701197].
5
[4] S. Colafrancesco, S. Profumo and P. Ullio, Phys. Rev. D
75 (2007) 023513 [arXiv:astro-ph/0607073].
[5] T. E. Jeltema and S. Profumo, "Searching for Dark Mat-
ter with X-ray Observations of Local Dwarf Galaxies"
arXiv:0805.1054 [astro-ph].
[6] D. Eckert, N. Produit, S. Paltani, A. Neronov and
T. L. Courvoisier, "INTEGRAL discovery of non-
thermal hard X-ray emission from the Ophiuchus cluster"
arXiv:0712.2326 [astro-ph].
[7] M. D. Johnston et al., Ap.J 245 (1981) 799.
[8] B. Y. Mills, O. B. Slee, E. R. Hill, Australian J. Phys., 13,
676 (1960); O. B. Slee, Australian J. Phys., Ap. Suppl.,
43, 1 (1977); O. B. Slee and C. S. Higgins, Australian
J. Phys., Ap. Suppl., 36, 60 (1975); J. R. Ehman, J. R.
Dixon and J. D. Kraus, Ap. J. 75 (1970) 351; B. B. Jones
and E. A. Finley, Australian J. Phys., 27, 687 (1974);
[9] O. Reimer, M. Pohl, P. Sreekumar and J. R. Mattox,
Astrophys. J. 588 (2003) 155 [arXiv:astro-ph/0301362].
[10] L. Feretti and G. Giovannini, arXiv:astro-ph/0703494.
[11] M. Rossetti and S. Molendi, Astron. Astrophys. 414, L41
(2004) [arXiv:astro-ph/0312447].
[12] C. Pfrommer, arXiv:0707.1693 [astro-ph].
[13] R. Fusco-Femiano et al., Astrophys. J. 513, L21 (1999)
[arXiv:astro-ph/9901018].
[14] M. Renaud, G. Belanger, J. Paul, F. Lebrun and R. Ter-
rier, arXiv:astro-ph/0606114.
[15] R. Fusco-Femiano, R. Landi and M. Orlandini, Astro-
phys. J. 624, L69 (2005) [arXiv:astro-ph/0504147].
[16] J. S. Sanders, A. C. Fabian, S. W. Allen and
R. W. Schmidt, Mon. Not. Roy. Astron. Soc. 349 (2004)
952 [arXiv:astro-ph/0311502]; J. S. Sanders, A. C. Fabian
and R. J. H. Dunn, Mon. Not. Roy. Astron. Soc. 360
(2005) 133 [arXiv:astro-ph/0503318].
[17] J. Nevalainen, T. Oosterbroek, M. Bonamente and
S. Colafrancesco, Astrophys. J. 608 (2004) 166
[arXiv:astro-ph/0311142].
[18] T. Sjostrand, S. Mrenna and P. Skands, arXiv:0710.3820
[hep-ph].
[19] J. S. Kaastra and R. Mewe, in Atomic Data Needs for
X-ray Astronomy, p. 161, (2000), ed. by M. A. Bautista,
T. R. Kallman, and A. K. Pradhan.
[20] See e.g. H. P. Nilles, Phys. Rept. 110 (1984) 1.
[21] H. Baer
020
JHEP 0510,
et
al.,
(2005)
[arXiv:hep-ph/0507282].
[22] S. Colafrancesco and P. Blasi, Astropart. Phys. 9 (1998)
227 [arXiv:astro-ph/9804262].
[23] H. A. Mayer-Hasselwander et al., Astron. Astrophys. 335
(1998) 161.
[24] F. Aharonian et al.
[H.E.S.S. Collaboration], Phys.
Rev. Lett. 97, 221102 (2006) [Erratum-ibid. 97, 249901
(2006)] [arXiv:astro-ph/0610509].
[25] J. F. Navarro et al., Mon. Not. Roy. Astron. Soc. 349
(2004) 1039 [arXiv:astro-ph/0311231].
[26] J. S. Bullock et al., Mon. Not. Roy. Astron. Soc. 321,
559 (2001) [arXiv:astro-ph/9908159].
[27] T. H. Reiprich and H. Boringer, Ap.J. 567 (2002) 716.
[28] J. Diemand, M. Zemp, B. Moore, J. Stadel and M. Car-
ollo, Mon. Not. Roy. Astron. Soc. 364 (2005) 665
[arXiv:astro-ph/0504215].
[29] See e.g. S. Profumo and P. Ullio, JCAP 0311 (2003)
006 [arXiv:hep-ph/0309220] and R. Catena et al.,
arXiv:0712.3173 [hep-ph].
[30] A. Cesarini, F. Fucito, A. Lionetto, A. Morselli
267 (2004)
and P. Ullio, Astropart. Phys. 21,
6
[arXiv:astro-ph/0305075].
[31] S. Profumo, Phys. Rev. D 72,
103521
(2005)
[arXiv:astro-ph/0508628].
B. Sadoulet, Phys. Rev. Lett. 74 (1995)
[arXiv:hep-ph/9412213].
5174
[39] S. Profumo and C. E. Yaguna, Phys. Rev. D 70, 095004
[32] N. W. Evans, F. Ferrer and S. Sarkar, Phys. Rev. D 69,
(2004) [arXiv:hep-ph/0407036].
123501 (2004) [arXiv:astro-ph/0311145].
[33] L. Bergstrom and D. Hooper, Phys. Rev. D 73, 063510
[40] http://www-glast.slac.stanford.edu/
[41] L. Bergstrom and P. Ullio, Nucl. Phys. B 504 (1997) 27
(2006) [arXiv:hep-ph/0512317].
[arXiv:hep-ph/9706232].
[34] S. Profumo and M. Kamionkowski, JCAP 0603, 003
[42] F. Ferrer, L. M. Krauss and S. Profumo, Phys. Rev. D
(2006) [arXiv:astro-ph/0601249].
[35] I. V. Moskalenko, S. W. Digel, T. A. Porter, O. Reimer
and A. W. Strong, Nucl. Phys. Proc. Suppl. 173 (2007)
44 [arXiv:astro-ph/0609768].
[36] W. de Boer, C. Sander, V. Zhukov, A. V. Gladyshev
and D. I. Kazakov, Astron. Astrophys. 444 (2005) 51
[arXiv:astro-ph/0508617].
[37] I. Gebauer, arXiv:0710.4966 [astro-ph].
[38] M. Kamionkowski, K. Griest, G. Jungman and
74 (2006) 115007 [arXiv:hep-ph/0609257].
[43] K.-T. Kim, P. P. Kronberg and P. C. Tribble, Ap.J. 379
(1991) 80.
[44] P. Blasi, S. Gabici and G. Brunetti, Int. J. Mod. Phys.
A 22 (2007) 681 [arXiv:astro-ph/0701545].
[45] S. Ando and D. Nagai, arXiv:0705.2588 [astro-ph].
[46] J. Kehayias, T. Jeltema and S. Profumo, in preparation.
|
astro-ph/0311239 | 1 | 0311 | 2003-11-11T01:01:38 | Rest-frame optical continua of L ~ L*, z>3 quasars: probing the faint end of the high z quasar luminosity function | [
"astro-ph"
] | Near-IR photometry for 20 radio-loud z>3 quasars, 16 of which are radio- selected, are presented. These data sample the rest-frame optical/UV continuum, which is commonly interpreted as emission from an accretion disk. In a previous study, we compared the rest-frame optical/UV continuum shapes of 15 optically bright (V<17.5) z>3 quasars with those of 27 low redshift (z~0.1) ones that were matched to the high redshift sample in evolved luminosity (i.e. having luminosities ranging from 1-7 times the characteristic luminosity, L*, where L*~(1+z)^{~3}) to look for signs of evolution in the central engines. We found the continuum shapes at z~0.1 and z>3 similar, consistent with no significant change in the ratio mdot/M, where mdot is the accretion rate with respect to the Eddington rate and M is the black hole mass. This study expands our earlier high redshift sample to lower luminosity, away from extreme objects and towards a luminosity overlap with lower redshift samples. The distribution of rest-frame optical/UV continuum shapes for this fainter sample is broader, extending further to the red than that of the brighter z>3 one. Three quasars from this fainter sample, two radio-selected and one optically-selected, have optical continuum slopes alpha<-1 (F_{nu}~nu^{alpha}). The optically-selected one, LBQS0056+0125, appears to be reddened by dust along the line of sight or in the host galaxy, whereas the radio-selected ones, PKS2215+02 and TXS2358+189, could derive their red continua from the contribution of a relatively strong synchrotron component to the rest-frame optical. These objects may represent a bridge to a population of very red high redshift quasars to which ongoing or future near-IR, optical and deep X-ray surveys will be sensitive. | astro-ph | astro-ph |
Mon. Not. R. Astron. Soc. 000, 000 -- 000 (0000)
Printed 8 March 2018
(MN LATEX style file v2.2)
Rest-frame optical continua of L ∼ L∗, z > 3 quasars:
probing the faint end of the high z quasar luminosity
function
Olga P. Kuhn
Joint Astronomy Centre, 660 N. A'ohoku Place, Hilo, HI, 96720
email: [email protected]
8 March 2018
ABSTRACT
Near-IR photometry for 20 radio-loud z > 3 quasars, 16 of which are radio-selected,
are presented. These data sample the rest-frame optical/UV continuum, which is com-
monly interpreted as emission from an accretion disk. In a previous study, we compared
the rest-frame optical/UV continuum shapes of 15 optically bright (V< 17.5) z > 3
quasars with those of 27 low redshift (z ∼ 0.1) ones that were matched to the high
redshift sample in evolved luminosity (i.e. having luminosities ranging from 1-7 times
∼ (1 + z)∼3) to look for signs of evolution
the characteristic luminosity, L∗, where L∗
in the central engines. We found the continuum shapes at z ∼ 0.1 and z > 3 similar,
consistent with no significant change in the ratio m/M, where m is the accretion rate
with respect to the Eddington rate and M is the black hole mass. This study expands
our earlier high redshift sample to lower luminosity, away from extreme objects and to-
wards a luminosity overlap with lower redshift samples. The distribution of rest-frame
optical/UV continuum shapes for this fainter sample is broader, extending further to
the red than that of the brighter z > 3 one. Three quasars from this fainter sample,
two radio-selected and one optically-selected, have optical continuum slopes α < −1
(Fν ∼ ν α). The optically-selected one, LBQS0056+0125, appears to be reddened by
dust along the line of sight or in the host galaxy, whereas the radio-selected ones,
PKS2215+02 and TXS2358+189, could derive their red continua from the contribu-
tion of a relatively strong synchrotron component to the rest-frame optical. These
objects may represent a bridge to a population of very red high redshift quasars to
which ongoing or future near-IR, optical and deep X-ray surveys will be sensitive.
Key words: quasars: general -- galaxies: evolution -- galaxies: high-redshift
1
INTRODUCTION
Within the past decade, there has been significant progress
towards characterizing the evolution of the quasar luminos-
ity function. Up to redshifts z < 2, it is well described as
luminosity evolution, with the characteristic luminosity L∗
increasing by a factor 40 − 50 from z = 0 to z = 2.3 (Boyle
et al. 2000). Quasar activity peaks around z∼ 2.5 − 3 (e.g.
Schmidt, Schneider & Gunn 1991) but from there to z > 4,
the space density drops by a factor ∼ 2 − 20 (e.g. Jarvis &
Rawlings 2000, Fan et al. 2001a, Vigotti et al. 2003). This
high redshift turnover is identified with the epoch of quasar
formation (Warren, Hewett & Osmer 1994), but the cumu-
lative effects of dust along the line of sight may also play a
role, the significance of which is not yet well determined.
Despite our improved knowledge of the statistical evo-
lution of quasars, a fundamental question, 'how do individ-
ual quasars form and evolve?', remains unanswered. Models
which combine theories of structure formation with those of
energy generation in quasars can reproduce the evolution of
the quasar luminosity function (e.g. Siemiginowska & Elvis
1997; Haehnelt, Natarajan & Rees 1998; Haiman & Menou
2000; Kauffman & Haehnelt 2000; Hatziminaoglou, Siemigi-
nowska & Elvis 2001), but these differ in details so do not
constrain the evolution of the central engine.
The optical/UV continua and emission lines originate
within the central parsec and reflect conditions in the cen-
tral engine. At low redshift, the emission line correlations
involved in Boroson & Green's (1992) eigenvectors 1 and 2
have been well studied, and within the past couple of years
in particular, the physical drivers of these have become bet-
ter understood, enabling them to be used as tracers of the
central mass and accretion rate (Marziani et al. 2001; Boro-
son 2002). The optical/UV continuum is often attributed to
emission from an accretion disk. Model fits can be used to
2 O. P. Kuhn
estimate the mass and accretion rate, though factors such
as intrinsic reddening or additional components, such as
synchrotron emission which may be important in the flat-
spectrum radio quasars (Serjeant & Rawlings 1997; Fran-
cis, Whiting & Webster 2000; Whiting, Webster & Francis
2001), complicate matters. Use of multiple datasets: emis-
sion line, continuum and polarization; should yield the best
estimates of central engine parameters, however, these are
feasible only for the brighter quasars.
In a previous study, we compared the rest-frame opti-
cal/UV continua of 15 z > 3 quasars with those of 27 z ∼ 0.1
quasars matched to the high redshift ones in evolved lumi-
nosity (Kuhn et al. 2001, hereafter Paper I). We found no
evidence for significant evolution of the continuum shape.
The high redshift sample in our previous study was se-
lected nearly a decade ago and was limited. We had aimed
to gather spectroscopic as well as photometric data. This
meant that the objects had to be bright and relatively few.
The resulting set which consisted of the optically brightest
z > 3 quasars was nearly evenly split between radio-loud
and radio-quiet quasars, and several of these were consid-
ered the most luminous objects in the universe when they
were discovered (e.g. HS 1946+7658; Hagen et al. 1992). Fi-
nally, our low-z/high-z sample matching criteria imposed a
strong redshift-luminosity degeneracy.
To move the high-redshift sample away from the most
extremely luminous objects, towards some overlap with the
low-redshift one (and other low or intermediate redshift sam-
ples drawn from the literature), and increase its size, a set
of 20 fainter z > 3 quasars was selected. For the previous
'bright z > 3' sample, a comparison of the optical/UV spec-
tral indices measured using both photometric and spectro-
scopic data with those measured from the photometry alone
shows that photometry alone is adequate at least for sin-
gle power law fits to continuum shape. This paper describes
the observations and distribution of optical/UV continua for
this new 'faint z > 3' sample. Section 2 discusses the sample
selection and section 3, the data acquisition and reduction.
In section 4, the rest-frame optical/UV continua are pre-
sented and, in section 5, compared with those of the 'bright
z > 3' sample. Section 6 discusses the reddest quasars with
α < −1, and conclusions are drawn in section 7. Except
in Figure 1, a Friedmann cosmology with Ho = 75 km s−1
Mpc−1, qo = 0.1 and Λ = 0 is used.
2 SAMPLE
2.1 Selection
To minimize optical selection effects, the 'faint z > 3' sample
was limited to radio-loud quasars. The targets were drawn
from a listing of all objects with z > 3 that were cat-
alogued in NED1 as both quasars and radio sources. To
be accessible from UKIRT and from La Palma (where we
sought optical data), a declination limit: −20◦ < δ < 60◦;
was imposed. A cross-check with the latest edition of the
V´eron-Cetty & V´eron catalogue (2000; VCV9) revealed a
few new quasars; and vice-versa, a few NED quasars were
1 NED, the NASA/IPAC Extragalactic Database
not in VCV9. Quasars appearing in one or the other cata-
logue were kept. Those noted as lensed were removed, and
a final list of 20 radio-loud quasars accessible during the
allocated nights, i.e. with right ascension between 22h and
08h30m, constituted the target sample, hereafter referred to
as z3f (for z > 3 faint; Table 1).
Of the 20 z3f quasars, 16 were radio-selected and 4 were
originally discovered in optical -- objective prism or grism
-- surveys. The radio-selected quasars tend to have high
radio-to-optical flux ratios (Table 1) and flat radio spec-
tra (Table 4). This is probably a consequence of the sur-
vey frequency (most are from 5GHz and 1415MHz surveys,
so sample still higher rest-frame frequencies) and sensitiv-
ity. Deeper and longer wavelength surveys would be more
effective in selecting steep-spectrum high redshift quasars.
This radio-selected subsample avoids the bias towards blue
optical/UV continua that may result from some optical sur-
veys, but on the other hand may include blazars in which
a synchrotron component contributes to and reddens the
optical/UV continua (Whiting et al. 2001). The 4 optically
selected radio-loud quasars were kept so that their colors
could be compared with those of the radio-selected ones.
2.2 Radio-loudnesses
The sample quasars are classified as radio-loud no matter
which criterion -- rest-frame radio-luminosity or radio-to-
optical flux ratio -- is used. Their rest-frame 1.4GHz lu-
minosities are computed using equation 1, from Stern et
al. (2000), where I have assumed the radio spectral index,
αr = −0.5 (Fν ∼ ν αr ), to be consistent with the assump-
tions made by Gregg et al. (1996) who classify as radio-loud
those quasars with log(L1.4GHz)> 32.5 [erg s−1 Hz−1] (for
Ho = 50, qo = 0.5).
L1.4GHz = 4πd2
L
F1.4GHz,obs
(1 + z)(1+αr )
(1)
In the equation, dL is
the luminosity distance and
F1.4GHz,obs(≡ FN V SS) is the observed flux at 1.4 GHz from
the NRAO-VLA Sky Survey (NVSS; Condon et al. 1998).
The values of F1.4GHz,obs and L1.4GHz are listed in the fourth
and fifth columns of Table 1. To compare the luminosities
L1.4GHz with the radio-quiet/radio-loud threshold of Gregg
et al. (i.e. to convert them to the cosmological parameters
that Gregg et al. adopted), a constant which depends on
redshift but ranges from 0.02 to 0.1 for this sample should
be subtracted. All of the z3f quasars have rest-frame 1.4
GHz luminosities above the radio-loud threshold of Gregg
et al (1996). The radio-to-optical flux ratio, RL, is calcu-
lated as in Bechtold et al. (1994), although the rest-frame
5GHz and 1450A luminosities are computed directly from
the radio data available from NED and the NVSS and from
the near-IR data published here:
RL = log10
L5GHz
L
1450A
(2)
Objects with RL > 1 are considered radio-loud. For the
three objects with only one radio measurement, a radio spec-
tral index, αr = −0.18 is assumed. This is the median αr for
the 16 radio-selected quasars. Values of RL for the sample
quasars are listed in column 6 of Table 1.
Rest-frame optical continua of L ∼ L∗ z > 3 quasars
3
2.3 Radio spectral indices
Most of the quasars were classified as flat-spectrum radio-
quasars. The spectra of a couple which had more exten-
sive radio data were determined to be peaked at Giga-Hz or
higher frequencies (as noted in Table 1). One, B30749+420,
was classified as a compact symmetric object (CSO). To
measure their radio spectral indices, I fit a power law to the
radio data taken from NED. The results confirm that none of
the quasars with greater than one radio-flux measurement,
i.e. the 16 radio-selected z3f quasars plus Q2311 − 036, is a
steep-spectrum quasar; all have αr > −0.5 (column 6 of Ta-
ble 4). Radio data at a greater range of wavelengths would
be needed to determine whether any more of these are Giga-
Hz peaked or CSOs. The tendency to find GPS quasars at
high redshift and their interpretation as young radio sources
(O'Dea 1998) suggests that there may be more within this
high redshift sample.
2.4 z−L coverage
A consideration in selecting the sample was to increase the
luminosity range at high redshift and reduce the redshift-
luminosity degeneracy that was inherent in our earlier study.
Figure 1 plots the redshift-absolute magnitude distribution
of the 'faint z > 3' sample, together with the earlier 'bright
z > 3' (hereafter z3b) and low redshift ones.
Cross-correlating the z > 3 quasars in the VCV9 catalog
with the positions of 2MASS sources in the 2nd incremental
data release2 yields 56 with detections at J, H and Ks. The
redshifts and absolute magnitudes of these 2MASS-detected
quasars are also indicated Figure 1. While not homogeneous,
the sample is large and as such provides a reference for the
other z > 3 samples.
3 OBSERVATIONS & DATA REDUCTION
3.1 Near-IR photometry
Near-infrared photometry was obtained for 18 of the z3f tar-
gets in the fall of 2000 using UFTI (UKIRT Fast-Track Im-
ager, Roche et al. 2002) at the United Kingdom Infrared
Telescope (UKIRT). Data for the remaining 2 were obtained
in the spring of 2001 using IRCAM at UKIRT. Deeper im-
ages of the field of PC0027+0525 were obtained in Au-
gust 2003, and magnitudes measured from these rather than
from the November 2000 ones were used in the anaylsis.
Table 2 lists the dates of the observations and results. All
20 quasars were observed through the J, H and K Mauna
Kea Observatory NIR filters (MKO-NIR; Tokunaga, Simons
& Vacca 2002), and 8 were also observed through the Z
(λcen ∼ 0.95µm) filter. Most of the observations were made
on 17 and 18 November (UT), though a few objects were ob-
served in September and October to be quasi-simultaneous
with optical data which had been obtained in late August
(Appendix A). All 20 were re-observed approximately one
year later (October 2001 - July 2002) at K to provide a rough
indication of their variability (see section 3.2).
All of the near-IR observations were carried out in a
2 http://www.ipac.caltech.edu/2mass/releases/second
-32
-30
-28
-26
-24
-22
0
1
2
3
4
5
z
Figure 1. Redshift-absolute magnitude distribution of three
z > 3 datasets considered here: (1) 17 of the 20 target (z3f)
quasars that are listed in, so have absolute magnitudes from,
V´eron-Cetty & V´eron (2001, VCV10; filled boxes); (2) the 15
previously studied bright z > 3 quasars (open triangles); and (3)
56 quasars with JHKs from 2MASS (crosses). For reference, the
redshift-absolute magnitude distribution of the 27 'Atlas' (Elvis
et al. 1994a) quasars used as a comparison sample in Paper I
are also plotted (filled triangles). All the absolute magnitudes are
from VCV10, which assume a k-correction based on a Fν ∼ ν−0.3
approximation to the quasar continuum. The solid curve traces
characteristic luminosity as a function of redshift: L∗ ∼ (1+z)3.34
up to z ∼ 2 (Boyle, Shanks & Peterson 1988); the dashed line
shows a later formulation which takes into account the slowing
of evolution at z ∼ 2 (note that in both cases, the curve is only
an extrapolation for z >∼ 2). Dotted lines mark tracks of con-
stant apparent magnitude. A Friedmann cosmology with Ho=50
km s−1 Mpc−1 and qo = 0 is used for this plot, since that is
what VCV10 adopt when computing absolute magnitudes and
it is one of the cosmologies for which Boyle et al. (1988, 1991)
parameterize luminosity evolution.
grid (or jitter) pattern. At J, H and K, this was either a 5 or
9-point jitter, depending on source brightness, of individual
60-second observations. For the faintest sources, the 9-point
jitter was repeated 2 or 3 times. At Z, longer integrations
were needed to be background limited, and most of the Z
band observations were carried out as 5 or 3-point jitters of
120 or 250-second integrations.
Standard stars from the set of UKIRT faint standards
(Hawarden et al. 2001) were observed approximately every
hour for flux calibration. While nights in September and
October were photometric, the two in November had some
cloud, during the second part of Nov 17 and first part of
Nov 18. Frequent observations of standards enabled the data
from the beginning of Nov 17 to be salvaged, and the two
quasars observed through cirrus at the end of that night
were re-observed with IRCAM in April 2001. Since there
was cirrus at the start of Nov 18, it was treated as non-
photometric and repeat observations to build up signal-to-
noise on several faint objects were made. Later on, skies
cleared and standard star observations looked reliable, so Z
band observations were made of several of the targets.
Reduction of the near-IR data was straightforward up
to the point of doing aperture photometry. The ORAC-
4 O. P. Kuhn
Table 1. Faint z > 3 radio-loud sample
Quasara
PKS 2215+02
MG3 J222537+2040
MG3 J225155+2217
Q2311-036
TXS 2342+342
TXS 2358+189
MG1 J000655+1416
PC0027+0525
LBQS0056+0125
UM672
MG3 J015105+2516
PKS 0201+113
MG3 J023222+2318
MG1J024614+1823
PKS 0335-122
PKS 0336-017
MG2 J062425+3855
Q0642+449
B3 0749+426
PMNJ0833+0959
18.4
20.5
−
mb
21.5h
R18.6
20.2
za
3.55
3.56
3.668
3.034 R18.8
3.053
3.10
3.20
4.099 R21.49
3.149
3.119
3.10
3.61
3.42
3.59
3.442 R20.2
3.197 R18.8
3.469 R18.7
18.49
3.396
R18.3
3.59
3.75
−
18.6
18
R19.8
R19.5
R19.9
−
Fc
N V SS
781
221
190
79
155
266
184
5
6
46
200
781
461
217
476
593
809
453
711
122
1.4GHz
Ld
35.6
35.1
35.1
34.5
34.8
35.0
34.9
33.6
33.4
34.3
34.9
35.7
35.4
35.1
35.4
35.4
35.6
35.4
35.6
34.9
RLe
4.6
3.0
3.9
2.9
2.9
4.5
2.9
3.1
1.9
2.5
3.9
3.9
4.1
3.5
4.5
3.7
4.2
3.8
3.2
3.1
notesg
FSRQ
FSRQ
FSRQ
FSRQ,ALS(1)
FSRQ,ALS(2)
FSRQ,ALS(2)
FSRQ
FSRQ
GPS(5),FSRQ(6),ALS(2)
FSRQ
FSRQ
FSRQ, ALS(4)
radio surveysf
87GB, TXS, WB92, PHJFS
87GB, WB92
87GB, WB92
PMN, opt(UM 659)
87GB, WB92
87GB, WB92
87GB, WB92, TXS
opt(SSG)
opt(SSG,LBQS)
opt(UM)
87GB, WB92
87GB, WB92, TXS
87GB, WB92, TXS
87GB, WB92
PMN, TXS
PMN, 87GB, WB92, TXS, PHJFS GPS(7)
B3, 87GB, WB92, S4, TXS
B3, 87GB, WB92, S4, TXS
87GB, WB92, TXS
87GB, WB92
FSRQ
HFP(8)
CSO(9)
FSRQ
ALS(3)
a Name as it appears in NED: coordinates used were from NED and not repeated here. Redshift z from
NED.
b optical magnitude, at V if not indicated as R, from VCV10, except when otherwise noted.
c FN V SS is the 21-cm flux in mJy from the NRAO-VLA Sky Survey (NVSS; Condon et al. 1998).
d Column lists log10(L1.4GHz ): L1.4GHz is the rest-frame luminosity at 1.4GHz in erg s−1 Hz−1.
e RL is the rest-frame radio(5GHz) to optical(1450A) flux ratio, as defined in Bechtold et al. (1994); RL> 1
is radio-loud. The radio and optical rest-frame fluxes are computed directly for each object from power law
fits to the radio and to the near-IR data.
f Radio and/or optical surveys which detected the quasar (indicated in this column and by the prefix
in column 1): PKS: 2.7 GHz, Parkes radio survey (e.g. Shimmins et al. 1966); PHJFS: PKS Half-Jansky
Flat-Spectrum Sample (Drinkwater et al. 1997); MG: 5 GHz MIT Green Bank 5GHz survey (e.g. Bennett
et al. 1986); 87GB: 5 GHz (Gregory & Condon 1991; Becker, White & Edwards 1991); TXS: 365 MHz
(Douglas et al. 1996); WB92: 1.4 GHz (White & Becker 1992); B3: 408 MHz, 3rd Bologna Catalog of Radio
Sources (Ficarra, Grueff & Tomassetti 1985); S4: 5 GHz, Fourth 'Strong' radio source survey (Pauliny-Toth
et al. 1978); PMN: 5 GHz, Parkes-MIT-NRAO radio survey (Griffith et al. 1994).
opt =discovered in an optical survey: LBQS -- objective-prism-selected (LBQS0056+0125 reported by
Chaffee et al. 1991); UM -- Univ. of Michigan, obj-prism selection (McAlpine & Feldman 1982); SSG --
grism survey (Schmidt, Schneider & Gunn 1987, Schneider, Schmidt & Gunn 1991).
g Notes:
ALS = line-of-sight absorption line system: (1) Lyα+metal system at z ∼ 2.7 (Bechtold 1994); (2) Lyα
absorption systems at high redshift (z >∼ 3) reported by White, Kinney & Becker (1993). The one toward
TXS 2358+189 may not be damped, but those toward TXS 2342+342 and PKS 0201+113 are; (3) z ∼ 2.7
damped Lyα system (Pettini et al. 1997); (4) z ∼ 3.178 damped Lyα system (Ellison et al. 2001).
FSRQ = flat spectrum radio quasar;
GPS = Giga-Hz Peaked: (5) O'Dea, Baum & Stanghellini (1991); (6) Stanghellini et al. (1998); (7) Savage
et al (1990); (8) HFP = High Frequency Peaker (Dallacasa et al. 2000); (9) CSO = Compact Symmetric
Object (Peck & Taylor 2000).
h Optical magnitude from NED
DR pipeline 3 (Currie et al. 1999) was used to process the
images: basically it (1) subtracts a dark, (2) applies a bad-
pixel mask, (3) combines the 3, 5 or 9 frames from each group
to make a flat field, masking the central region where the
object should fall, (4) divides this flat into each observation
of the group and finally (5) shifts and averages these to
produce the final mosaic.
Aperture photometry was done using the IRAF 4 task
digiphot.apphot.phot. The difficulties in aperture pho-
tometry arose primarily because of greater-than-average and
variable seeing. For the standard stars, a 45-pixel radius
aperture was adopted (diameter 8.′′2). The curves of growth
for many of the quasars became noisy at radii larger than
10 or 15 pixels, however, so an aperture correction was nec-
essary. For consistency, all of the quasar data were cor-
3 ORAC-DR is an online Data Reduction Pipeline developed at
the Joint Astronomy Centre by Frossie Economou and Tim Jen-
ness in collaboration with the UK Astronomy Technology Centre
as part of the ORAC project.
4 IRAF is distributed by the National Optical Astronomy Obser-
vatories, which are operated by the Association of Universities for
Research in Astronomy, Inc., under cooperative agreement with
the National Science Foundation.
Rest-frame optical continua of L ∼ L∗ z > 3 quasars
5
Table 2. Near-IR observations
Quasar
PKS2215+02
Za
19.63 ± 0.05
...
Ja
18.64 ± 0.04
...
Ha
18.05 ± 0.03
...
MG3J222537+2040
17.69 ± 0.04
...
17.08 ± 0.03
...
16.70 ± 0.04
...
MG3J225155+2217
Q2311-036
...
...
...
18.02 ± 0.03
...
Ka
17.06 ± 0.02
17.10 ± 0.02
16.13 ± 0.02
16.07 ± 0.03
18.16 ± 0.05
17.98 ± 0.04
16.18 ± 0.04
...
16.23 ± 0.11b
16.34 ± 0.01
16.34 ± 0.02
17.89 ± 0.04
...
18.44 ± 0.06
16.49 ± 0.04
16.48 ± 0.03
19.10 ± 0.10
19.05 ± 0.04
15.66 ± 0.04
15.66 ± 0.06
15.75 ± 0.04
15.74 ± 0.11b
17.78 ± 0.04
17.78 ± 0.04
16.80 ± 0.03
15.65 ± 0.03
17.26 ± 0.03
17.40 ± 0.03
16.30 ± 0.05
16.24 ± 0.06
17.56 ± 0.03
17.75 ± 0.04
16.65 ± 0.03
...
16.52 ± 0.05
16.38 ± 0.03
...
16.81 ± 0.03
15.70 ± 0.03
...
15.75 ± 0.06
...
15.53 ± 0.01d
15.47 ± 0.03
...
16.62 ± 0.02d
16.61 ± 0.02
UT date
2000 Sep 17
2001 Oct 6
2000 Sep 17
2001 Oct 6
2000 Nov 18
2001 Oct 6
2000 Nov 17
2000 Nov 18
2002 Jul 17
2000 Oct 13
2001 Oct 6
2000 Nov 18
2000 Oct 13
2001 Oct 6
2000 Nov 17
2001 Oct 8
2000 Nov 18
2003 Aug 08
2000 Nov 17
2001 Oct 8
2000 Nov 17
2002 Jul 17
2000 Nov 17
2001 Oct 6
2000 Nov 17
2001 Oct 6
2000 Sep 17
2002 Feb 16
2000 Nov 17
2001 Oct 6
2000 Nov 18
2002 Feb 16
2000 Nov 18
2000 Oct 12
2001 Oct 5
2000 Nov 17
2000 Nov 18
2001 Oct 5
2000 Nov 17
2000 Nov 18
2001 Oct 5
2000 Nov 18
2001 Apr 15
2002 Feb 16
2000 Nov 18
2001 Apr 15
2002 Feb 16
comments
Z-band data non-photometric
∆K = +0.04
∆K = −0.06
∆K = −0.18
∆K = +0.05
∆K = 0
scaled to Nov 17
∆K = +0.55
∆K ∼ 0
scaled to Nov 17
c
∆K = 0
∆K ∼ 0
∆K = 0
∆K = −1.15
∆K = +0.14
∆K = −0.06
JHK scaled to Sep 17 & Nov 17
∆K = +0.19
∆K = −0.13
∆K = +0.43
∆K = +0.05
IRCAM
∆K = −0.06
IRCAM
∆K ∼ 0
19.34 ± 0.07
...
18.67 ± 0.06
...
17.48 ± 0.04
...
16.76 ± 0.03
...
...
...
17.26 ± 0.02
...
16.76 ± 0.02
...
19.70 ± 0.06
19.69 ± 0.06
...
18.76 ± 0.04
...
...
17.16 ± 0.03
...
16.98 ± 0.04
...
20.55 ± 0.09
20.46 ± 0.04
17.24 ± 0.05
...
19.81 ± 0.08
...
16.54 ± 0.05
...
16.99 ± 0.04
...
16.41 ± 0.04
...
19.03 ± 0.04
...
18.72 ± 0.05
...
17.95 ± 0.03
...
17.54 ± 0.03
...
18.47 ± 0.03
...
17.94 ± 0.04
...
17.86 ± 0.03
...
17.33 ± 0.03
...
19.50 ± 0.03
...
19.04 ± 0.05
...
18.50 ± 0.03
...
18.01 ± 0.04
...
...
...
18.20 ± 0.03
...
...
17.73 ± 0.06
...
17.16 ± 0.05
...
...
18.24 ± 0.03
...
17.54 ± 0.04
17.56 ± 0.04
...
17.21 ± 0.03
...
...
17.73 ± 0.03
...
17.19 ± 0.02
...
...
...
17.18 ± 0.04
...
16.56 ± 0.05
...
...
...
...
...
16.48 ± 0.02d
16.04 ± 0.01d
...
...
...
...
17.69 ± 0.02d
17.08 ± 0.02d
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
TXS2342+342
TXS2358+189
MG1J000655+1416
PC0027+0525
LBQS0056+0125
UM672
MG3J015105+2516
PKS0201+113
MG3J023222+2318
MG1J024614+1823
PKS0335-122
PKS0336-017
MG2J062425+3855
Q0642+449
B30749+426
PMNJ0833+0959
a All magnitudes except those obtained with IRCAM and those obtained 2002 July 17 were measured within an aperture of
radius 10 pixels (1.′′8 diameter) and an aperture correction applied to determine the magnitude within the 45 pixel radius
(8.′′19 diameter) aperture used for the standard stars. The quasar magnitudes obtained using UFTI on 2002 July 17 were
measured within apertures of radius, R=20(3.′′6 diameter), and corrected to the R=30(5.′′ 4 diameter) aperture used in
determining the zeropoint. Magnitudes are determined assuming only an extinction correction; a synthetic color correction
to account for the different spectral shapes of Vega and the quasar is incorporated in the zero-magnitude flux.
b 2002 Jul 17 was not strictly photometric and there were not enough stars for relative photometry, but the data appeared
to be of sufficient quality to show that neither of these quasars varied appreciably.
c The quasar magnitudes obtained for PC0027+0525 on 2003 Aug 08 (in sub-half-arcsecond seeing) were measured within
apertures of radius, R=10(1.′′8 diameter), and corrected to the R=30(5.′′4 diameter) aperture used in determining the
zeropoint. Bad pixels were interpolated over before magnitudes were computed; this affected the J-band magnitude. The
quasar was barely detected on the 2000 Nov 17/18 data, though it was clearly seen on the 2003 Aug 08 images. The Nov
17/18 magnitudes are only listed to show their rough consistency with the 2003 magnitudes.
d JHK magnitudes for these two objects were measured with IRCAM. For these observations, aperture corrections were
made from an R=20(3.′′ 25 diameter) to R=40(6.′′ 5 diameter) aperture.
rected from a small aperture of radius 10 pixels (diameter
1.′′8) to the 45 pixels standard star one. The variable seeing
meant that the aperture correction had to be determined
individually for each mosaic. Many fields had only one star
bright enough to use in determining an aperture correction.
In this case, the difference between its instrumental magni-
tudes measured within the R=10 pixel and R=45 pixel aper-
tures was taken to be the aperture correction, AC, and the
quadrature sum of the errors in these instrumental magni-
tudes was adopted as the error in it. A few fields had several
stars bright enough for aperture corrections. For these, the
aperture correction was determined from the star with the
smoothest curve of growth and the error was taken to be the
standard deviation, σN −1, in the measurements. The aper-
ture corrections determined in this way, for each frame, were
reasonable -- not too scattered and generally following the
seeing variations. On a single night with good seeing (∼ 0.′′5,
2000 October 13), average aperture corrections ranged from
−0.18 ± 0.06 at J to −0.12 ± 0.01 at K, while on a night with
poor seeing (0.′′8 to 2′′, 2000 November 17) they ranged from
−0.52±0.17 at J to −0.41±0.11 at K. The error here reflects
the seeing variations and is smaller for an individual frame.
Zero-points, ZP , and extinction coefficients, kext, were
computed for each night at each filter. The quasar magni-
6 O. P. Kuhn
tudes, mQ, were then determined from their instrumental
magnitudes within the R=10 aperture, m10, as:
mQ = m10 − AC − kext(χ − 1) + ZP
(3)
where χ is the airmass. Errors are quadrature sums of the
errors in the instrumental magnitude, aperture correction
and zero point.
3.2 Variability
Prior to these observations, most of the target quasars had
not been extensively observed; in particular, there is little
information on their variability properties. Section 2.3 con-
firms that most of the radio-selected z3f quasars have flat
radio spectra. The set of flat-spectrum radio-loud quasars
includes blazars, which contain a beamed synchrotron com-
ponent that may contribute significantly to the rest-frame
optical (e.g. Serjeant & Rawlings 1997), thus complicating
the use of the rest-frame optical/UV continuum shape in
probing the central mass and accretion rate. Signatures of
beamed emission include rapid, high-amplitude variability
(e.g. Marcha et al. 1996). Through service observations at
UKIRT, all 20 targets were re-observed at K, approximately
one year after the initial run. Conditions were not always
photometric, and differential photometry was done for most
of the fields by comparison with the earlier images. The re-
sulting magnitudes are listed in Table 2. Three quasars var-
ied by at least 0.4 mag: TXS2358+189, MG2 J062425+3855
and PKS0201+113, which brightened by 1.2 mag at K in 11
months. It is likely that beamed synchrotron emission con-
tributes to the rest-frame optical continua of these three,
however, the data do not rule out the possibility of such a
component in the other quasars, even though they varied
less. It should be possible to select radio-quiet samples from
optical multi-color or grism surveys or deep near-IR sur-
veys that are not biased to blue optical/UV continua and
avoid the complications of radio-optical synchrotron contri-
butions.
4 SPECTRAL ENERGY DISTRIBUTIONS
4.1 Magnitude-to-flux conversion
No color-correction was made to the IR magnitudes listed in
Table 2. However, a set of zero-magnitude fluxes which in-
corporate a synthetic color correction were computed. These
were determined to be the fluxes, at the central wavelengths
of each of the IR bands, of a power law: Fν,P L = n ν −0.46
(representative of the shape of the composite spectrum of
657 quasars from the FBQS, Brotherton et al. 2001); which
has been normalized to have the same broad band magni-
tude as Vega, i.e.
n =
R φλFλ,V egadλ
R φλFλ,P Ldλ
,
(4)
where φλ represents the transmission through the filter, mul-
tiplied by the atmospheric transmission and detector re-
sponse when available.
Color-corrected zero-magnitude fluxes were computed
for each of the broadband filters used in UFTI. For these,
φλ incorporates both the filter and atmospheric transmis-
sion functions. Data from two other instruments (Paper I):
the MMT-IR photometer (Rieke 1984) and OSIRIS (De-
Poy et al. 1993) at CTIO; are used to re-compute rest-
frame continuum shapes for the z3b quasars (section 5.2).
For these, φλ was simply assumed to have a tophat shape
starting and ending at the 50% cut-on and cut-off wave-
lengths of the MKO-NIR filters. For each filter and trans-
mission function (UFTI or 'tophat'), the central wavelengths
and color-corrected zero-magnitude fluxes are listed in the
second and fourth columns of Table 3. For comparison, the
zero-magnitude fluxes for Vega (i.e. the flux of Vega at each
of the central wavelengths), are also included in the Table.
The zero-magnitude fluxes were found to depend little on
the shape of φλ, so for simplicity the color-corrected zero-
magnitude fluxes computed for UFTI were used to convert
all the IR magnitudes to fluxes.
4.2 Galactic extinction
The fluxes were corrected for Galactic extinction, assum-
ing the values of E(B-V) from Schlegel, Finkbeiner & Davis
(1998; as reported in NED), and the near-IR and optical
Galactic reddening law of Rieke, Lebofsky & Low (1985)
and Savage & Mathis (1979).
4.3 Spectral energy distributions
For the 20 z3f quasars, rest-frame radio-to-UV spectral en-
ergy distributions (SEDs) were constructed from the data
presented here and from radio data from the literature
(listed in NED). These are plotted in Figure 2.
5 SPECTRAL INDICES
5.1 α for the faint z > 3 sample
Rest-frame optical/UV spectral indices were determined by
fitting a power law through the rest-frame SEDs between
1500 and 5990 A (log ν=14.7-15.3). This wavelength range
was chosen since it includes J, H and K for z = 3 − 4.1, the
range of redshifts in the sample. The power law fits were
made with and without the Z-band point, which was not
obtained for all the objects, but the difference is minimal.
For the analysis, the fits to just J, H and K were used.
The 1500-5990A spectral indices determined from these fits
(with and without the emission line correction made to the
photometry; see Appendix B) are listed in Table 4 and their
distribution is plotted in Figure 3.
5.2 Bright z > 3 sample
5.2.1 α for the bright z > 3 sample
The original aim of this project was to expand, in number
and luminosity range, the set of 15 bright z > 3 quasars for
which Kuhn et al. (2001) constructed rest-frame optical/UV
SEDs (Paper I, table 1). In this section, the SEDs of the new
z3f quasars are compared with those of the bright z3b ones,
to determine whether increasing the number has reduced the
scatter, and whether there is any trend with luminosity. The
Rest-frame optical continua of L ∼ L∗ z > 3 quasars
7
48
46
44
48
42
46
44
42
48
46
44
48
42
46
44
42
48
46
44
42
10
12
14
10
12
14
Figure 2. Rest-frame radio/optical/UV spectral energy distributions (SEDs) of the 20 'faint z > 3' sample quasars. Radio data (open
boxes) are taken from NED, and the optical (stars; see Appendix A) and IR data (filled boxes for the 'original' JHK data; open triangles
for K band 'variability check' measurements and a few J datapoints not simultaneous with those used in the power law fits; and X's for
the Z band points) are from this paper. The mean SED of ∼ 18 low redshift (z ∼ 0.1) radio-loud quasars (dotted line; Elvis et al. 1994a),
normalized to the quasar's rest-frame 4400A luminosity, is overplotted for comparison. Power law fits to the radio continua and to the
rest-frame optical/UV continua (from JHK only) are plotted (solid line segments -- these extend beyond the frequency ranges of the fits
so that they would be easy to see).
8 O. P. Kuhn
48
46
44
48
42
46
44
42
48
46
44
48
42
46
44
42
48
46
44
42
10
12
14
10
12
14
Figure 2 -- Continued
Rest-frame optical continua of L ∼ L∗ z > 3 quasars
9
Table 3. Zero-magnitude fluxes
Filtera
UFTI Z
UFTI MKO-NIR J
'tophat' J (1.17 − 1.33µm)
UFTI MKO-NIR H
'tophat' H (1.49 − 1.78µm)
UFTI MKO-NIR K
'tophat' K (2.03 − 2.37µm)
λb
0
0.9525
1.25
1.25
1.635
1.635
2.20
2.20
Fλ(Vega)c
5.239
2.982
2.982
1.192
1.192
0.4151
0.4151
Fλ(PL)c
6.921
3.040
3.021
1.208
1.203
0.4192
0.4189
a Name of the filter. I did not have filter curves for the MMT IR photometer or OSIRIS so approximate
the filter by a tophat and give the adopted cut-on and cut-off values in the table.
b The central wavelength (A) of the filters used in UFTI. Since these filter curves are close to a tophat
shape and the cut-on and cut-off values are set mainly by the atmosphere, these central wavelengths are
used for the MMT and OSIRIS data as well.
c Zero-magnitude flux [10−10 erg s−1 cm−2 A−1]: the flux, Fλ(λ0), of Vega or of a ν−0.46 power law
normalized to have the same broad band magnitude as Vega.
Table 4. Rest-frame optical and radio spectral indices
Quasar
αa,b
opt
PKS 2215+02
−1.30 ± 0.05
MG3 J222537+2040 −0.21 ± 0.04
MG3 J225155+2217 −0.31 ± 0.13
−0.65 ± 0.05
Q2311-036
−0.08 ± 0.04
TXS 2342+342
TXS 2358+189
−1.37 ± 0.09
MG1 J000655+1416 +0.52 ± 0.08
−0.77 ± 0.09
PC 0027+0525
−1.05 ± 0.10
LBQS 0056+0125
UM 672
−0.50 ± 0.09
MG3 J015105+2516 −0.44 ± 0.09
PKS 0201+113
−0.23 ± 0.07
MG3 J023222+2318 −0.35 ± 0.07
MG1 J024614+1823 −0.82 ± 0.09
−0.67 ± 0.05
PKS 0335-122
PKS 0336-017
+0.08 ± 0.05
MG2 J062425+3855 −0.40 ± 0.04
−0.79 ± 0.06
Q0642+449
−0.11 ± 0.03
B3 0749+426
PMN J0833+0959
−0.38 ± 0.04
La,b
4400A
46.22
46.62
45.85
46.43
46.42
45.70
46.41
45.70
46.62
46.62
45.79
46.36
46.13
46.49
45.96
46.34
46.48
46.72
46.88
46.50
αa,c
opt(corr)
−1.24 ± 0.05
−0.16 ± 0.04
−0.34 ± 0.13
−0.75 ± 0.05
−0.10 ± 0.04
−1.51 ± 0.09
+0.36 ± 0.08
−0.94 ± 0.09
−1.19 ± 0.10
−0.66 ± 0.09
−0.61 ± 0.09
−0.33 ± 0.07
−0.43 ± 0.07
−0.98 ± 0.09
−0.57 ± 0.05
+0.06 ± 0.05
−0.34 ± 0.04
−0.77 ± 0.06
−0.13 ± 0.03
−0.33 ± 0.04
La,c
4400A
(corr)
46.13
46.55
45.77
46.40
46.38
45.66
46.36
45.58
46.57
46.58
45.75
46.29
46.06
46.43
45.90
46.28
46.41
46.65
46.81
46.44
αd
r
−0.45 ± 0.06
−0.35 ± 0.02
−0.02 ± 0.05
−0.30 ± 0.00
−0.20 ± 0.09
−0.05 ± 0.07
−0.44 ± 0.03
−0.18e
−0.18e
−0.18e
+0.22 ± 0.08
+0.25 ± 0.13
−0.08 ± 0.03
+0.05 ± 0.06
−0.38 ± 0.04
−0.39 ± 0.05
−0.31 ± 0.08
+0.46 ± 0.05
−0.22 ± 0.08
−0.16 ± 0.11
Ld
5GHz
45.15
44.52
44.45
43.92
44.27
44.39
44.43
43.22e
42.92e
43.79e
44.32
45.02
44.89
44.48
44.87
44.94
45.30
44.96
44.93
44.34
a Power law fit through log νrest = 14.7 − 15.3 (λrest ∼ 1500 − 5990A) and monochromatic rest-frame luminosity [erg
s−1] of this power law fit at λrest = 4400A.
b No correction has been made for emission lines.
c Emission line contribution has been estimated and removed (see Appendix B).
d Power law fit through log νrest = 9 − 11.5 and luminosity [erg s−1] of this power law fit at νrest = 5 GHz.
e Only one radio data point was found, from NVSS and at 1415MHz. To estimate the rest-frame luminosity at 5GHz,
the median spectral index determined for the 16 radio-selected quasars, αr = −0.18, was assumed.
first step in making the comparison is to re-determine the
spectral indices of the z3b quasars following the same meth-
ods as described in sections 4 and 5.1 for the z3f sample,
namely from the broad band magnitudes alone. Dropping
the requirement for IR spectroscopy adds 2 quasars to the
z3b sample, SP82 1 and Q1442+101, for which we had ob-
tained near-IR magnitudes (table 6 of Paper I) but failed to
obtain spectra. One quasar which was missing photometry
at J (Q0114-089) had to be dropped. This yields a net total
of 16 quasars in the revised z3b sample. Finally, I observed
the lensed quasar, Q1208+101, on 21 May 2001 at UKIRT
and obtained the following magnitudes (for both compo-
nents): Z=16.60 ± 0.01, J=16.09 ± 0.01, H=15.63 ± 0.01 and
K=15.16 ± 0.01. At J and K, these are 0.15 and 0.26 mag-
nitudes brighter than what Kuhn et al. (2001) measured in
1993; the differences are within range of what is observed
for quasar variability (Giveon et al. 1999; Hook et al. 1994
remark on the anomalously high variability of Q0055 − 269,
another quasar from the z3b sample that was discovered in
a similar manner as Q1208 + 101). In this paper, only these
new data are used for Q1208+101.
For the quasars that had been observed with OSIRIS,
one photometric point was sufficient to normalize the entire
1 − 2.5µm cross-dispersed spectrum, so we had not obtained
photometry at all three bands: J, H and K. To derive broad
band magnitudes from the spectra, I integrated over the
product of these with the curves of filter-plus-atmospheric
transmission (tophat with cut-on and cut-off values as listed
in Table 3), first, to determine the scaling factor needed to
match the integral with the broad band photometry (typ-
10 O. P. Kuhn
6
4
2
0
0.5
0
-0.5
-1
-1
-0.5
0
0.5
Figure 4. Comparison of the optical spectral indices measured
from fits to broad band photometry, with (open triangles and dot-
ted error bars) and without (filled squares and solid error bars)
a correction for emission line contributions, against those deter-
mined from narrow band monochromatic luminosities (Paper I,
Table 15). The line, α(photometry) = α(narrowband), is drawn
for reference.
Figure 4, making the emission line correction (Appendix B)
improves the agreement for some objects but worsens it for
others.
5.3 Comparing the bright and faint
As the histograms in Figure 3 show, the z3f sample includes
redder quasars than the z3b one. Three of the z3f quasars
have spectral indices α < −1 and are "red quasars" accord-
ing to the definition of Gregg et al. (2002). A Kolmogorov-
Smirnov (K-S) test gives a probability of 5% that the dis-
tributions for the z3b and z3f samples are drawn from the
same parent population. A Student's t-test shows that the
mean spectral indices for each sample differ significantly --
the probability of randomly obtaining a separation as large
as measured, P (t), is 1.4%.
A number of explanations can be put forth:
First, the z3f sample contains only radio-loud quasars
whereas the z3b sample is evenly divided between radio-quiet
and radio-loud. Several of the z3f quasars varied, which sug-
gests that they may be blazars in which a non-thermal con-
tribution to the rest-frame optical could redden their rest-
frame optical/UV continua (Whiting et al. 2001). If the two
radio-selected quasars in the red tail are removed from the
z3f sample, the K-S probability that the bright and faint
samples are drawn from the same parent population in-
creases to 13% -- there are no longer sufficient objects in
the tail to render the distributions of spectral indices signif-
icantly different. However the mean spectral indices for each
sample are still significantly different (P (t) = 4%).
Second, dust reddens and extincts the emitted contin-
uum, so in the optical/UV, redder objects would be expected
to be fainter. Richards et al. (2001) note this trend in the
sample of ∼ 2600 SDSS quasars for which they present op-
tical photometry, and it is evident in Figure 5 which plots
spectral index against optical luminosity for the faint and
bright z > 3 samples as well as the 56 z > 3 quasars with
Figure 3. Distributions of rest-frame optical spectral indices for
the 20 z3f quasars (top) and 16 z3b quasars (bottom). Superposed
on these are the distributions for the 16 radio-selected z3f quasars
(top; right slanted thick hash marks) and the 8 radio-loud z3b
ones (bottom; left slanted thick hash marks). Note the red tail
to the z3f sample, with 3 of the quasars, 2 radio-selected and 1
optically-selected, having α < −1.
ically at H), and second, to determine magnitudes in the
bands for which we had not obtained photometry.
The near-IR magnitudes, determined from the OSIRIS
spectra or directly, for all 16 z3b quasars were processed in
the same way as described above for the z3f quasars: they
were converted to fluxes (section 4.1), dereddened by the
Galactic value (section 4.2), corrected for emission line con-
tributions (Appendix B) and blueshifted to the rest-frame
(section 4.3).
5.2.2 Broad-band photometry vs. Narrow band
luminosities
To confirm the validity of using broad band magnitudes to
measure rest-frame optical spectral indices, for 14 of the 15
bright z > 3 quasars studied in Paper I (all but Q0114−089),
the optical spectral indices computed from the photometry
alone (this time using the zero-magnitude fluxes listed in
Table 9 of Paper I) were compared with those determined
by fitting a line through a set of narrow band monochro-
matic luminosities (Paper I, Table 15) which covered ap-
proximately the same rest-frame region as did J, H and K:
i.e. 3023A, 4200A, 4750A and 5100A for z = 3 − 3.5; 2660A,
3023A, 4200A and 4750A for z = 3.6 − 3.8; and 2500A,
2660A, 3023A and 4200A for z > 4. Figure 4 compares the
optical spectral indices determined by these two methods
and shows that, while not in perfect agreement, the dis-
tributions of spectral indices computed by fitting through
narrow-band line-free windows and through only broad band
photometry do not differ significantly. As can be seen in
Rest-frame optical continua of L ∼ L∗ z > 3 quasars
11
Figure 5. Rest-frame optical spectral index, α, vs. luminosity for several samples: the 20 z3f sample quasars (filled boxes); the 16 quasars
in the z3b sample (filled triangles; the radio-loud ones are circled); and 56 z > 3 quasars, 17 radio-loud (open boxes) and 39 radio-quiet
(crosses) from VCV9 for which J, H and Ks magnitudes were published in the 2nd incremental data release of the 2MASS PSC. Starting
at α = −0.46, reddening vectors are plotted for E(B-V) = 0 to 0.2 in steps of 0.05 and assuming both Galactic and Small Magellanic
Cloud extinction laws.
J, H and Ks magnitudes from the 2MASS 2nd incremental
data release.
Finally, the redder colors of the z3f with respect to
the z3b quasars could represent an intrinsic difference in
their central engines. The spectrum emitted by a thin ac-
cretion disk is expected to peak at a frequency deter-
mined by the apparent disk temperature, which for op-
tically thick/geometrically thin disks follows the relation:
log T ∼ 1
− 2.4(cos θ − 1) where m is the accretion
rate relative to the Eddington limit, M is the black hole
mass, and θ is the inclination of the disk axis to the line
of sight (Sun & Malkan 1989, McDowell et al. 1991). The
thermal disk spectra emitted by the fainter set would be sys-
tematically redder if, for example, both sets of quasars had
similar central masses, but the fainter ones were accreting
at lower rates.
4 log m
M
The importance of dust at high redshift (e.g. Andreani,
Franceschini & Granato 1999; Warren, Hewett & Foltz
2000), together with the possibility of a significant non-
thermal component in these radio-loud quasars, leads to
the suspicion that extrinsic factors are responsible for the
on-average redder colors of these z3f quasars. These com-
plicate the use of optical/UV continua to probe the central
engines and point to the need for a similar study of radio-
quiet quasars and for spectroscopic data.
6 RED QUASARS
Three of the sample quasars have α < −1; two of these,
PKS2215+02 and TXS2358+189, are radio-selected, and
one, LBQS 0056+0125, was discovered in a grism survey
(PC0056+0125, Schmidt, Schneider & Gunn 1987).
6.1 LBQS 0056+0125
LBQS0056+0125 has an optical spectral index α = −1.05
(Table 4). In the discovery paper, Schmidt, Schneider &
Gunn (1987) remarked on the presence of narrow absorp-
tion lines redwards of Lyα. Since then several line-of-sight
Lyα absorption line systems have been detected; one at
z = 2.7771 (Schneider, Schmidt & Gunn 1991; Pettini et
al. 1997) is damped (DLA). To redden an average quasar
continuum (the FBQS composite) emitted at the quasar
12 O. P. Kuhn
redshift, z = 3.149, to α = −1.05, requires an extinction
E(B-V) ∼ 0.12 in the DLA, assuming the Small Magellanic
Cloud (SMC) extinction law (Pr´evot et al. 1984, Bouchet et
al. 1985). Pettini et al. (1997) measure the neutral hydro-
gen column for this DLA to be log(NHI ) = 21.11 ± 0.07.
The above estimate for E(B-V) is consistent with this if
the gas-to-dust ratio in the DLA is approximately 2 times
that in the Milky Way(MW; 4.8 × 1021 cm−2 s−1, Savage &
Mathis 1979) -- a factor ∼ 5 − 15 times lower (dustier) than
existing data imply for DLAs (Pei, Fall & Bechtold 1991,
Pettini et al. 1997). The above assumes no intrinsic extinc-
tion. If the SMC law is used also to describe the extinction
in the host galaxy, and the extinction in the DLA is fixed at
E(B-V)=0.027, what is implied by log(NHI ) = 21.11 and a
gas-to-dust ratio 10 times that in the MW, then an emitted
continuum (FBQS composite) would need to undergo an in-
trinsic extinction, E(B-V)∼ 0.1 mag, which corresponds to
an intrinsic column density NHI ∼ 5 × 1020f cm−2 where
f is the ratio of the gas-to-dust ratio in the quasar host to
that in the MW. This is larger than the column densities
measured for low redshift optically selected quasars (Laor
et al. 1997), but consistent with the values inferred for sets
of X-ray selected (Puchnarewicz et al. 1996) and high red-
shift radio-loud quasars (e.g. Elvis et al. 1994b). So the red
color of LBQS0056+0125 can be explained by intrinsic and
line-of-sight extinction, but with gas-to-dust ratios that are
on the low side for the intervening absorbers and quasar host
galaxies. The emitted continuum may itself be redder than
the FBQS composite. Also, the gas-to-dust ratios in DLAs
do show a lot of scatter (Pettini et al. 1997); this might also
explain why other quasars from the z3f sample with high
redshift DLAs (e.g. TXS 2342+342 and PKS0201+113) do
not also appear very red. Warren et al. (2001) have recently
imaged the field of LBQS0056+0125 with NICMOS(NIC2)
on HST and detect two possible candidates for the galaxy
counterpart to the DLA.
PKS2215+02 was detected in a 14.7 ksec observation with
the ROSAT HRI, with an unabsorbed 0.1 − 2.4 keV flux
(observed frame) equal to 3.2×10−13 erg s−1 cm−2 (Siebert
& Brinkmann 1998). Siebert & Brinkmann noted that this
quasar is optically very faint and has the highest X-ray-
to-optical and radio-to-optical flux ratios of the radio-loud
z > 3 quasars they studied. They suggest that it is heav-
ily absorbed although note that in the optical/UV, PKS
2215+02 does not stand out as redder than the other Parkes
Half Jansky quasars that Francis et al. (2001) studied. The
rest-frame optical continuum presented here is slightly red-
der than but not inconsistent with that of Francis et al.
(2001). Ellison et al. (2001) find no evidence for a damped
Lyα system along the line-of-sight to PKS 2215+02, so if
the red color derives from extinction, this must occur at the
quasar. The data presented here, which show the quasar
as one of the strongest radio emitters of the z3f sample
and as having one of the highest radio-to-optical flux ratios
(RL = 4.6; Table 1), together with evidence for variability
at 5GHz (Siebert & Brinkmann 1998) and in the rest-frame
optical, favor a non-thermal contribution rather than ex-
tinction as the cause of the red color. Further monitoring,
or other tests for blazar-like activity, are needed to confirm
this hypothesis.
TXS2358+189 (z = 3.10), like PKS2215+02, has a very
high radio-to-optical flux ratio, RL = 4.5. It has a flat ra-
dio spectrum, but insufficient data to measure radio vari-
ability. At K, TXS2358+189 dimmed by 0.55 magnitudes
from September 2000 to October 2001; this was the second-
largest one-year variation seen in the sample and suggests
the presence of beamed synchrotron emission. White, Kin-
ney & Becker (1993) find evidence in its spectrum for a Lyα
absorption system, though the data do not conclusively show
whether it is damped. In sum, it is probable that the red
optical continua of both TXS2358+189 and PKS2215+02
derive from the blending of a synchrotron component.
6.2 PKS2215+02 and TXS2358+189
6.3 Characteristics of the α < −1 quasars
The other two quasars with α < −1 are radio-selected. They
have some of the the highest radio-to-optical flux ratios of
the z3f sample (see Table 1). PKS2215+02 belongs to the
Parkes Half Jansky sample (Drinkwater et al. 1997), which
was found to contain sufficient numbers of red quasars to
lead to the suggestion, later proved well-founded at least
for low redshift AGN (Cutri et al. 2000), that optical sur-
veys were missing a large fraction of quasars (Webster et
al. 1995). The Parkes Half Jansky sample has been well
studied in the optical and IR (Francis et al. 2000; Whiting
et al. 2001) and X-ray (Siebert et al. 1998). Francis et al.
(2000) measured the following magnitudes for PKS2215+02
in early September 1997: B=21.84 ± 0.31; V=20.42 ± 0.10;
R=20.14 ± 0.12; I=20.00 ± 0.20; J=19.20 ± 0.50; H=18.21 ±
0.29; Kn=19.34 ± 1.78. These are consistent with the mea-
surements presented here, which were made 3 years later,
except at J and K5, although they quote a large uncertainty
in the Kn magnitude. Between September 2000 and Octo-
ber 2001 the quasar did not vary at K. In 1995 May/June,
5 Though Francis et al. used the Kn filter, they normalized their
magnitudes to the K band zero point.
It is not too surprising to find several red quasars among the
z3f sample. First, if the z = 2.5− ∼ 4 decline in quasar space
density signals the formation epoch of quasars, then some of
the z3f quasars may be intrinsically young. If a ULIRG (e.g.
Sanders et al. 1989) or a dust-enshrouded phase (e.g. Egami
et al. 1996; Fabian 1999) is the first stage in the evolution
of quasars, then young quasars might be expected to be red
(but see Yu & Tremaine 2002). Second, the largely radio-
selected sample included objects with high radio-to-optical
flux ratios and with flat radio spectra, several of which, from
their variability, appear to be blazars and thus may be red-
dened by a non-thermal contribution to the rest-frame opti-
cal (Whiting et al. 2001). And third, the probability of DLAs
and Lyα systems along the line of sight increases with red-
shift.
The 3 α < −1 z3f quasars can be interpreted as red-
dened by line-of-sight dust or as containing a strong syn-
chrotron component. Ongoing and planned optical, near-IR
and deep X-ray surveys will be more sensitive to reddened
quasars than previous surveys; here we compare the prop-
erties of these 3 α < −1 quasars with the limits of such
surveys as the SDSS, WFCAM/UKIDSS and Chandra deep
fields.
Rest-frame optical continua of L ∼ L∗ z > 3 quasars
13
The 3 α < −1 z3f quasars are relatively faint:
LBQS0056+0125 has r=18.9 (Schneider et al. 1991) and
K=15.7; PKS2215+02 has V=20.6 and K=17.1 and
TXS2358+189 has V∼ 21.3 and K=18.4.
From the SDSS commissioning data, Fan et al. (2001b)
selected a uniform sample of high redshift quasars, setting
a flux limit at i∗ ∼ 20; this would have missed one or two of
the α < −1 z3f quasars. In the near-IR, the LAS (large area
survey), which should complement the SDSS, will reach to
K=18.4 (http://www.ukidss.org/surveys/surveys.html) so
would have picked up all (TXS2358+189 lies at its limit).
The Chandra deep (∼ 0.1 − 1 Ms) fields (Barger et al. 2001;
Hornschemeier et al. 2001; Giacconi et al. 2002) reach down
to flux levels of order 6 × 10−17 to 4 × 10−15 erg s−1 cm−2
(2 − 10 keV). Few of the z3f quasars have X-ray data, but
the two detected by ROSAT-HRI have Fx ∼ 3 − 6 × 10−13
erg s−1 cm−2 and the one upper limit was consistent with a
flux, Fx ∼ 10−14 erg s−1 cm−2 (Siebert & Brinkmann 1998).
The unabsorbed 0.5 − 2 keV (observed frame) X-ray fluxes
for a set of z > 4 quasars range from 1 − 30 × 10−15 erg
s−1 cm−2 (Vignali et al. 2001), so future deep X-ray surveys
should turn up more quasars similar to those in the z3f
sample.
7 CONCLUSIONS
This paper presents near-IR photometry for 20 radio-loud
(16 radio-selected) z > 3 quasars.
Some of the targets had not previously been extensively
observed -- NED lists references only to the radio surveys
in which they were detected and the source of the optical
identification -- so the photometry presented here are the
first for these.
While the underlying aim is to look for evolution in the
central engines, uncertainties due to extinction and the pos-
sible contribution of a non-thermal component complicate
efforts to estimate key central engine parameters from these
data alone.
The main conclusions that can be drawn from this
dataset are:
(i) First, the continuum shapes of faint radio-loud z > 3
quasars show more scatter than those of the bright z > 3
quasars from Paper I. Choosing objects with luminosities
close to L∗, rather than the extremely high luminosity
quasars of the bright z > 3 sample, has not led to a conver-
gence in the optical/UV continuum shapes of z > 3 quasars.
(ii) And second, the distribution of continuum shapes of
these faint radio-loud z > 3 quasars is slightly redder than
that of the bright z > 3 ones, which were a mix of radio-loud
and radio-quiet quasars, with 3 having α < −1. These 3 are
not all radio-selected, so while a synchrotron contribution
cannot be ruled out as a cause for the red colors of the radio-
selected quasars, the red optically-selected quasar probably
suffers intrinsic or line-of-sight reddening. Study of a care-
fully selected set of high redshift radio-quiet quasars would
eliminate the possibly of synchrotron reddening. Near-IR
and optical spectroscopy would complement the more easily
obtained photometry, enabling measures of the central mass
via the Hβ (e.g. Laor 1998), MgII (McLure & Jarvis 2002)
or CIV (Vestergaard 2002) emission line widths, and an es-
timate of intrinsic reddening via the Lyα/Hβ ratio (Netzer
et al. 1995, Bechtold et al. 1997).
The 3 α < −1 quasars may represent a bridge to a larger
population of high redshift quasars that are either intrinsi-
cally reddened or reddened by line-of-sight systems (Warren
et al. 2000), the discovery of which is anticipated from on-
going and future optical, near-IR, and deep X-ray surveys.
ACKNOWLEDGMENTS
The bulk of the data presented here were obtained at
UKIRT, the United Kingdom Infrared Telescope, which is
operated by the Joint Astronomy Centre on behalf on the
U.K. Particle Physics and Astronomy Research Council.
Some of the data were obtained as part of the UKIRT Service
Programme. I thank the staff at UKIRT for their help with
obtaining and reducing the data, in particular Andy Adam-
son, Malcolm Currie, Chris Davis, Tom Kerr, Sandy Leggett
and Thor Wold and John Davies for obtaining the optical
photometry presented here. Also I thank the referee, Dr.
Stephen Warren, for comments which helped me to improve
the paper. Construction of the optical/UV continua for the
bright z > 3 dataset used a large suite of instruments and
telescopes, as described in Paper I. One of the instruments
mentioned in this paper, OSIRIS, is the Ohio State Infrared
Imager/Spectrometer, and was built using funds from NSF
award AST 90-161112 and AST 92-18449 by the OSU As-
tronomical Instrumentation Facility. This publication makes
use of data products from the Two Micron All Sky Survey,
which is a joint project of the University of Massachusetts
and the Infrared Processing and Analysis Center/California
Institute of Technology, funded by the National Aeronautics
and Space Administration and the National Science Foun-
dation. This research has also made use of the NASA/IPAC
Extragalactic Database (NED) which is operated by the Jet
Propulsion Laboratory, California Institute of Technology,
under contract with the National Aeronautics and Space
Administration.
REFERENCES
Allen's Astrophysical Quantities, 4th Edition, 2000, ed. A.
N. Cox, AIP Press, New York.
Andreani, P., Franceschini, A., Granato, G. 1999, MNRAS,
306, 161.
Barger, A. J., Cowie, L. L., Steffen, A. T., Hornschemeier,
A. E., Brandt, W. N., Garmire, G. P.. 2001, ApJ, 560, 23.
Bechtold, J. 1994, ApJS, 91, 1.
Bechtold, J., et al. 1994, AJ, 108, 759.
Bechtold, J., Shields, J., Rieke, M., Ji. P., Scott, J., Kuhn,
O., Elvis, M., Elston, R. 1997, in Emission Lines in Active
Galaxies: New Methods and Techniques, ASP Conference
Series, Vol 113, eds. B. M. Peterson, F.-Z. Cheng and A.
S. Wilson, Astron Soc. Pac., San Francisco, p. 122.
Becker, R. H., White, R. L., Edwards, A. L. 1991, ApJS,
75, 1.
Bennett, C.L., Lawrence, C.R., Burke, B.F., Hewitt J.N.,
Mahoney, J. 1986 ApJS, 61, 1.
Boroson, T. A., Green, R. F. 1992, ApJS, 80, 190.
Boroson, T. A. 2002, ApJ, 565, 78.
14 O. P. Kuhn
Bouchet, P., Lequeux, J., Maurice, E., Pr´evot, L., Pr´evot-
Burnichon, M . L. 1985, A&A, 149, 330.
Boyle, B. J., Shanks, T., Peterson, B. A. 1988, MNRAS,
235, 935.
Boyle B. J., Jones L.R., Shanks T., Marano B., Zitelli V.,
Zamorani G., 1991, in The Space Distribution of Quasars,
ASP Conference Series, Vol 21, ed. D. Crampton, Astron
Soc. Pac., San Francisco, p. 191.
Boyle, B. J., Shanks, T., Croom, S. M., Smith, R. J., Miller,
L., Loaring, N., Heymans, C. 2000, MNRAS, 317, 1014.
Brotherton, M. S., Tran, H. D., Becker, R. H., Gregg, M.
D., Laurent-Muehleisen, S. A., White, R. L. 2001, ApJ,
546, 775.
Chaffee, F. H., et al. 1991, AJ, 102, 461.
Condon, J. J., Cotton, W. D., Greisen, E. W., Yin, Q. F.,
Perley, R. A., Taylor, G. B., Broderick, J. J. 1998, AJ, 115,
1693.
Currie, M. J., Wright, G. S., Bridger, A., Economou, F.
1999, in Astronomical Data Analysis Software and Sys-
tems VIII, ASP Conference Series, Vol 172, eds. D. M.
Mehringer, R. L. Plante, D. A. Roberts, Astron Soc. Pac.,
San Francisco, p. 175.
Cutri, R. M., Nelson, B. O., Kirkpatrick, J. D., Huchra, J.
P., Smith, P. S. 2000, in The New Era of Wide-Field As-
tronomy, ASP Conference Series, Vol 232, eds. R. Clowes,
A. Adamson, G. Bromage, Astron Soc. Pac., San Francisco,
p. 78.
Dallacasa, D. Stanghellini, C., Centonza, M., Fanti, R.
2000, A&A, 363, 887.
DePoy, D. L., Atwood, B., Byard, P., Frogel, J., O'Brien,
T. 1993, Proc. SPIE 1946, 667.
Douglas, J. N., Bash, F. N., Bozyan, F. A., Torrence, G.
W., Wolfe, C. 1996, AJ, 111, 1945.
Drinkwater, M. J., et al. 1997, MNRAS, 284, 85.
Egami, E., Iwamuro, F., Maihara, T., Oya, S., Cowie, L.
L. 1996, AJ, 112, 73.
Ellison, S. L., Yan, L., Hook, I. M., Pettini, M., Wall, J.
V., Shaver, P. 2001, A&A, 379, 393.
Elvis, M., et al. 1994a, ApJS, 95, 1.
Elvis, M., Fiore, F., Wilkes, B. J., McDowell, J. C. 1994b,
ApJ, 422, 60.
Fabian, A. C. 1999, MNRAS, 308, L39.
Fan, X. et al. 2001a, AJ, 121, 54.
Fan, X. et al. 2001b, AJ, 121, 31.
Ficarra, A., Grueff, G., Tomassetti, G. 1985, A&AS, 59,
255.
Forster, K., Green, P. J., Aldcroft, T. L., Vestergaard, M.,
Foltz, C. B., Hewett, P. C. 2001, ApJS, 134, 35.
Francis, P. J., Whiting, M. T., Webster, R. L., 2000, PASA,
53, 56.
Giacconi, R., et al. 2002, ApJS, 139, 369.
Giveon, U., Maoz, D., Kaspi, S., Netzer, H., Smith, P. S.
1999, MNRAS, 306, 637.
Gregg, M. D., Becker, R. H., White, R. L., Helfand, D. J.,
McMahon, R. G., Hook, I. M. 1996, AJ, 112, 407.
Gregg, M. D., Lacy, M., White, R. L., Glikman, E.,
Helfand, D., Becker, R. H., Brotherton, M. S. 2002, AJ,
564, 133.
Gregory, P. C., Condon, J. J. 1991, ApJS, 75, 1011.
Griffith, M. R., Wright, A. E., Burke, B. F., Ekers, R. D.
1994, ApJS, 90, 179.
Haehnelt, M. G. , Natarajan, P., Rees, M. J. 1998, MNRAS,
T. G.,Leggett,
300, 817.
Hagen, H.-J. et al. 1992, A&A, 253, L5.
Haiman, Z.,Menou, K. 2000, ApJ, 531, 42.
Hatziminaoglou, E.,Siemiginowska, A.,Elvis, M. 2001, ApJ,
547, 90.
Hawarden,
S. K.,Letawsky, M.
B.,Ballantyne, D. R.,Casali, M. M. 2001, MNRAS,
325, 563.
Hook,I.M., McMahon,R.G., Boyle,B.J., Irwin,M.J. 1994,
MNRAS, 268, 305.
Hornschemeier, A. E., et al. 2001, ApJ, 554, 742.
Jarvis, M. J., Rawlings, S. 2000, MNRAS, 319, 121.
Kauffmann, G., Haehnelt, M. 2000, MNRAS, 311, 576.
Kuhn, O., Bechtold, J., Elvis, M., Elston, R. 2001, ApJS,
136, 225. (Paper I)
Landolt, A. 1992, AJ, 104, 340.
Laor, A., Fiore, F., Elvis, M., Wilkes, B. J., McDowell, J.
C. 1997, ApJ, 477, 93.
Laor, A. 1998, ApJ, 505, L83.
McAlpine, G. M., Feldman, F. R. 1982, ApJ, 261, 412.
McDowell, J., Kuhn, O., Elvis, M., Wilkes, B. 1991, in
"Structure and Emission Properties of Accretion Disks",
Proc. of the Sixth IAP Astrophysics Meeting/IAU Collo-
quium No. 129, eds. C. Bertout, S. Collin-Souffrin, J. P.
Lasota, J. Tran Thanh Van, Gif sur Yvette Cedex, France:
Editions Fronti`eres, 473.
McIntosh, D. H., Rieke, M. J., Rix, H.-W., Foltz, C. B.,
Weymann, R. J. 1999, ApJ, 514, 40.
McLure, R. J., Jarvis, M. J. 2002, MNRAS, 337, 109.
Marcha, M. J. M., Browne, I. W. A., Impey, C. D., Smith,
P. S. 1996, MNRAS, 281, 425.
Marziani, P., Sulentic, J. W., Zwitter, T, Dultzin-Hacyan,
D., Calvani, M. 2001, ApJ, 558, 553.
Netzer, H., Brotherton, M. S., Wills, B. J., Han, M., Wills,
D., Baldwin, J. A., Ferland, G. J., Browne, I. W. A. 1995,
ApJ, 448, 27.
O'Dea, C. P., Baum, S. A., Stanghellini, C. 1991, ApJ, 380,
660.
O'Dea, C. P. 1998, PASP, 110, 493.
Pauliny-Toth, I. I. K., Witzel, A., Preuss, E., Kuhr, H.,
Kellerman, K. I., Fomalont, E. B. 1978, AJ, 451.
Peck, A. B., Taylor, G. B. 2000, ApJ, 534, 90.
Pei, Y. C., Fall, S. M., Bechtold, J. 1991, ApJ, 378, 6.
Pettini, M.. King, D. L., Smith, L. J., Hunstead, R. W.
1997, ApJ, 478, 536.
Pr´evot, M. L., Lequeux, J., Maurice, E., Pr´evot, L., Rocca-
Volmerange, B. 1984, A&A, 132, 389.
Puchnarewicz, E. M., et al. 1996, MNRAS, 315, 1.
Richards, G. T., et al. 2001, AJ, 121, 2308.
Rieke, G. H. 1984, in MMTO Visiting Astronomer Infor-
mation, Multiple Mirror Telescope Observatory Technical
Report No. 13.
Rieke, G. H., Lebofsky, M. J., Low, F. J. 1985, AJ 90, 900.
Roche P.F. et al. 2002, Proc SPIE 4841, Instrument Design
and Performance for Optical/IR Ground-Based Telescopes,
eds. M Iye and A.F Moorwood.
Sanders, D. B., Phinney, E. S., Neugebauer, G., Soifer, B.
T., Matthews, K. 1989, ApJ, 347, 29.
Savage, A., Jauncey, D. L., White, G. L., Peterson, B. A.,
Peters, W. L., Gulkis, S., Condon, J. J. 1990, AuJPh, 43,
241.
Savage, B. D., Mathis, J. S. 1979, ARA&A, 17, 73.
Rest-frame optical continua of L ∼ L∗ z > 3 quasars
15
Schlegel, D. J., Finkbeiner, D. P., Davis, M. 1998, ApJ,
500, 525.
Schmidt, M. A., Schneider, D. P., Gunn, J. E. 1987, ApJ,
316, L1.
Schmidt, M. A., Schneider, D. P., Gunn, J. E. 1991, in The
Space Distribution of Quasars, ASP Conf Series, Vol 21,
ed. D. Crampton, Astron Soc. Pac., San Francisco, p. 110.
Schneider, D. P., Schmidt, M. A., Gunn, J. E. 1991, AJ,
101, 2004.
Schneider, D. P., Schmidt, M. A., Gunn, J. E. 1997, AJ,
114, 36.
Serjeant, S., Rawlings, S. 1997, Nature, 379, 304.
Shimmins, A. J., Day. G. A., Ekers, R. D., Cole, D. J. 1966,
AuJPh, 19, 837.
Siebert, J., Brinkmann, W. 1998, A&A, 333, 63.
Siebert, J., Brinkmann, W., Drinkwater, M. J., Yuan, W.,
Francis, P. J., Peterson, B. A., Webster, R. L. 1998, MN-
RAS, 301, 261.
Siemiginowska, A., Elvis, M. 1997, ApJ, 482, L9.
Stanghellini, C., O'Dea, C. P., Dallacasa, D., Baum, S. A.,
Fanti, R., Fanti, C. 1998, A&AS, 131, 303.
Stern, D.,Djorgovski, S. G., Perley, R. A., de Carvalho, R.
R., Wall, J. V. 2000, AJ, 119, 1526.
Sun, W.-H., Malkan, M. A. 1989, ApJ, 346, 68.
Tokunaga, A. T., Simons, D. A., Vacca, W. D. 2002, PASP,
114, 180.
V´eron-Cetty, M.-P., V´eron, P. 2000, ESO Sci Rep, 19, 1
(VCV9).
V´eron-Cetty, M.-P., V´eron, P. 2001, A&A, 374, 92
(VCV10).
Vestergaard, M. 2002, ApJ, 571, 733.
Vignali, C. Brandt, W. N., Fan, X., Gunn, J. E., Kaspi, S.,
Schneider, D. P., Strauss, Michael A. 2001, AJ, 122, 2143.
Vigotti, M., Carballo, R., Benn, C. R., DeZotti, G., Fanti,
R., Gonzalez Serrano, J. I., Mack, K.-H., Holt, J. 2003,
ApJ, 591, 43.
Warren, S. J., Hewett, P. C., Osmer, P. J. 1994, ApJ, 421,
412.
Warren, S. J., Hewett, P. C., Foltz, C. B. 2000, MNRAS,
312, 827.
Warren, S. J., Møller, P., Fall, S. M., Jakobsen, P. 2001,
MNRAS, 326, 759.
Webster, R. L., Francis, P. J., Peterson, B. A., Drinkwater,
M. J., Masci, F. J. 1995, Nature, 375, 469.
White, R. L., Becker, R. H. 1992, ApJS, 79, 331.
White, R. L., Kinney, A. L., Becker, R. H. 1993, ApJ, 407,
456.
Whiting, M. T., Webster, R. L., Francis, P. J. 2001, MN-
RAS, 323, 718.
Yu, Q., Tremaine, S. 2002, MNRAS, 335, 965.
APPENDIX A: OPTICAL PHOTOMETRY
Optical, V and R-band, photometry for several of the target
quasars was obtained by J. K. Davies over the course of
his 5 night run (2000 August 24-28 UT) at the Jacobus
Kapteyn Telescope in La Palma. Since these data were not
obtained for all of the target quasars, they were not used in
the analysis. However they are plotted in Figure 2, which
shows that, except for the reddest quasars with α < −1, the
optical data generally trace the UV turnover that is seen in
the low redshift mean SED (Elvis et al. 1994a).
The optical images were reduced using the ccdproc
and apphot packages in IRAF. All frames were bias sub-
tracted, and master sky and dome flats were produced for
each filter. The bias-subtracted images were first divided by
the sky or dome flats to remove small scale features, and
then, to reduce the residual gradient, they were divided by
a heavily smoothed median of the sky- or dome-flattened
science frames. Standard stars from Landolt (1992) were
observed for flux calibration. Zero points, extinction and
color coefficients were determined from fits to standard star
instrumental magnitudes measured through an aperture of
radius 20 pixels (diameter 13.′′2). As with the near-IR pho-
tometry, aperture corrections were needed. In this case, the
radius of the small aperture was 6 pixels (diameter 4′′) ex-
cept for one instance when it had to be decreased to 4 (di-
ameter 2.′′6), as noted in Table A1. The aperture corrections
were determined for each image. The telescope jumped dur-
ing some of the integrations, causing E-W elongated images
with multiple peaks. These are noted in Table A1. The dis-
tance between the peaks was comparable to or smaller than
the 20-pixel radius aperture, and the magnitudes were mea-
sured in the same way as from the round images, however the
results should be regarded with more uncertainty than the
error in table A1 indicates. Since some objects were observed
either at V or R only, two sets of transformation equations,
with and without the color terms, were fit. The appropriate
set was used to determine the magnitudes listed in Table
A1. For the two objects observed only at V or R, a synthetic
color correction was incorporated in the magnitude-to-flux
conversion, by using the zero magnitude fluxes for a ν −0.46
power-law rather than for Vega (as discussed in Section 4.1).
Effective wavelengths and zero-magnitude fluxes for Vega
and for the ν −0.46 power-law were computed for both the
Harris V and R filters used at the JKT. For V, these are
0.545µm, 36.46 × 10−10 and 36.11 × 10−10 erg s−1 cm−2A−1,
respectively. For R, these are 0.639µm, 22.44 × 10−10 and
22.11 × 10−10 erg s−1 cm−2A−1, respectively.
APPENDIX B: ESTIMATED EMISSION LINE
CONTRIBUTIONS TO THE BROAD BAND
FLUXES
For z = 3−4, the emission lines of MgII, Hβ, [OIII]4959,5007
as well as the blended FeII lines fall within the Z, J, H and
K bands. I used the FBQS composite spectrum of Brother-
ton et al. (2001) to compute, as a function of redshift, the
relative contribution of emission lines to the continuum flux
within the band. The bands were defined as discussed in sec-
tion 4.1 and listed in Table 3. Following Brotherton et al.
(2001), the continuum level was taken to be a power law fit
through the composite at 1285, 2200, 4200 and 5770A, which
had a spectral index, α = −0.46. The ratios of continuum
to total flux within a band range from ∼ 0.7 − 0.96 and are
plotted as a function of redshift in Figure B1. For z > 3, the
blended FeII and Balmer continuum falls within J and its
contribution to the broad band flux is greater than the emis-
sion line contribution at K, which explains why the spectral
indices computed from line-corrected fluxes are redder than
those computed from uncorrected fluxes (Table 4). Differ-
16 O. P. Kuhn
Table A1. Optical observations
Quasar
PKS2215+02
MG3J222537+2040
TXS2342+342
TXS2358+189
MG1J000655+1416
MG3J015105+2516
PKS0201+113
MG3J023222+2318
PKS0335-122
PKS0336-017
Va
20.58 ± 0.14b
18.78 ± 0.03
18.89 ± 0.04
21.45 ± 0.19
21.21 ± 0.16
18.49 ± 0.04
20.46 ± 0.09
19.90 ± 0.05
20.12 ± 0.07
20.15 ± 0.10
20.30 ± 0.08
...
...
Ra
20.19 ± 0.05
18.42 ± 0.02
18.52 ± 0.03
...
...
18.12 ± 0.03
20.45 ± 0.09
19.23 ± 0.03
19.81 ± 0.06
20.13 ± 0.08
18.53 ± 0.08
18.50 ± 0.07
UT date
2000 Aug 28
2000 Aug 25
2000 Aug 24
2000 Aug 28
2000 Aug 28
2000 Aug 26
2000 Aug 25
2000 Aug 27
2000 Aug 25
2000 Aug 28
2000 Aug 28
2000 Aug 28
2000 Aug 28
comments
tel jumped 1x during V image
tel jumped 1x, not color corrected
tel jumped 2x, not color corrected
tel jumped 1x during R image
not color corrected
not color corrected
a All magnitudes were measured within an aperture of radius 6 pixels (4′′diameter) except PKS2215+02 at R, and an
aperture correction was applied to determine the magnitude within the 20 pixel radius (13.′′2 diameter) aperture used for
the standard stars. Magnitudes are determined assuming an extinction and color correction except when the quasar was
observed in only one band, in which case a synthetic color correction is incorporated into the magnitude to flux conversion.
b On the R-band image, the aperture correction had to be made from an aperture of radius 4, rather than 6, pixels.
1.1
1
0.9
0.8
3
3.5
4
4.5
Figure B1. Ratio of continuum to total (continuum plus emis-
sion line) flux within the Z, J, H and K bands as a function of
redshift. The thick lines are used for the UFTI Z, J, H and K
filters and thin dotted lines for the 'tophat' approximations to
J, H, K. Inset is the mean FBQS composite of Brotherton et al.
(2001; dotted line) and continuum fit (a Fν ∼ ν−0.46 power law;
solid line) that were used in this calculation.
ences in line strengths from object to object (e.g. McIntosh
et al. 1999, Forster et al. 2001), however, are a significant
source of uncertainty. In the analysis, I have opted to use
uncorrected spectral indices since these involve one fewer
assumption and more easily compared to results from other
studies.
This paper has been typeset from a TEX/ LATEX file prepared
by the author.
|
0808.2582 | 1 | 0808 | 2008-08-19T12:24:17 | Muon Charge Ratio of Ultrahigh Energy Cosmic Rays | [
"astro-ph",
"hep-ex",
"hep-ph",
"nucl-ex"
] | The muon charge ratio of ultrahigh energy cosmic rays may provide information to detect the composition of the primary cosmic rays. We propose to extract the charge information of high energy muons in very inclined extensive air showers by analyzing their relative lateral positions in the shower transverse plane. | astro-ph | astro-ph | November 16, 2018 12:26 WSPC/INSTRUCTION FILE
mabq
Modern Physics Letters A
c(cid:13) World Scientific Publishing Company
8
0
0
2
g
u
A
9
1
]
h
p
-
o
r
t
s
a
[
1
v
2
8
5
2
.
8
0
8
0
:
v
i
X
r
a
Muon Charge Ratio of Ultrahigh Energy Cosmic Rays∗
BO-QIANG MA
School of Physics and State Key Laboratory of Nuclear Physics and Technology,
Peking University, Beijing 100871, China
[email protected]
The muon charge ratio of ultrahigh energy cosmic rays may provide information to
detect the composition of the primary cosmic rays. We propose to extract the charge
information of high energy muons in very inclined extensive air showers by analyzing
their relative lateral positions in the shower transverse plane.
The most high energy particles can be observed by human being are from cosmic
rays. The study of them belongs to frontiers of human knowledge in combination of
cosmology, astrophysics, and particle physics, and can provide better understanding
of the universe from most small to most big, i.e., connecting quarks to the cosmos.
The universe is not empty, but full of background relic particles from the big bang.
It has long been anticipated that the highest energy cosmic rays would be protons
from outside the galaxy, and there is an upper limit of the highest energy in the
observed proton spectrum, commonly referred to as the GZK cutoff 1, as the protons
traveling from intergalactic distances should experience energy losses owing to pion
productions by the photons in the cosmic background radiation. Although there
have been attentions for the cosmic ray events above the GZK cutoff, it is natural
to expect that these ultrahigh energy cosmic rays come from sources within the GZK
zone 2, i.e., not far from us in more than tens of Mpc. Recently there are also reports
on the observation of the GZK cut-off by new experiments 3. However, questions
about the composition of such ultrahigh energy cosmic ray particles, e.g., whether
they are protons, neutrons, or anti-nucleons 4, are still open to investigations.
Muons in the air showers are mainly from decays of pions and kaons produced
in the interactions of the primary cosmic rays with the atmosphere. The very high
energy secondary pion and kaon cosmic rays can be considered as from the current
fragmentation of partons in deep inelastic scattering of the primary cosmic rays with
the nucleon targets of the atmosphere in a first approximation 5. We also consider
only the favored fragmentation processes, i.e., the π+, which is composed of valence
u and ¯d quarks, is from the fragmentation of u and ¯d quarks in the nucleon beam,
∗Invited talk at 2007 International Symposium on Cosmology and Particle Astrophysics
(CosPA2007), November 13-15, 2007, Taipei.
1
November 16, 2018 12:26 WSPC/INSTRUCTION FILE
mabq
2 BO-QIANG MA
and the π−, which is composed of valence ¯u and d quarks, is from the fragmentation
of ¯u and d quarks 6. Similarly, the K +, which is composed of valence u and ¯s, is from
the fragmentation of u and ¯s quarks, and the K −, which is composed of valence
¯u and s, is from the fragmentation of ¯u and s quarks. The µ+ is from the decay
of a π+ or a K + and the µ− is from the decay of a π− or a K −. We can roughly
estimate the muon charge ratio by
µ+
µ− =
R 1
0 dx (cid:8)(cid:2)u(x) + ¯d(x)(cid:3) + κ [u(x) + ¯s(x)](cid:9)
R 1
0 dx {[d(x) + ¯u(x)] + κ [¯u(x) + s(x)]}
,
(1)
where q(x) is the quark distribution with flavor q for the incident hadron beam
and κ ∼ 0.1 → 0.3 is a factor reflecting the relative muon flux and fragmentation
behavior of K/π. Secondary collisions do not influence the above estimation, since
the current parton beams still keep their flavor content and act as the current
partons after the strong interactions with the partons in the atmosphere targets.
Adopting a simple model estimation of the parton flavor content in the nucleon
without any parameter 7, we find that µ+/µ− ∼ 1.7 for proton and µ+/µ− ∼ 0.7
for neutron. This simple evaluation is in agreement with the empirical expectation
of µ+/µ− ≈ 1.66 for proton and µ+/µ− ≈ 0.695 for neutron 8 as well as that in an
extensive Monte Carlo calculation 9, thus it provides a clear picture to understand
the dominant features for the muon charge ratio by the primary hadronic cosmic
rays. For the µ+/µ− ratio for antiproton, it is equivalent to the µ−/µ+ ratio for
proton by using Eq. (1), thus we find µ+/µ− ∼ 0.6 for antiproton, which is close to
that for neutron. The µ+/µ− ratio for antineutron is also equivalent to the µ−/µ+
ratio for neutron, and it is µ+/µ− ∼ 1.4, which is close to that for proton. It
is hard to distinguish between the primary neutrons and antiprotons (or protons
and antineutrons) by the µ+/µ− ratio of the air shower, unless very high precision
measurement is performed and also our knowledge of the muon charge ratio for each
nucleon species is well established.
The study of cosmic rays with primary energies above 105 GeV are typically
based on the measurements of extensive air showers (EAS) that they initiate in the
atmosphere. The ground detector array records the secondary particles produced in
shower cascades, including photons, electrons (positrons), muons, and some hadrons.
Then their arrival times and density profiles are used to infer the primary energy
and composition of the incident cosmic ray particle, usually through comparison
with simulated results. Photons, electrons and positrons are the most numerous
secondary particles in an EAS event. However, for very inclined showers, these
electromagnetic components would travel a long slant distance and are almost com-
pletely absorbed before they reach the ground. On the other hand, muons are decay
products of charged mesons in shower hadronic cascades. Most high energy muons
survive their propagation through the slant atmospheric depth, during which they
lose typically a few tens of GeV's energy. These high energy muons carry impor-
tant information about the nature of the primary cosmic ray hadron, which will be
extracted from their energy spectrum and lateral distribution.
November 16, 2018 12:26 WSPC/INSTRUCTION FILE
mabq
Muon Charge Ratio of Ultrahigh Energy Cosmic Rays
3
As discussed by Hwang and I 4, the ratio of positive versus negative muons
µ+/µ− is a significant quantity which can help to discern the primary composition,
and at high energies this charge ratio also reflects important features of hadronic
meson production in cosmic ray collisions. In order to obtain such muon charge
information, we would need a way to distinguish between positive and negative high
energy muons. Unfortunately, existing muon detectors available at shower arrays,
usually scintillators and water Cerenkov detectors, are not commonly equipped with
magnetized steel to differentiate the muon charges. Even if they were, the limited
region of the magnetic field prevents definite determination of high energy muons'
track curvature.
This invites us to think of the geomagnetic field as a huge natural detector for
muon charge information. Apparently, after being produced high in the atmosphere,
a positively charged muon would bend east on its way down while a negatively
charged muon would bend west, introducing an asymmetry into the density profile
of the shower front. If their separation is large enough as compared with other
circularly symmetric "background" deviations, it will be possible to distinguish the
positive muons from the negative ones.
To see such an effect, Xue and I 10 analyzed the possibility of obtaining the
charge information of high energy muons in very inclined extensive air showers.
We have demonstrated that positive and negative high energy muons in sufficiently
inclined air showers can be distinguished from each other through their opposite ge-
omagnetic deviations in the transverse plane. We developed a revised Heitler model
to calculate this distinct double-lobed distribution, and studied the condition for
the two lobes of either positive or negative muons to be separable with confidence.
From our criterion of resolvability, we concluded that a zenith angle 75◦ ≤ θ ≤ 85◦
will be most suitable for our approach.
There are already some results from full air shower simulations that take into
account the geomagnetic effect on muon propagation 11,12,13,14,15,16,17,18. They
illustrated remarkable double-lobed muon lateral density profile in very inclined
air showers, which is in agreement with our expectation qualitatively. However, no
present study has fully considered the high energy part of muon content, which
can be used to compare with our results. Thus we would like to propose future
simulations of very inclined extensive air showers that focus on the behavior of high
energy muons. They also have to keep track of the muon charges and the relation
to their lateral positions. For more detailed analysis and discussion, please refer to
Ref.10.
In summary, we propose to extract the charge information of high energy muons
in very inclined extensive air showers by analyzing their relative lateral positions in
the shower transverse plane. This muon charge information is helpful to detect the
composition of cosmic rays, e.g., the neutron or antiproton content of the ultrahigh
energy cosmic rays.
November 16, 2018 12:26 WSPC/INSTRUCTION FILE
mabq
4 BO-QIANG MA
Acknowledgments
I am very grateful to Pauchy Hwang and the organizers for their invitation and
warm hospitality. I also thank Pauchy Hwang and BingKan Xue for the collab-
orated results in this talk. This work is partially supported by National Natural
Science Foundation of China (Nos. 10721063, 10575003, 10528510), by the Key
Grant Project of Chinese Ministry of Education (No. 305001), and by the Research
Fund for the Doctoral Program of Higher Education (China).
References
1. K. Greisen, Phys. Rev. Lett. 16, 748 (1966); G.T. Zatsepin and V.A. Kuzmin, Pis'ma
Zh. Eksp. Tero. Fiz. 4, 114 (1966) [JETP Lett. 4, 78 (1966)].
2. F.W. Stecker, Phys. Rev. Lett. 21, 1016 (1968). For an extensive review, see,
F.W. Stecker, astro-ph/0101072.
3. R. Abbasi et al., HiRes Collaboration, Phys. Rev. Lett. 100, 101101 (2008). Also Pierre
Auger Observatory reported the observation of the GZK cut-off.
4. W-Y.P. Hwang and B.-Q. Ma, Eur. Phys. J. A 25, 467 (2005).
5. See, e.g., B.-Q. Ma, I. Schmidt, J. Soffer, and J.-J. Yang, Nucl. Phys. A 703, 346 (2002).
6. See, e.g., B.-Q. Ma, I. Schmidt, and J.-J. Yang, Phys. Rev. D 65, 034010 (2002).
7. Y.-J. Zhang, B. Zhang, and B.-Q. Ma, Phys. Lett. B 523, 260 (2001).
8. R.K. Adair, Phys. Rev. Lett. 33, 115 (1974); R.K. Adair et al., Phys. Rev. Lett. 39,
112 (1977); O.C. Allkofer et al., Phys. Rev. Lett. 41, 832 (1978).
9. J.N. Capdevielle and Y. Muraki, Astropart. Phys. 11, 335 (1999). See, e.g., Fig. 7.
10. B. Xue and B.-Q. Ma, Astropart. Phys. 27, 286 (2007).
11. M. Ave, R. A. Vazquez and E. Zas, Astropart. Phys. 14, 91 (2000).
12. H. H. Aly, M. F. Kaplon, and M. L. Shen, Nuovo. Cim. 31, 905 (1964).
13. J. R. Horandel, J. Phys. G 29, 2439 (2003).
14. C. A. Ayre et al., J. Phys. A 5, L102 (1972).
15. T. Hebbeker and C. Timmermans, Astropart. Phys. 18, 107 (2002).
16. S. Ostapchenko, Nucl. Phys. Proc. Suppl. 151, 147 (2006).
17. See, e.g., W. D. Apel et al. [KASCADE Collaboration], arXiv:astro-ph/0510810.
18. P. Hansen, T. K. Gaisser, T. Stanev and S. J. Sciutto, Phys. Rev. D 71, 083012 (2005);
A. Cillis and S. J. Sciutto, arXiv:astro-ph/9908002.
|
astro-ph/0504322 | 1 | 0504 | 2005-04-14T11:58:51 | Optimal column density measurements from multiband near-infrared observations | [
"astro-ph"
] | We consider from a general point of view the problem of determining the extinction in dense molecular clouds. We use a rigorous statistical approach to characterize the properties of the most widely used optical and infrared techniques, namely the star count and the color excess methods. We propose a new maximum-likelihood method that takes advantage of both star counts and star colors to provide an optimal estimate of the extinction. Detailed numerical simulations show that our method performs optimally under a wide range of conditions and, in particular, is significantly superior to the standard techniques for clouds with high column-densities and affected by contamination by foreground stars. | astro-ph | astro-ph | Astronomy & Astrophysics manuscript no. 2267ms
(DOI: will be inserted by hand later)
July 2, 2018
5
0
0
2
r
p
A
4
1
1
v
2
2
3
4
0
5
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
Optimal column density measurements from multiband
near-infrared observations
M. Lombardi1,2
1 European Southern Observatory, Karl-Schwarzschild-Strasse 2, D-85748 Garching bei Munchen, Germany
2 University of Milan, Department of Physics, via Celoria 16, I-20133 Milan, Italy
Received ***date***; Accepted ***date***
Abstract. We consider from a general point of view the problem of determining the extinction in dense molecular
clouds. We use a rigorous statistical approach to characterize the properties of the most widely used optical and
infrared techniques, namely the star count and the color excess methods. We propose a new maximum-likelihood
method that takes advantage of both star counts and star colors to provide an optimal estimate of the extinction.
Detailed numerical simulations show that our method performs optimally under a wide range of conditions and, in
particular, is significantly superior to the standard techniques for clouds with high column-densities and affected
by contamination by foreground stars.
Key words. dust, extinction -- methods: statistical -- ISM: clouds -- infrared: ISM -- stars: formation
1. Introduction
A detailed comprehension of the structure and physical
properties of dark molecular clouds is a critical step to
understand fundamental processes such as star and planet
formation. Still, after several decades of investigations,
very little is known about dark clouds, about their re-
lationship to the diffuse interstellar medium, and about
the composition and evolution of the dust grains present
in the clouds.
Molecular clouds are composed of approximately 99%
molecular hydrogen and helium, but due to the absence
of a dipole moment these molecules are virtually unde-
tectable at the low temperature (∼ 10 K) that charac-
terize these objects. As a result, astronomers have been
using different tracers to study the density distribution
of molecular clouds. Historically, the first technique was
based on a statistical analysis of the angular density of
stars observed through a cloud (Wolf 1923; Bok 1937).
This method, known as star counts, uses the dust (which
is responsible for the extinction of light) as a tracer of H2
and He.
More recently, radio observations of carbon monox-
ide (CO) and its isotopes have been used to study the
spatial distribution and the physical conditions of molec-
ular clouds, under the assumption that CO is a reli-
able tracer of the matter in the cloud, i.e. that the ra-
tio N (CO)/(cid:0)N (H2) + N (He)(cid:1) is approximately constant.
Send offprint requests to: M. Lombardi
Correspondence to: [email protected]
Radio spectroscopy techniques have provided extremely
effective in studying the most dense regions (above
1022protons cm−2) and, more importantly, give dynamical
information on the cloud structure. However, with the ad-
vent of various dust detectors in the near-infrared (NIR),
far infrared, millimeter, and sub-millimeter bands, it has
become clear that several poorly constrained processes
(e.g., deviations from thermodynamic equilibrium, opac-
ity variations, chemical evolution, and more importantly
depletion of molecules) can significantly affect the results
of analyses based on radio observations (e.g. Lada et al.
1994; Alves et al. 1999; Harjunpaa et al. 2004).
The reddening of background stars offers a natural
method to study the distribution of dust in molecu-
lar clouds, and thus the hydrogen column density. This
technique can be better applied to the infrared bands,
which compared to optical bands are less affected by
extinction and are less sensitive to the physical prop-
erties of the dust grains (Mathis 1990). Before the ad-
vent of large format array cameras, the lack of instru-
mental sensitivity clogged infrared observations to small,
dense clouds (e.g. Jones et al. 1980; Frerking et al. 1982;
Jones et al. 1984; Casali 1986). More recently, infrared ar-
rays have made it possible to measure thousands of stars
and to extend the original technique to entire molecular
cloud complexes (Lada et al. 1994, 1999; Alves et al. 2001;
Lombardi & Alves 2001). Such measurements are free of
the complications that plague molecular-line data and per-
mit a detailed analysis of the cloud density distribution.
2
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
Although the success of the NIR color excess method
is evident, there is still a need for a deeper understanding
of its limitations, statistical biases, and uncertainties. The
aim of this paper is twofold: (i) to study in detail the sta-
tistical properties of the star count and color excess meth-
ods and (ii) to describe a new, optimal method based on
a maximum-likelihood analysis. The method is described
here for NIR observations, but could equally well be ap-
plied to other infrared bands for which the extinction law
and the intrinsic color distribution are known with good
accuracy. The paper is organized as follows. In Sect. 2
we introduce our formalism and consider from a statisti-
cal point of view the extinction of stars by a foreground
cloud. The two main techniques used to obtain extinction
measurements, namely the star count and the NIR color
excess methods, are discussed in Sect. 3. In Sect. 4 we
describe the new maximum-likelihood method and show
by means of numerical simulations that it performs excel-
lently even in the presence of a significant contamination
by foreground stars. Our conclusions are briefly reported
in Sect. 5. Finally, in Appendix A we consider the sta-
tistical properties of the median and of related estimators
used often in infrared studies of dark clouds to remove the
effects of foreground stars.
2. Basic relations
Suppose that N stars with magnitudes (on a given band)
{ mn} are observed in a given region of the sky (here-
after we will use the hat to denote measured quanti-
ties). These observed magnitudes depend on the original,
unreddened, absolute star magnitudes {Mn}, on the star
distances {Dn}, on the individual extinctions {An}, and
on the photometric errors {ǫn}. In particular, we have
mn = Mn + 5 log10 Dn − 5 + An + ǫn = mn + ǫn , (1)
where the distances {Dn} are taken to be expressed in
parsecs. The quantity mn that appears in the r.h.s. of this
equation is the reddened magnitude of the n-th star free
from measurement errors (i.e., the magnitude that would
be observed in the limit of an extremely long exposure
time).
2.1. Single-band probability distributions
In order to consider the most general situation in a proper
statistical way, we introduce several probability distribu-
tions:
ρ(M, D): the probability distribution (density) of stars
with absolute magnitude M at distance D;
pA(AD): the probability distribution of extinction for ob-
jects located at distance D, i.e. the probability of an
extinction of A magnitudes on a star given that its
distance is D;
pǫ( mm): the probability distribution of photometric
measurement errors, i.e. the probability of measuring
a magnitude m for a star given that its true reddened
magnitude is m.
These probability distributions are now defined in the
most general way. Later, however, we will consider spe-
cial forms of them that allow us to simplify the relevant
equations. Note also that we take the extinction A at dis-
tance D as a random variable. Hence, we have introduced
above the probability distribution pA(AD) rather than
a (deterministic) function A(D). This general approach
can be used to describe molecular clouds with "patchy"
column densities, which seems to be quite common (see,
e.g., Lada et al. 1999; Cambr´esy 1999). Finally, we use a
general form for the photometric error that includes the
common case where the error depends on the magnitude
of the star. Moreover, we also consider through pǫ( mm)
the case of undetected objects, i.e. the case where a star
of true magnitude m close to the detection limit produces
no detectable flux; in this case we write m = null. Hence,
the quantity
c(m) = 1 − pǫ(nullm)
(2)
represents the completeness of our observations for stars
with true magnitude m.
We assume that both pA(AD) and pǫ( mm) are nor-
malized to unity, i.e.
Z ∞
pǫ(nullm) +Z ∞
−∞
0
pA(AD) dA = 1
pǫ( mm) d m = 1
∀D ,
∀m .
(3)
(4)
Moreover, we take ρ(M, D) to be normalized so that the
quantity
dN = D2ρ(M, D) dD dΩ dM
(5)
represents the expected number of stars located at a dis-
tance in the range [D, D + dD] in a patch of the sky of
solid angle dΩ, and with absolute magnitudes between M
and M + dM .
Equation (1) can be used to obtain the probability dis-
tribution of reddened magnitudes pm(m). Using the pre-
vious definitions we find
pm(m) =Z ∞
0
dD D2Z ∞
× ρ(m − 5 log10 D + 5 − A, D) .
dA pA(AD)
0
(6)
Finally, pm(m) can be converted into the distribution of
observed magnitudes p m( m) by convolving it with the
measurement error probability distribution pǫ( mm):
p m( m) =Z ∞
Because of the way pǫ( mm) is defined, Eq. (7) includes
also the case m = null. Note also that p m( m) is normalized
so that p m( m) d m represents the expected angular density
of stars with observed magnitudes on the range [ m, m +
d m].
pǫ( mm)pm(m) dm .
(7)
∞
Equations (6) and (7) are basic equations of this paper.
However, in the form they are written, they are by far too
general to be useful in practical cases.
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
3
2.2. Simplifications
In order to make use of the general relations written above
we introduce a number of useful simplifications.
First, we note that so far we have considered obser-
vations carried out on a single band. In order to threat
multi-band data, we introduce a reddening law as follows:
for any band λ, the extinction Aλ is given by
Aλ = kλAV ,
(8)
where kλ is a constant and AV is the extinction in the V
band. In other words, Eq. (8) states that the ratio Aλ/Aλ′
between the extinction in two bands λ and λ′ is a constant
for a given cloud. In reality, this ratio is known to vary
with the environment for short wavelengths, but is almost
universal in the NIR (and, according to a recent work of
Indebetouw et al. 2004, for wavelengths in the range 3 --
10 µm as well). In the rest of this paper we will follow the
usual notation and express all extinctions in terms of AV .
A second important assumption that we will use re-
gards the functional form of ρ(M, D). First, we will as-
sume that this function can be separated in M and D,
i.e. can be written as the product of two functions, one
involving M only, and one involving D only: ρ(M, D) =
pM (M )ρ(D). In practice, this assumption is equivalent to
saying that we observe a single "population" of stars at
all distances. Since observations show that star number
counts are well approximated by an exponential law of
the form pm(m) ∝ 10αm (where the exponent α ≃ 0.34 is
approximately independent of the band considered in the
NIR), we also assume that pM (M ) follows a similar law.
We then write
ρ(M, D) = νλ10αM ρ(D) ,
(9)
where νλ is a normalization factor. In the following we will
assume that ρ(D), similarly to α, is independent of the
band considered, while νλ is not. Note also that, because
of the exponential form of this equation in M , a change
of the normalization factor νλ is in practice equivalent to
a change of the zero-point used for the magnitude M .
Using Eqs. (9) and (8) we can slightly simplify Eq. (6)
by performing the integration over D:
pm(m) = Z ∞
0
dD D2−5αρ(D)Z ∞
× νλ105α10αm10−αkλAV
0
dAV pAV (AV D)
≡ aνλ10αmZ ∞
0
dAV pAV (AV )10−αkλAV , (10)
where we have defined a distance-weighted probability dis-
tribution for AV as
105α
dD D2−5αρ(D)pAV (AV D) , (11)
and where a is a numerical factor introduced in Eq. (10)
to ensure that pAV (AV ) is normalized to unity:
a Z ∞
pAV (AV ) =
0
a = 105αZ ∞
0
dD D2−5αρ(D) .
(12)
V ) = δ(A′
Note that pAV (AV ) has a simple interpretation, being
the probability distribution for stars to have undergone
a given extinction regardless of their distance. For exam-
ple, if all stars were subject to the same extinction AV , we
would have pAV (A′
V − AV ). Note also that, since
α ≃ 0.34 can be taken to be approximately independent of
the band considered, pAV (AV ) is also band-independent.
Equation (10) shows an interesting property of extinc-
tion studies: the probability distribution pm(m) depends
on pAV (AV D) only through pAV (AV ). This point is par-
ticularly important since it shows an intrinsic limitation
of extinction measurements. Indeed, all observables are
derived from p m( m), the distribution of observed magni-
tudes, and this function depends only on pm(m) (other
than observational "limiting factors" such as pǫ( mm) or
c( m)). Since, as noted above, pm(m) does not depend di-
rectly on pAV (AV D), any estimator based on observed
magnitudes only1 will not provide any information on
pAV (AV D) directly but (in the best case scenario) only
on pAV (AV ). Since pAV (AV ) is a sort of convolution of
pAV (AV D), in general it is not possible to have a complete
knowledge on pAV (AV D), not even if we know the star
density distribution ρ(D); this limitation, among other
things, prevents us from gathering information on the
three-dimensional structure of a molecular cloud.
A case where, instead, the function pAV (AV D) can be
recovered is a deterministic model for the extinction, i.e.
a cloud complex whose extinction AV depends directly on
the distance D (and, thus, is not a random variable). In
this case we have
pAV (AV D) = δ(cid:0)AV − AV (D)(cid:1) .
The extinction AV (D) has a simple relationship with the
integrated gas column density ρgas:
(13)
AV (D) =
1
β Z D
0
dD′ ρgas(D′) ,
(14)
where β ≃ 2 × 1021 cm−2 mag−1 is the ratio between gas
density and V -band extinction (Lilley 1955; Bohlin et al.
1978). Suppose now that we carry out observations on the
cloud and measure pAV (AV ). In order to show that we can
recover the structure of the cloud, let us call D(AV ) the
inverse of the function AV (D) defined in Eq. (14), i.e. the
distance at which the integrated extinction is AV . Then
we have
105α
pAV (AV ) =
=
105α
a (cid:2)D(AV )(cid:3)2−5α
ρ(cid:0)D(AV )(cid:1)
3 − 5α
a
d
ρ(cid:0)D(AV )(cid:1)D′(AV )
dAV h(cid:2)D(AV )(cid:3)3−5αi .
(15)
1 Through this paper we assume that the distance of individ-
ual stars is not an observable; if, instead, the distance of each
star can be estimated, then in principle one can also directly
measure pAV (AV D).
4
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
(cid:2)D(AV )(cid:3)3−5α
In other words, we have obtained a differential equation in
. If ρ(D) is known, using the boundary con-
dition D(0) = 0 we can in principle solve this differential
equation and obtain D(AV ) which, in turn, can be in-
verted into AV (D) and ρgas(D). It is superfluous to stress
that this process in reality is far from being trivial.
Finally, we consider a very simple deterministic model
for pAλ(AλD) that describes a thin cloud at a distance
D0 with uniform column density AV :
pAV (A′
V D) =(δ(A′
δ(A′
V )
if D < D0 ,
V − AV ) otherwise .
(16)
The related distribution pAV (AV ) takes then the simple
form
pAV (A′
V ) = f δ(A′
V ) + (1 − f )δ(A′
V − AV ) ,
(17)
where f is a real number in the range [0, 1] given by
f =(cid:20)Z D0
0
dD D2−5αρ(D)(cid:21)(cid:30)(cid:20)Z ∞
0
dD D2−5αρ(D)(cid:21) .
(18)
Hence, f represents the fraction of foreground stars (i.e.
stars at distance D < D0) present in any apparent mag-
nitude bin in regions with negligible extinction. In the
following, we will refer to the simple case described in
Eqs. (16) and (17) as "thin cloud approximation" (the
term is borrowed from gravitational lensing theory).
2.3. The completeness function
Above, in Eq. (2), we introduced the completeness func-
tion c(m), which gives the probability to detect a star of
magnitude m. Typically, c(m) is close to unity for bright
stars, and vanishes for very faint stars; note however that
c(m) might be smaller than unity at low m in crowded
fields or close to bright objects. The completeness func-
tion c(m) enters naturally in the definition of the error
probability distribution pǫ( mm), which can be written as
pǫ( mm) =(c(m)pǫ( mm)
if m 6= null ,
if m = null .
(19)
1 − c(m)
This representation reflects the measurement process: for
any star, there is first a random process that "decides"
whether the object is detected or not (with probability
fixed by c(m)); then, if the star is detected, there is a
second random process that generates the observed mag-
nitude m according to pǫ( mm). Note also that this latter
probability distribution is normalized [cf. Eq. (4)].
In order to carry out some simplifications (see below
Sect. 4.2), we also consider a different representation of
the completeness function in which the order of the two
random processes described above is swapped: this leads
to a completeness function c( m) defined in terms of the
observed magnitude. In other words, we first generate for
every star the "observed" magnitude m according to a
probability distribution pǫ( mm), and then decide whether
the star is really observed using a second random process
controlled by the completeness function c( m). This im-
plies, among other things, a modification of Eq. (7), that
now becomes
−∞
pǫ( mm)pm(m) dm .
p m( m) = c( m)Z ∞
This equation is valid for detected stars, i.e. if m 6= null.
Since c( m) represents the probability that a star with mea-
sured magnitude m be detected, we evaluate the proba-
bility that the star is not detected as
(20)
p m(null) =Z ∞
−∞
d m(cid:2)1− c( m)(cid:3)Z ∞
−∞
dm pǫ( mm)pm(m) .
(21)
An important point to observe here is the fact that we have
modified Eq. (7) into Eqs. (20) and (21) by defining two
functions, the new error distribution pǫ( mm) and the new
completeness function c( m), which replace, respectively,
pǫ( mm) and c(m). Indeed, both these couples of functions
are intimately related, and this was implicitly taken into
account above by using a single distribution pǫ( mm) to
describe both the photometric errors and the completeness
of the observations. Interestingly, it is possible to find a
relationship between the couple pǫ( mm), c( m) and the
couple pǫ( mm), c(m) by requiring that, for any reddened
magnitude probability distribution pm(m), the observed
magnitude probability distribution p m( m) evaluated using
Eq. (7), or using Eqs. (20) and (21), agrees. We find then
c(m)pǫ( mm) = c( m)pǫ( mm) ,
1 − c(m) =Z ∞
−∞
pǫ( mm)(cid:2)1 − c( m)(cid:3) d m .
if we integrate
Since pǫ( mm) is normalized to unity,
Eq. (22) over m and substitute the result into Eq. (23),
we find
Z ∞
−∞
pǫ( mm) d m = 1 ,
(24)
i.e. pǫ is also normalized to unity. In summary we have:
(22)
(23)
(25)
(26)
−∞
c(m) =Z ∞
pǫ( mm) =
R ∞
−∞ c( m)pǫ( mm) d m
c( m)pǫ( mm) d m ,
c( m)pǫ( mm)
.
Since Eq. (25) involves a convolution, it is in general not
possible to invert it and express c( m) in terms of c(m).
Note that Eq. (26) is essentially Bayes' theorem applied
to pǫ( mm).
Although Eqs. (25) and (26) show that the description
of the completeness c(m) in terms of the true reddened
magnitude m is more general than the description c( m)
in terms of the observed magnitude m, we argue that the
latter is more practical to use:
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
5
-- Operationally, the completeness function is usually
evaluated from the observed magnitudes by compar-
ing the expected number of stars in a given magnitude
bin with the number of stars detected in the same bin.
-- Often, it can be sensible to use some a posteriori cuts
in the star catalog. For example, if we know that ob-
servations of faint stars are particularly unreliable, we
can just discard all objects with observed magnitude
m larger than a given threshold mlim. This cut could
be easily described in terms of c( m):
c( m) = H(mlim − m) =(1 if m ≤ mlim ,
0 otherwise .
(27)
-- The use of c( m) allows us to evaluate analytically
p m( m) for some special simple probability distribu-
tions. This point is particularly valuable for the practi-
cal applications that we will describe below in Sect. 4.
2.4. Multi-band probability distributions
So far we have focused our analysis on single-band mea-
surements. In order to extend the discussion to observa-
tions carried out in different bands, we need to generalize
the relevant probability distributions.
We denote M = {M1, M2, . . . , MΛ} the magnitudes
of a star in Λ different bands; in general, we use bold
symbols such as k to indicate quantities that have different
values in the various bands. We will threat these as vector
quantities; in this way, e.g., the generalization of Eq. (1)
can written as
mn = M n + 5 log10 Dn− 5 + kAV + ǫn = mn + ǫn . (28)
Since we are now working with observations in differ-
ent bands, we need also to generalize ρ(M, D). We define
thus ρ(M , D), the probability to have a star with absolute
magnitudes M at distance D from us. Using this distri-
bution we can rewrite Eq. (6) as
pm(m) =Z ∞
0
dD D2Z ∞
× ρ(m − 5 log10 D + 5 − kAV , D) .
dAV pAV (AV D)
0
(29)
The generalization of Eq. (7) is also simple. Since
photometric measurements in different bands are inde-
pendent, we have to perform Λ different convolutions
with the measurement error probability distributions
(cid:8)pǫλ( mλmλ)(cid:9) corresponding to the various bands:
p m( m) =Z dΛm pm(m)
pǫλ( mλmλ) .
Λ
Yλ=1
(30)
Note that we consider a star detected if it is detected in
at least one band.
The integral of p m( m) over all admissible values of m
gives the expected local density of stars σ:
Since σ gives the normalization of p m( m), the conditional
probability that a star with magnitudes m be observed
given the fact that the star is detected is p m( m)/σ.
When using the alternate completeness functions
(cid:8)cλ( mλ)(cid:9) expressed in terms of the observed magnitudes
{ mλ}, Eq. (30) can be rewritten as
Yλ=1hδ( mλ − m′
p m( m) = Z dΛm′
λ)cλ( m′
λ)
Λ
λ)(cid:1)i
+ δ( mλ − null)(cid:0)1 − cλ( m′
Λ
(32)
pǫλ( m′
λmλ) .
×Z dΛm pm(m)
Yλ=1
The combination of delta distributions inside the brackets
in this equation ensures that m′
λ = mλ if mλ 6= null, and
b)(cid:1) (the probability of not
detecting the star) if mλ = null; the last integration is the
usual convolution with the measurement errors. Similarly,
Eq. (31) can also be written as
that we integrate over(cid:0)1− cλ( m′
σ = Z dΛm(cid:20)1 −
Λ
Yλ=1(cid:16)1 − cλ( mλ)(cid:17)(cid:21)
Λ
×Z dΛm pm(m)
pǫλ( mλmλ) .
Yλ=1
(33)
2.5. Further simplifications
As for the single-band case, we consider a number of re-
alistic and useful simplifications that will allow us to take
advantage of the formalism introduced above in practical
cases.
First, we still take ρ(M , D) to be separable into
ρ(M , D) = pM (M )ρ(D). Furthermore, we adopt for
pM (M ) a simple functional form, which is sufficiently ac-
curate for our purposes. Since we know that pm(m) ∝
10αm with α approximately independent of the band, and
since stars appear to have a limited scatter in their NIR
colors, we write
pM (M ) = ν10αM1
× exp(cid:20)−
(Ma − M1 − χa)C−1
ab (Mb − M1 − χb)
2
(cid:21) ,
(34)
where we have used Einstein's convection on repeated in-
dexes. In other words, we suppose that M1 is exponen-
tially distributed, and that star colors are distributed as
Gaussian random variables with averages hMa−M1i = χa;
C represents the covariance of these colors. This distribu-
tion can be rewritten in a more convenient way as (see
also below Sect. 4.2)
σ =Z(R∪{null})Λ\{null}
p m( m) dΛm .
(31)
pM (M ) = exp(cid:20)−
M TP M + 2QTM + R
2
(cid:21) .
(35)
6
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
The form (34) for pM (M ) is particularly convenient for
several reasons. First, we can greatly simplify Eq. (29) and
obtain a result similar to Eq. (10):
pm(m) = aZ ∞
0
dAV pAV (AV )pM (m − kAV ) ,
(36)
where pAV (AV ) and a are still given by Eqs. (11) and
(12). Again, we observe that the fact that pm(m) de-
pends only on the distance-weighted pAV (AV ), implies
that pAV (AV D) is not an observable. Moreover, the forms
(34) and (35) are unmodified by a reddening M 7→ m =
M + 5 log10 D − 5 + kAV and by a convolution with
Gaussian photometric errors (see below Sect. 4.2).
3. Extinction measurements
We will describe in this section the star counts and color
excess methods in order to better describe the advantages
and limitations of them.
3.1. Star counts
In the 18th century the English astronomer William
Herschel noted that some regions presented few stars and,
following Newton's idea of a perfectly transparent space,
interpreted these "holes in the sky" as a real lack of stars.
This misconception survived the discovery of individual
dark clouds (Barnard 1919) and had serious consequences
on Shapley's calibration of the distance scale for Cepheids.
The "discovery" of dust in our Galaxy took place only in
1930 when Trumpler showed its importance in dimming
the light coming from distant open clusters. Finally, in
recent decades dust has no longer been seen only as an-
noying "fog" obscuring the light of background sources,
and has been shown to have a tremendous impact on the
evolution of galaxies and on the formation of stars and
stellar systems (see Li & Greenberg 2003 for a detailed
historical review).
It has long been recognized (Wolf 1923) that measure-
ments of the local density of stars in different regions of
the sky can be used to map the extinction. The original
technique consisted in comparing the number of stars in
magnitude bins in regions subject to extinction with the
number of stars in regions where the extinction is (sup-
posedly) negligible. This technique was then improved by
Bok (1956) which suggested to use counts up to a limiting
magnitude to reduce the error. In the past, the star count
technique was mainly applied to optical data (typically vi-
sually inspected Smith plates; e.g. Dickman 1978; Mattila
1986; Andreazza & Vilas-Boas 1996). In the last decade,
however, near-infrared digital data have been available,
and the star count technique has been finally applied
to NIR observations (e.g. Cambresy et al. 1997). In this
respect, a key role has been played by large NIR sur-
veys such as the Two Micron All Sky Survey (2MASS;
Kleinmann et al. 1994) and the Deep NIR southern Sky
Survey (DENIS; Epchtein et al. 1997).
λ
J
H
K
L
M
N
kλ
kλ
(Rieke & Lebofsky 1985)
(van de Hulst 1946)
0.282
0.175
0.112
0.058
0.023
0.052
0.246
0.155
0.089
0.045
0.033
0.013
Table 1. The extinction law in different infrared bands
(taken from Rieke & Lebofsky 1985 and van de Hulst
1946).
The star count method is easily described using our
notation. We first note that, as for multi-band probability
distributions, the integral of p m( m) over m gives the local
density of stars σ [cf. Eq. (31)]:
σ ≡ hσi = Z p m( m) d m
= Z d m c( m)Z dm pm(m)pǫ( mm) .
Inserting here Eq. (10), we immediately obtain
σ = σ(0)Z dAV pAV (AV )10−αkλAV ,
(37)
(38)
where σ(0) is the average density expected in absence of
extinction:
σ(0) = aνλZ d m c( m)Z dm 10αmpǫ( mm) .
(39)
Equation (38) shows that the ratio σ/σ(0) between the
average densities expected in presence and in absence of
extinction is simply related to pAV (AV ).
Clearly, a single measurement of σ/σ(0) cannot be used
to derive the distribution pAV (AV ). However, in the sim-
plest case where all stars are background to a thin cloud,
so that pAV (A′
V ), we find the classical re-
lation
σ
σ(0) = 10−αkλAV ,
V ) = δ(AV − A′
(40)
which can be immediately inverted to obtain AV (Bok
1956). Operationally, both densities σ and σ(0) are mea-
sured by dividing the number of stars observed against the
cloud and in a control field by the angular size of the re-
gions considered. Hence, the measurement errors on these
quantities are due to the randomness on the local number
of stars. If we ignore the correlation on the star positions,
and thus assume that these are a homogeneous Poisson
process (see, e.g., Cressie 1993), then the local number of
stars follows a simple Poisson distribution.
Although the estimator (40) can be applied to thin
clouds only, in principle more general situations can also
be handled by taking advantage of the particular form of
Eq. (38). We first observe that the integral appearing in
the r.h.s. of this equation is reminiscent of a Laplace's
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
7
transform. Given a real function f (x) defined for non-
negative values x, its Laplace's transform is defined as
L[f ](k) =Z ∞
0
f (x) exp(−kx) .
(41)
An important property of Laplace's transform is that it
can be inverted: in other words, it is possible to obtain
f (x) provided one knows L[f ](k) for any non-negative
k. Using the notation just introduced, we can rewrite
Eq. (38) as
σ = σ(0)L[pAV ](αkλ ln 10) .
(42)
Hence, if we were able to measure the ratio σ/σ(0) for all
positive values of the product αkλ, we could in principle
derive pAV . As indicated above, the constant α ≃ 0.34 is
approximately independent of the band considered; how-
ever, kλ strongly depends on the band used to carry out
the observations (see Table 1). Hence, the product αkλ
that appears in Eq. (42) can be varied among a limited
set of values. In conclusion, although complete knowledge
of pAV is clearly impossible, we can still use multiband ob-
servations to investigate pAV in more complex situations
than the one considered in Eq. (40).
Let us consider, for example, the case of the thin cloud
approximation described in Eq. (17). Since, in general,
the fraction f of foreground stars is not known, we want
to estimate both AV and f in a given patch of the sky.
[Although f can be taken to be constant in many cases,
changes on this quantity have to be expected for different
regions of large cloud complexes because of the geometry
of the cloud and of changes of star densities due to the
Galactic structure.] For this, we insert Eq. (17) in Eq. (38),
thus obtaining a simple generalization of Eq. (40):
σ
σ(0) = f + (1 − f )10−αkλAV .
(43)
Hence, we can in principle use two different measurements
of ¯σ in two different bands to deduce both f and AV on
the region considered. In practice, it is preferable to follow
the approach described below in Sect. 4.
It is interesting to investigate in more detail the esti-
mator derived from Eq. (43), i.e.
AV = −
1
αkλ
log10(cid:20) N − Aσ(0)f
Aσ(0)(1 − f )(cid:21) ,
(44)
where N is the number of stars found in the region inves-
tigated, and A is its area. Assuming that σ(0) is measured
without significant errors (this is possible by using a large
control field), we can deduce the expected error on AV by
noting that N follows a Poisson distribution with average
[cf. Eq. (43)]
hNi = Aσ(0)(cid:2)f + (1 − f )10−αkλAV(cid:3) .
Using a first order approximation (valid for small relative
errors) we find
(45)
Var(cid:0) AV(cid:1) =(cid:18)
1
αkλ ln 10 ·
1
hNi − Af σ(0)(cid:19)2
hNi .
(46)
1
0.8
0.6
0.4
0.2
0
y
t
i
l
i
b
a
b
o
r
p
−5
0
5
10
15
20
AV
(from the
Fig. 1. The
cumulative probability distribution for
AV . The various curves are relative to a true ex-
tinction AV = [0, 2, 4, 6, 8, 10]
to
the right); the average values measured are instead
h AV i = [0.24, 2.33, 4.47, 6.64, 8.79, 10.76] (the scatters
around these values ranges from 1.9 to 4.3). In all cases we
assumed α = 0.34, kλ = 0.175, f = 0.1, and Aσ(0) = 20.
Along the curves the dots mark the actual possible val-
ues of AV that can be measured for different values of the
observed number of stars N .
left
4
2
0
)
V
A
(
r
r
E
,
V
A
−
i
V
A
h
−2
−4
−6
−8
Aσ(0) = 10
Aσ(0) = 20
Aσ(0) = 50
0
5
10
AV
15
20
Fig. 2. The bias (thick lines) and error (thin lines) on AV
evaluated using Eq. (44) for different values of Aσ(0). For
this figure we used the same parameters as in Fig.1, i.e.
α = 0.34, kλ = 0.175, and f = 0.1.
In case of vanishing f (i.e., if all stars are background to
the cloud), this expression takes a simpler form
Err(cid:0) AV(cid:1) =qVar(cid:0) AV(cid:1) =
1
1
αkλ ln 10
phNi
=
11.4
,
phNi
(47)
where the last expression is valid for the K band (cf.
Tab. 1).
Further statistical properties of the estimator (44) can
be better evaluated using numerical methods. Figure 1
shows, as an example, the expected cumulative probabil-
8
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
ity distributions for AV for some typical cases. As ex-
pected, the cumulative distributions reach 0.5 around the
true value for AV , but the expected measurement errors
can be very large, especially at high column densities.
The numerical analysis of Eq. (44) also shows that the
estimator is significantly biased. In general, in the limit
σA ≫ 1, the estimator is biased toward large values of
AV , i.e. (cid:10) AV(cid:11) > AV ; the opposite happens for σA ∼ 1
(see Fig. 2). This change in behavior can be understood
by observing that, if AV is large, then Aσ is small and
thus we simply do not have enough background stars to
probe high column densities.
Interestingly, Eq. (46) is a reasonably good approxi-
mation of the true variance of AV . Typically, this approx-
imation slightly underestimates the true variance of AV ,
but again the opposite happens for low values of σA.
3.2. Near-infrared color excess (NICE and NICER)
The dust present in molecular clouds produces differ-
ent amounts of obscuration in different color bands (see
Table 1), so that background stars appear reddened.
Hence, the color excess in the red part of the spectrum
of background stars can be used to measure the extinc-
tion along the line of sight.
Historically, the color excess technique has been im-
peded by the limited instrumental sensitivity to small re-
gions (Jones et al. 1980; Frerking et al. 1982; Jones et al.
1984). However, in the early 1990s, with the advent of
near-infrared arrays, it became possible to accurately mea-
sure the NIR magnitudes of many stars from single point-
ing observations. This new technology was first exploited
by Lada et al. (1994), and since then has been successfully
applied to many molecular clouds (e.g. Horner et al. 1997;
Lada et al. 1999; Alves et al. 2001).
Lada's technique (called "near-infrared color excess"
or Nice) is based on measurements of a NIR color (e.g.,
H − K) of many stars. Since stars have relatively well
defined colors in the infrared, a significant intervening ex-
tinction can be detected as a reddening. Note that a key
point of the Nice method (and of similar methods based
on the reddening of stars; see, e.g., Schultheis et al. 1999)
is the assumption that all stars belong to a homogeneous
population.
The Nice method can be quantitatively described us-
ing the notation introduced so far. In particular, from
Eq. (29) we have
pm(m) = p(0)
m (m − kAV ) ,
(48)
where we have denoted p(0)
m (m) the l.h.s. of Eq. (29) for
AV = 0. Naively, Eq. (48) can be used to obtain a simple
estimate of the intervening infrared extinction AV . For
example, we find
hmi = hmi(0) + kAV ,
(49)
where again we used the superscript (0) to denote quan-
tities measured in a control region where AV = 0.
Equation (49) is better rewritten in terms of star colors.
Calling χλ = mλ − m1 and κλ = kλ − k1, we have
hχi = hχi(0) + κAV .
(50)
This equation, applied to a single color, is essentially the
Nice technique. More precisely, this technique uses the
simple average of a set of N angularly close stars to eval-
uate the column density:
AV =
1
N
N
Xn=1
χn − ¯χ(0)
κ
,
(51)
where ¯χ(0) is an estimate of hχi(0) (obtained, e.g., by mea-
suring the star colors on a control field where presumably
AV ≃ 0). As an example, consider the Nice method ap-
plied to the χ = H − K color. Since stars have a typical
scatter of 0.09 mag in this color, we expect an error on AV
of Err(cid:0) AV(cid:1) ≃ 1.4/√N . Hence, even in the presence of sig-
nificant photometric errors, the Nice method gives signif-
icantly more accurate results than the star count method
[cf. Eq. (47)].
As shown by Lombardi & Alves (2001), one can indeed
take full advantage of observations carried out in differ-
ent bands to obtain more accurate column density mea-
surements. The improved technique, called Nicer (Nice
Revised) optimally balances the information from different
bands and different stars. As a by-product of the analysis,
Nicer also allows us to evaluate the expected error on the
column density map, which is useful to estimate the sig-
nificance on the detection of substructures and cores. The
Nicer technique can be described using the following sim-
ple argument. Equation (50) written above can be taken
to be a system of (Λ − 1) equations to be approximately
solved for AV , the approximation being made necessary
because we can only measure hχi and hχi(0) with limited
accuracy. The "best" solution for AV can been obtained
by minimizing the chi-square quantity
χ2 =
N
Xn=1(cid:2) χn− ¯χ(0)−κAV(cid:3)T
(C +E)−1(cid:2) χn− ¯χ(0)−κAV(cid:3) ,
(52)
Consistently with the notation used above in Eq. (34), we
called C the covariance matrix of the star colors; moreover,
the symbol E was used to denote the covariance matrix of
measurement errors [the two covariance matrices have to
be added up in Eq. (52) in order to properly estimate the
expected scatter on star colors]. The best estimate of AV ,
obtained by minimizing the χ2 of Eq. (52), is precisely the
Nicer estimator.
Both the Nice and Nicer estimators appear to be
unbiased provided that: (i) there are no foreground stars
and (ii) the measured colors χn are unbiased estimates of
the true colors χn.
In reality, even if the two conditions considered above
are satisfied, both color excess methods can still be biased
because of selection effects introduced by the completeness
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
9
0.15
0.1
i
V
A
h
−
V
A
0.05
0
0
30
20
i
V
A
h
10
0
f = 0.01
f = 0.03
f = 0.1
f = 0.3
10
AV
20
30
Fig. 3. The bias for the Nice method due to the different
completeness at low and high column densities. This plot
is evaluated by generating 100 000 stars for various values
of AV (marked as dots). Note that, as described in the
text, for AV ≃ 6.7 we observe a rapid increase in the bias
due to the change between a selection in the K band and
a selection in the H band (see Fig. 4). The small scale
oscillations on the plot are due to numeric effects.
0
10
20
AV
30
40
Fig. 5. The bias for the Nice method due to the contam-
ination by foreground stars. Solid lines represents, for dif-
ferent fractions of foreground stars, the median of the dis-
tribution of the measured column densities. Dashed lines,
instead show the expectation value for the median of AV
obtained from the measurements of 15 reddenings.
H
=
1
5
.
5
5
.
5
1
=
K
AV = 12
K
−
H
AV = 6
AV = 0
K
Fig. 4. A graphic explanation of the bias plotted in Fig. 3.
At low column densities, we select stars mainly on their
K magnitude, while at high column densities we mainly
select using their H magnitude. As a result, the average
intrinsic color (dotted line) of the observed stars (gray
bands) changes toward lower H − K values.
function. To understand this point let us make a simple
example. Suppose that we carry out our observations in
two bands, λ1 and λ2, and that both completeness func-
tions cλ(mλ) are not vanishing only on a narrow magni-
tude range. In this case we would always have χ ≃ χ(0)
(because a star is observed in both bands only if m1 and
m2 are inside the narrow detection window), and thus we
would always measure AV ≃ 0, independently of the real
column density. Although unrealistic, this example shows
that we must expect a bias for the Nice method; the ar-
gument for the Nicer method is similar and leads to the
same conclusion.
The bias of these two methods depends on the details
of the probability distribution pM (M ) and of the com-
pleteness functions cλ(mλ). However, as an example, we
evaluated the expected bias for various values of the col-
umn density and for the typical probability distributions
that we expect for the 2MASS catalog. As shown in Fig. 3,
the bias in the case considered appears to be limited, be-
low 0.2 in AV , and has a characteristic shape: it vanishes
for AV = 0, increases quickly for AV ∼ 7, and finally
saturates for AV > 11. This behavior has a simple expla-
nation:
-- Since the colors of unreddened stars are evaluated us-
ing a control field (where supposedly AV = 0), the bias
has to vanish for AV = 0.
-- The general trend of bias on AV can be understood
with the help of Fig. 4. At low AV , stars that are ob-
served in the K band are almost certainly also observed
on the H band, because of the values of the limiting
magnitudes (approximately 14.9 in H and 14.3 in K)
and of the average star colors (hH−Ki = 0.18). When,
instead, the reddening is large, we have the opposite
situation (because K is less affected by reddening than
H). This different selection at different column densi-
ties is the source of the observed bias.
-- Finally, at large column densities an asymptotic value
is reached because now only the H band is used to
select stars.
A more serious problem is related to the contamina-
tion by foreground stars, which can strongly bias our re-
sults in the direction of lower column densities. For regions
10
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
0.23
0.96
y
t
i
l
i
b
a
b
o
r
p
e
v
i
t
a
l
e
r
9.31
0
10
20
30
AV
Fig. 6. The effect of a foreground star contamination
f = 0.1 on the distribution of column densities for differ-
ent values of AV . From the bottom to the top we show, for
AV ∈ {0, 10, 20, 30}, both the distribution of background
stars (solid plot, gray filled), and foreground stars (dashed
plot). Both distributions are taken to be Gaussian, the
first centered on AV , and the second on 0. As the col-
umn density increases, the relative contribution of fore-
ground stars also increases because of selection effects. At
AV ≃ 16, they become predominant and the median of the
whole distribution quickly moves toward the foreground
distribution.
with low extinction, where the expected number of fore-
ground stars is small compared to the expected number of
background stars, foreground star contamination is usu-
ally reduced by using a median estimator for AV , i.e. by
averaging the individual column densities measured in the
direction of each star with the median instead of the sim-
ple mean (e.g. Cambr´esy et al. 2002; Lombardi & Alves
2001). As shown in Fig. 5, this simple technique is very
effective and leads to almost unbiased results at low col-
umn densities (see Appendix A for a statistical discussion
on the median estimator). However, for a given fraction of
foreground stars f (which, as described above, is evaluated
in regions with negligible extinction) there is a threshold
for AV after which the cloud extinction makes the ex-
pected number of background stars smaller than the ex-
pected number of foreground stars. This threshold can be
evaluated using Eq. (43) and is
AV =
1
αkλ
log10
1 − f
f
.
(53)
Because of the way the median estimator is defined, we
observe in Fig. 5 an abrupt change on the measured col-
umn density AV close to this value (see also Fig. 6). Note
that the relatively smooth transition observed in Fig. 5
(solid lines) is due to the intrinsic scatter in the star col-
ors; indeed, in absence of any scatter, we would observe
an "instantaneous" change from h AV i = AV at low col-
umn densities to h AV i = 0 at high column densities. In
Fig. 5 we also plot the expectation value of a more inter-
esting quantity, the median over N = 15 measured col-
umn densities. This quantity differs from the median over
the whole distribution because of the statistical variations
on the local number of foreground stars. In other words,
since N = 15 is a relatively small number, in different
realizations of our simulations we can have a significantly
different number of foreground stars. For example, even at
low column density, we can have a large fraction of fore-
ground stars; similarly, even at very large column density,
there is a finite probability to have the majority of stars
in the background. As a result, the dashed line in Fig. 5
has a much smoother transition around the value given
by Eq. (53). This is at the same time a good and a bad
news. From one side, this means that we can be signifi-
cantly biased for values of AV smaller than the threshold
value of Eq. (53); from the other side, this also means that
we can still partially make use of the median estimator at
relatively large column densities.
For regions with very high column density, it is nor-
mally quite easy to identify and remove foreground stars
because of their anomalous colors with respect to the red-
dened background stars (see, e.g., Alves et al. 2001). Some
authors (e.g. Cambr´esy et al. 2002) make use of this infor-
mation to also remove stars in less dense regions using the
following strategy. The angular density σf of foreground
stars is determined using dense regions (where the fore-
ground/background identification is easy); then, for any
region of the cloud, the k bluer stars are thrown away,
where Nf = σfA is the expected number of foreground
stars in the analyzed region (deduced from the foreground
density measured in the dense regions). This technique is
quite simple and reasonably effective, but unfortunately
introduces a bias at low extinctions. Consider, indeed,
the limiting case where we have a negligible extinction
AV ≃ 0. The distinction between foreground and back-
ground stars is in this situation ambiguous, and thus the
Nf bluer objects will likely include also some background
stars. Hence, the results will be biased toward higher col-
umn density. The exact evaluation of the bias is non triv-
ial, but can be carried out using the techniques described
in App. A. Here we report only an approximated result
valid for AV ≃ 0 [see Eq. (A.15)],
(54)
Nf
2N
,
√2πErr(cid:0) AV(cid:1)
h AV i ≃
where Err(cid:0) AV(cid:1) is the average error of the measured ex-
tinction from a single star. Note that the result given in
Eq. (54) can be taken to be an upper limit for the bias,
since we expect this to decrease at high column densities,
where the identification of foreground stars is secure.
4. A maximum-likelihood approach
From the discussion above, it is clear that both the star
count and the Nice(r) methods can produce unsatisfac-
tory results. On one hand, the star count technique has a
low signal-to-noise ratio and produces significantly biased
results at high column densities (see Fig. 2). On the other
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
11
hand, the color excess technique is more sensitive and has
a smaller bias, but can be severely affected by the con-
tamination of foreground stars, especially for large values
of AV .
As pointed out by Cambr´esy et al. (2002), we can take
advantage of the different behavior of the star count and
the color excess techniques in the various column density
regimes to partially solve the problem of the contamina-
tion by foreground stars. In particular, Cambr´esy et al.
note that it is better to use the color excess method at
low column densities because of its higher sensitivity, and
to switch to the star count method at very large column
densities where the Nice method is completely unreliable.
The optimal "turning point" where we need to change the
technique can be evaluated empirically by comparing the
individual results of the two methods at different column
densities.
The solution proposed by Cambr´esy et al. has the sig-
nificant advantage of being relatively simple to implement
and reasonably effective, but clearly is suboptimal in many
respects:
-- The choice of the "turning point" and of the matching
strategy is to some extent arbitrary.
-- The overall estimate of AV still remains significantly
biased at high column densities because at large AV
the star counts are used (and this method has a non-
negligible bias, see Fig. 2).
-- The density of foreground stars must still be deter-
mined separately and is taken to be constant on the
whole field.
Using the theoretical framework developed so far, it is
possible to design and implement an efficient maximum-
likelihood approach to the problem.
4.1. Likelihood
Suppose again that in a region of the sky of area A we
observe in various bands N stars with magnitudes { mn}.
Assuming that the area of the sky is small enough so that
there are no significant changes on the relevant parameters
of the problem (such as the unreddened density σ(0), local
expected density σ, or the reddening probability distribu-
tion pAV (AV )), then we can easily evaluate the joint prob-
ability distribution for such a star configuration. First, we
note that the number of stars inside the region follows a
Poisson distribution with average Aσ:
pN (N ) = e−Aσ (Aσ)N
N !
.
(55)
The joint star probability distribution, i.e. the likelihood,
is given by
L(cid:0){ mn}(cid:1) = pN (N )
=
e−AσAN
N !
p m( mn)
σ
p m( mn) .
N
Yn=1
Yn=1
N
Note that L depends on unknown quantities, such as
pAV (AV ), through pN (N ), p m( m), and σ: hence, assum-
ing that there is no prior for these unknown quantities,
we can obtain an estimate of them by maximizing L or,
equivalently, ln L. In the following subsections we will in-
vestigate in more detail this maximum-likelihood estima-
tor.
4.2. Implementation
We implemented the maximum-likelihood estimator using
the simplification described above. In particular, we used
the forms (34) and (35) for the source magnitude proba-
bility distribution pM (M ). A simple calculation gives the
following relationships between the coefficients of Eq. (34)
and Eq. (35):
Pab = C−1
ab − δ1aXa′
+ δ1aδ1bXa′,b′
C−1
a′b − δ1bXb′
C−1
a′b′ ,
C−1
ab′
(57)
(58)
C−1
a′bχb ,
ab χb + δ1aXa′
Qa = −δa1α ln 10 − C−1
R = −2 ln n + C−1
ab χaχb .
One of the advantages of the quadratic expression (35)
is that we can write the effects of reddening as a simple
transformation of the three quantities P , Q, and R. In
particular, the transformation M 7→ M + kAV can be
rewritten as
P 7→ P ,
R 7→ R − 2AV QTk + A2
Q 7→ Q − AV P k
V kT P k .
(60)
(59)
Hence, reddening does not change the functional form of
the probability distribution (35) but only the three pa-
rameters involved.
For the following discussion, it is also convenient to in-
troduce a new vector µ ≡ (M , 1) = (M1, M2, . . . , MΛ, 1),
and a (Λ + 1) × (Λ + 1) matrix S defined as
P11 ··· P1Λ Q1
...
...
PΛ1 ··· PΛΛ QΛ
Q1 . . . QΛ R
QT R(cid:19) =
S−1 =(cid:18) P Q
Then we can write the quadratic form appearing in
Eq. (35) simply as
(61)
. . .
...
.
pM (M ) = exp(cid:20)−
µTS−1µ
2
(cid:21) .
(62)
More importantly, the action of measurement errors can
be described as simple transformations of S, provided that
the measurement error on each band can be described as a
simple Gaussian with vanishing average and variance v. In
other words, assuming that the errors can be represented
as a convolution with the Gaussian kernel
(56)
pǫ(ǫ) =
1
(2π)Λ/2QΛ
λ=1 vλ
exp(cid:20)−
Λ
Xλ=1
ǫ2
λ
2v2
λ(cid:21) ,
(63)
12
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
then the convolution pM ∗ Σ can be written as
(pM ∗ Σ)(µ) =r det S
(cid:21) ,
exp(cid:20)−
µTT −1µ
2
where T = S + diag(v2
det T
1, v2
2, . . . , v2
Λ, 0).
(64)
In conclusion, the simple implementation of the
maximum-likelihood method considered here can be sum-
marized in the following items:
1. The quantities P , Q, and R appearing in Eq. (35) are
determined in a control field free from any extinction.
Similarly, for each band the photometric errors, i.e. the
functions pǫλ( mλmλ), are calculated.
2. A cloud model (for example, the thin cloud approx-
imation) is chosen, and the relevant parameters (for
example, AV and f ) are taken to be unknown and
constant on the field.
3. For a given combination of the cloud parameters, the
likelihood function (56) is evaluated using the vari-
ous observed magnitudes. Note that the evaluation of
the expected star density σ and of the star probabil-
ity p m( m) when the star is undetected in one or more
bands is non-trivial and requires integrations over the
completeness functions [cf. Eq. (32)].
4. The last step is repeated for different values of the
cloud parameters and the ones corresponding to the
minimum of the likelihood function are taken as best
estimates.
5. The local curvature of L (i.e., the matrix of its second
derivatives) is used to estimate the errors on the cloud
parameters.
Clearly, one key point here is the speed of the likeli-
hood function, which needs to be evaluated several times
in our maximum-likelihood approach. In our implementa-
tion, the likelihood function has been optimized by per-
forming the relevant integrations (cf. point 3 above) using
appropriate bounds. In other words, when an integration
of p m( m) was requested, we estimated the area in the
magnitude space where this function was significantly dif-
ferent from zero, and performed the integral inside that
area (as opposed to performing the integral over the whole
parameter space). This optimization was found to have a
significant impact on the overall speed of our implemen-
tation.
4.3. Simulations
reliability and effectiveness of
The
the maximum-
likelihood approach were assessed through extensive nu-
merical simulations. The simulations were designed to
reproduce with reasonable accuracy the 2MASS near-
infrared data. We simulated star observations in three
bands, J, H, and K, and used the various parameters
described in Table 2.
We initially considered an area of the sky A and a thin
cloud characterized by a fraction of foreground stars f [cf.
Eq. (18)] and a reddening AV . We randomly generated
there stars inside this area using the following recipe:
2(cid:2)ln L − ln Lmin(cid:3)
150
100
50
0
6
8
AV
10
12
14
1.0
0.8
0.0
0.2
0.6
0.4
f
Fig. 7. Log-likelihood surface plot. The plot shows the
logarithmic of the likelihood ratio as a function of the two
unknown parameters AV and f (the real values used for
the simulation are 10 and 0.2 respectively). The simula-
tion has been carried out using σA = 25, but the actual
number of stars generated were N = 19 (because of the
Poisson noise on the number of stars). On the bottom
we also plot contours corresponding to 68.2%, 95.5%, and
99.7% confidence level regions. Note that the likelihood is
very smooth and has only a single well defined minimum.
1. We evaluated the expected local star density σ using
Eqs. (31) and (32). We found that the needed inte-
grations could be performed more efficiently using a
Monte-Carlo technique.
2. We calculated the effective number of stars N by gen-
erating a random integer distributed according to a
Poisson distribution with average Aσ.
cedure:
(a) We generated the unreddened magnitudes in the
3. For each of the N star we adopted the following pro-
three bands according to Eq. (34).
(b) We then uniformly generated a random number in
the range [0, 1], and considered the star to be in
Par.
α
χ1
χ2
C11
C12
Value Description
0.34
0.18
0.82
0.0078 Variance h(H − K − χ1)2i
0.0112 Covariance
Slope of number counts
Average color hH − Ki
Average color hJ − Ki
h(J − K − χ1)(H − K − χ2)i
C22
k1
k2
k3
mlim
mlim
mlim
3
v1
2
1
v2
v3
0.0375 Variance h(J − K − χ2)2i
0.112
Reddening law in K band
Reddening law in H band
0.175
Reddening law in J band
0.282
Limit magnitude in K band
14.3
Limit magnitude in H band
14.9
15.8
Limit magnitude in J band
Average photometric error
0.05
h( m1 − m1)2i1/2 on K band
Average photometric error
h( m2 − m2)2i1/2 on H band
Average photometric error
h( m3 − m3)2i1/2 on J band
0.05
0.05
ref. Eq.
(34)
(34)
(34)
(34)
(34)
(34)
(8)
(8)
(8)
(20) & (27)
(20) & (27)
(20) & (27)
(20)
(20)
(20)
Table 2. The common parameters used for all numerical
simulations.
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
13
30
(cid:3)
20
n
i
m
L
n
l
−
L
n
10
l
(cid:2)
2
ML
Nice (mean)
Nicer (mean)
Nice (median)
Nicer (median)
0.2
0.1
0
V
A
−
i
V
A
h
−0.1
−0.2
0
6
8
10
AV
12
14
Fig. 8. The log-likelihood (ratio) as a function of AV for
different source densities, extremized for f . The largest
curve is obtained using σA = 10 (in this particular re-
alization we had N = 8); the mid curve using σA = 20
(N = 12), and the most peaked curve using σA = 40
(N = 35). In all cases the true value of the column den-
sity was AV = 10, and the fraction of foreground stars was
f = 0.2. Note that all curves are very well approximated
by parabolas. The intersections of the log-likelihood func-
tions with the three dashed lines mark the 68.2%, 95.5%,
and 99.7% confidence level regions.
the foreground if this number was smaller than f
[defined according to Eq. (18)].
(c) Background stars were reddened by adding kλAV
to each magnitude; the magnitudes of foreground
stars were left unchanged.
(d) We added photometric errors to all stars; these, for
simplicity, were taken to be Gaussian distributed
with standard deviation vλ = 0.05 independent of
the band and of the original magnitude.
(e) For each band, we uniformly generated a random
number in the range [0, 1], and took the star to be
detected in the band if this number was smaller
than the completeness function cλ( mλ). In the
simulations discussed here we used for simplicity
Heaviside functions for the completeness functions.
(f) We finally retained the star if it was detected in
at least one band; otherwise, we repeated all steps
above from point (a).
In summary, at the end of a single star generation we had
for each star the three magnitudes in the bands J, H, and
K (with possibly some magnitudes mλ = null) and the
associated measurement errors vλ.
We then used this dataset to test the reliability and ef-
ficiency of the maximum-likelihood estimator, and to com-
pare it with the other column density estimators consid-
ered in Sect. 3. The maximum-likelihood method was im-
plemented as described in Sect. 4.2, and was tested against
the data generated as described in the items above.
Figure 7 shows the log-likelihood surface plot obtained
in a typical simulation. The surface appears to be very
0
10
AV
20
30
Fig. 9. The bias h AV i − AV of various column density
measurement methods. The bias is evaluated from the av-
erages of 1 000 simulated fields with no foreground contri-
bution (f = 0) and with average number of stars σA = 25.
The maximum-likelihood method, marked in the legend as
ML for brevity, has negligible bias, especially for AV < 20.
smooth with a well defined minimum, an essential con-
dition for the reliability of the maximum-likelihood ap-
proach. Moreover, the log-likelihood function is found to
have approximately a parabolic shape, which further sim-
plifies the interpretation of the results obtained. For exam-
ple, this property allows us to use the likelihood ratio (see,
e.g., Eadie et al. 1971) to draw confidence level regions on
the parameter space (see the contours of Fig. 7).
Figure 8 represents the log-likelihood as a function of
AV for three different datasets. The figure was obtained
by minimizing the log-likelihood function with respect to
f for each value of AV in the range [5, 15], and by plotting
the value of this function.
In order to assess more quantitatively the merits of
the maximum-likelihood method, we compared the statis-
tical properties of various column density estimators. In
particular, we simulated a large number of "observations"
using the technique described above, and we studied the
distribution of the column densities estimated using the
maximum-likelihood method, the Nice, and the Nicer
methods. Simulations were carried out using a thin cloud
with true extinction in the range AV ∈ [0, 30] and with
different values of the foreground fraction f . For the Nice
and Nicer estimators we used both the simple mean and
the median of the individual extinction measured for each
star; moreover, assuming that the density of foreground
stars σf could be determined separately, we discarded in
each simulation the σfA bluer stars, and used only the re-
maining (redder) stars in the analysis. Note that in some
cases, for large values of AV and relatively large values of
f we had no usable star for the Nice and Nicer analysis;
in other words, all stars left after the foreground selec-
tion had only one band available (typically the K band).
In this case we just obtained a lower limit on AV by us-
ing the redder usable star (even if this star was originally
considered to be in the foreground).
14
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
ML
Nice (mean)
Nicer (mean)
Nice (median)
Nicer (median)
0.9
0.8
0.7
0.6
0.5
0.4
0.3
)
V
A
(
r
r
E
0
10
20
30
AV
of various col-
umn density measurement methods, evaluated as in Fig. 9.
Fig. 10. The total error (cid:10)(cid:0) AV − AV(cid:1)2(cid:11)1/2
1
0
V
A
−
i
V
A
h
−1
−2
ML
Nice (mean)
Nicer (mean)
Nice (median)
Nicer (median)
ML
Nice (mean)
Nicer (mean)
Nice (median)
Nicer (median)
)
V
A
(
r
r
E
6
4
2
0
0
10
20
30
AV
Fig. 12. As for Fig. 10, but with f = 0.02.
1
0
V
A
−
i
V
A
h
−1
−2
ML
Nice (mean)
Nicer (mean)
Nice (median)
Nicer (median)
0
10
20
30
AV
Fig. 11. As for Fig. 9, but with f = 0.02.
0
10
20
30
AV
Fig. 13. As for Fig. 9, but with f = 0.05.
Figures 9 and 10 show, respectively, the bias and the
total error obtained from the three methods considered
here for AV ∈ [0, 30] and f = 0. From these plots it is
evident that the maximum-likelihood estimator does not
have any significant bias up to AV = 20 and a very small
one (of the order of 0.1) for larger column densities. Since
for f = 0 we never have foreground stars, the bias of the
Nice and Nicer techniques does not change if we use a
mean or a median estimator. Note also that the bias in
Fig. 9 for these two methods is the one discussed in detail
above (cf. Fig. 3). Regarding the total error, we observe a
steady increase of it for large values of AV . This can be
explained by noting that, although the average number
of stars σA = 25 is kept constant for all column densi-
ties, when AV is large most stars are only detected in the
K band and thus do not provide reddening information.
Figure 10 also shows that the maximum-likelihood method
is clearly superior, although Nicer (with the mean esti-
mator) also performs well. As expected, both median es-
timators are more noisy than the simple mean (which, for
f = 0, is optimal).
The situation changes quite dramatically when f > 0.
Figures 11 -- 16 show the bias and the total error of the var-
ML
Nice (mean)
Nicer (mean)
Nice (median)
Nicer (median)
)
V
A
(
r
r
E
8
6
4
2
0
0
10
20
30
AV
Fig. 14. As for Fig. 10, but with f = 0.05.
ious methods for increasing values of f . A careful study of
these results can reveal several interesting characteristics
of the Nice and Nicer techniques.
Let us initially focus on the bias plots, shown in
Figs. 11, 13, and 15. At low column densities, i.e. for
AV < 10, we find again the bias described in Sect. 3.2
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
15
1
0
−1
−2
−3
−4
−5
V
A
−
i
V
A
h
ML
Nice (mean)
Nicer (mean)
Nice (median)
Nicer (median)
0
10
AV
20
30
Fig. 15. As for Fig. 9, but with f = 0.10.
ML
Nice (mean)
Nicer (mean)
Nice (median)
Nicer (median)
)
V
A
(
r
r
E
12
10
8
6
4
2
0
0
10
AV
20
30
Fig. 16. As for Fig. 10, but with f = 0.10.
and plotted in Fig. 3 (see also above Fig. 9 for the case
f = 0). At median column densities, i.e. for AV ∼ 20 (or
AV ∼ 30 for f = 0.02), the bias can be explained by the
selection effect due to the correction of foreground stars.
As explained above [see Eq. (54)], removing the Nf = σfA
bluer stars introduces a systematic error on the extinc-
tion estimate. This bias is positive (i.e. toward larger ex-
tinctions) and can be as large as 1 mag in AV . Finally,
at high column densities (AV ∼ 30) both methods sys-
tematically underestimate the extinction. This is due to
the heavy contamination by foreground stars present at
these large values of reddening. To better understand this
point, we note that the subtraction of the Nf bluer stars
is a simplistic approximation because the actual number
of foreground stars is not fixed (it is a Poisson random
variable with average Nf ). Depending on the number of
foreground stars with respect to Nf we can have three
different situations:
-- If the number of foreground stars is exactly Nf , then
at high column densities no bias is introduced and the
estimate of AV is correctly performed;
-- For a simulation with a number of foreground stars
smaller than Nf , a small bias toward higher column
densities is expected, because some of the bluer back-
ground stars are discarded;
-- Finally, if the number of foreground stars is underesti-
mated, a very large bias toward smaller column densi-
ties is introduced.
Clearly, for large values of AV the third effect is expected
to dominate. Indeed, Figs. 13 and 15 show that both the
Nice and Nicer methods significantly underestimate the
reddening for large values of AV . In theory, as mentioned
above, when applying the Nice or Nicer technique to
high extinction regions, it should be relatively straightfor-
ward to identify foreground stars by their color, and thus
the bias of these method could be smaller than suggested
by the plot shown here. Note that apparently Nicer
presents a larger bias compared to Nice. This is due to
the larger flexibility of the Nicer method, which is able
to obtain a column density estimate when any two of the
three bands are available. As a result, Nicer is more af-
fected by the contamination by foreground stars described
in the items above. Indeed, our simulations show that if
we force Nicer to use only stars with observed H and K
bands (i.e. essentially the same stars as the ones used by
Nice), its bias and its noise are drastically reduced and
become compatible with the ones of Nice.
Figures 12, 14, and 16 show that the total error of the
Nice and Nicer methods increases very rapidly at large
column densities, where the contamination by foreground
stars is very likely; the maximum-likelihood estimator, in-
stead, has an almost flat error curve. Hence, our novel
method is able to "recognize" the presence of foreground
stars; moreover, the inclusion of the background density in
the likelihood expression allows this estimator to "switch"
to the number count technique in regions with large ex-
tinction. Figure 12, in particularly, shows that even in ex-
treme cases with a large foreground star contamination
the maximum-likelihood method is still very reliable and
accurate. To better appreciate this point, we note that, in
our simulations, for AV = 25 and f = 0.2 on average only
one tenth of the ∼ 25 stars are background to the cloud.
4.4. Limitations
The maximum-likelihood approach to the extinction mea-
surements presents clear advantages with respect to the
standard techniques in the simplified framework consid-
ered in this section (uniform AV over the cloud patch
analysed, thin-cloud approximation, simple model for the
star intrinsic magnitude distribution). One could thus le-
gitimately ask whether the maximum-likelihood technique
is applicable to more realistic and complex situations.
4.4.1. Small-scale inhomogeneities
Most clouds present clear sign of substructure at dif-
ferent scales and a statistical analysis of the radio and
NIR observations seems to indicate the that these ob-
jects can be well described in terms of turbulent mod-
16
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
els (see, e.g. Miesch & Bally 1994; Padoan et al. 1997). In
presence of significant inhomogeneities, most methods (in-
cluding the maximum-likelihood one) are expected to be
biased toward low values because the background stars
will no longer be randomly distributed in the patch of
the sky used to estimate the local extinction value, but
will be preferentially detected in low-extinction regions
[cf. Eq. (43)]. Although a detailed discussion of this effect
is behind the scope of this paper, it is worth considering
the following points:
-- For the HK Nice, for the JHK Nicer, and for the
JHK maximum-likelihood methods, small scale inho-
mogeneities become important when the local varia-
tions of AV on the patch of the sky considered are
of order ∆ = 1/(αk2 ln 10), where k2 = AH /AV
[cf. Eq. (8)]. For typical 2MASS observations we find
∆ ≃ 7.3 mag, and hence all methods should still be ap-
plicable to the analysis of regions with relatively low
column densities (approximately AV < 10 mag).
-- The effect of substructures is studied in detail in a
forthcoming paper (Lombardi 2005), where it is also
presented a method to avoid the bias introduced by
small-scale inhomogeneities. The application this novel
technique to a Nicer map of the Pipe nebula con-
firms the expectations summarized in the last item.
In particular,
for the Pipe nebula the "standard"
Nicer has a negligible small bias (below 0.2 mag) for
AV < 10 mag, while the bias increases dramatically
for larger extinction values (e.g., it reaches 1 mag at
AV = 15 mag). A similar bias behavior is expected
for the maximum-likelihood technique described in this
section.
-- Since small-scale inhomogeneities are believed to be
due to turbulent motions (Larson 1981; Padoan et al.
1997; Heyer & Brunt 2004), the probability distribu-
tion pAV (AV ) is expected to be a log-normal:
(ln AV − M )2
pAV (AV ) =
1
√2πSAV
exp(cid:20)−
2S2
(cid:21) . (65)
Independently from the exact form of pAV (AV ), the
maximum-likelihood approach can be used in this more
general framework to obtain the relevant parameters
of pAV (e.g., M and S in the case of the log-normal
of Eq. (65)); moreover, the parameters estimated are
asymptotically unbiased (see Eadie et al. 1971). We
will consider the use of a non-trivial extinction proba-
bility distributions similar to the one of Eq. (65) in a
follow-up paper.
4.4.2. Star magnitude distribution
In the implementation of the maximum-likelihood tech-
nique discussed in this section, we used the simple model
for the magnitude probability distribution pM (M ) [cf.
Eq. (34)]. However, the functional form of pM (M ) used
can be inaccurate in describing real data for several rea-
sons:
-- Different stellar populations can produce multiple
peaks in the color distribution of stars. For example,
giant and dwarf stars produce two distinct peaks in
the J − H histogram (the two peaks are also clearly
visible in a JHK color-color diagram).
-- Different stellar populations can also have different
slopes of the number counts, which could different sig-
nificantly from the "nominal" value α ≃ 0.34 (see, e.g.
Cambr´esy et al. 2002). If the various stellar popula-
tions also have different intrinsic mean colors, then the
simple model (34) could lead to inaccurate extinction
measurements.
-- The average NIR colors of stars in regions free of ex-
tinction are not completely independent of the star
luminosity. For example, a color-magnitude plot of
2MASS stars shows that the average J − K color
slightly increases as the K magnitude increases. Since
this effect is rather small, the associated bias is prob-
ably negligible in most cases; moreover, this effect can
be included in the expression (35) with a suitable
choice of the coefficients P , Q, and R.
All issues described above are strictly related to the sim-
plified description for the probability pM (M ) used here,
and not to the maximum-likelihood method itself. In other
words, it is possible (and relatively easy) to implement
a maximum-likelihood estimator based on a more realis-
tic probability distribution for the star magnitudes (e.g.,
synthetic stellar population models; see Robin et al. 2004;
Jarrett et al. 1994). For this purpose, we note that the
most computationally effective way to generalize pM (M )
is by writing it as a linear combination of functions of
the form (34) (with each function representing, de facto,
a different star population).
4.4.3. General remarks
On of most significant drawback of the maximum-
likelihood approach is its speed. The implementation used
in this paper is approximately one order of magnitude
slower than the Nicer method (at least on a typical work-
station), and this might prevent large applications of the
method proposed here. On the other hand, the fast tech-
nological progress in the computer speed justifies the work
presented in this paper, in the sense that soon it will be
possible to use the maximum-likelihood method on large
datasets composed of millions of stars.
Another possible issue related to the technique pre-
sented here is the need for a more detailed knowledge of
the properties of the data used. As a comparison, we note
that the original Nice technique makes use only of the
H and K magnitudes of stars and of the average color
hH − Ki of stars in the control field. In addition to these
data, Nicer also requires the estimated errors of star mag-
nitudes in the NIR bands used. Finally, the maximum-
likelihood method requires a detailed knowledge of the
probability distribution pM (M ), of the measurement er-
rors pǫλ( mλmλ) of each star, and of the completeness
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
17
functions cλ( mλ). In reality, in the Bayesian approach im-
plicitly adopted in this paper, the maximum-likelihood
method can also be used if an approximate knowledge of
these parameters is available. Suppose, for example, that
the parameters P , Q, and R that characterize pM (M ) [see
Eq. (35)] are only approximately known from the mea-
surements in the control field. In this case we can take
these parameters as unknowns in the expression for the
likelihood, and we can thus minimize L with respect to
them as well as with respect to the parameters that char-
acterize pAV (AV ) (AV and f in the case considered here).
As customary in standard maximum-likelihood problems,
we include the knowledge on P , Q, and R as a prior in
the function L. Note that this way the dimension of the
space over which we need to minimize L greatly increases
(and this can pose severe computational problems), but in
principle the schema proposed is applicable to real cases.
Interestingly, the Nice and Nicer technique can be
seen as special cases of the maximum-likelihood method
when no prior knowledge on R is available: the complete
lack information on the normalization of pM (M ) forces
our method to use the only color information of the stars.
Similarly, the number counts method can be recovered as
special case when the knowledge on the average colors χa
is absent [or, equivalently, when the scatter in the colors
Cab is very large; cf. Eq. (34)]. This suggests that the
Nice(r)) techniques are not more robust of the maximum-
likelihood one, but just simpler.
Finally, we mention that the presence of young stellar
objects (that could be for example embedded in the dense
cores) can in principle affect the extinction measurements.
Hence, these objects should be "manually" removed before
using any extinction measurement method, including the
maximum-likelihood described in this paper.
5. Discussion and conclusions
In this paper we have considered the problem of an accu-
rate determination of the extinction toward a dark molec-
ular cloud. The results obtained here can be summarized
in the following items:
-- The extinction and the reddening of background stars
have been considered from a general statistical point
of view.
-- The bias and uncertainties of the two main NIR tech-
niques used to map the extinction, the star count and
the color excess methods, have been discussed in de-
tail. We have shown that, although the color excess
method has generally a smaller error, it is affected by
a large bias in presence of contamination by foreground
stars. We have also shown that both Nice and Nicer
are affected by a relatively small bias (approximately
0.2mag in AV) as a result of selection effects.
-- A new optimal maximum-likelihood method has been
presented and tested with extensive simulations.
The simulations described in Sect. 4.3 have shown that
in the simple case of a thin cloud the maximum-likelihood
estimator performs significantly better than the Nice and
Nicer estimators (since the number count method is
known to have a larger noise, we did not report a de-
tailed comparison with this method here). However, the
maximum-likelihood approach also allows us to consider
more general cloud configurations, which cannot be easily
dealt with using standard techniques. For example, the
maximum-likelihood techniques could be used to measure
directly on the same patch of the sky both the column
density AV and the fraction of foreground stars f ; alter-
natively, it would be also possible to determine f globally
on the cloud and take it as a constant on the whole field
(a good approximation for small clouds).
In Sect. 4.4.1 we discussed one of the most serious lim-
itations of NIR color excess studies in molecular clouds,
namely the bias introduced by substructures. In our orig-
inal formulation, the maximum-likelihood method can
be used not only in the thin cloud approximation, but
in general for any functional form of pAV (AV ). Hence,
as mentioned above, we could implement the maximum-
likelihood method using a more realistic probability dis-
tribution for AV , such as the one of Eq. (65).
Another possibility offered by the maximum-likelihood
approach is the generalization of the thin-cloud approxi-
mation to a multi-layer case, where two or more (thin)
clouds located at different distances are observed on over-
lapping areas of the sky. For example, in case of a double
cloud we could write [cf. Eq. (17)]
pAV (AV ) = f (1)δ(AV ) + f (2)δ(AV − A(1)
V )
+(cid:0)1 − f (1) − f (2)(cid:1)δ(cid:0)AV − A(1)
V − A(2)
V (cid:1) .
(66)
This configuration, for example, might be appropriate to
study clouds close to the galactic center, where the su-
perposition of different complexes is very likely. Such a
method could effectively disentangle the effects of the two
clouds provided the values of f (1) and f (2) are sufficiently
different. Alternatively, one could use a double cloud as a
null test, i.e. to check that indeed the thin-cloud approx-
imation is appropriate (in this case one expects to find
A(1)
V ≃ 0, A(2)
V ≃ 0, or f (1) ≃ 0). Some of these possibili-
ties will be investigated in detail in a follow-up paper by
using 2MASS data.
Acknowledgements. We are very grateful to G. Bertin and
J. Alves for useful and stimulating discussions, and to the ref-
eree, L. Cambresy for providing stimulating comments and for
helping us improve the paper significantly.
Appendix A: The median and related estimators
The goal of this appendix is to derive the probability distribu-
tion of the median of n independent identical random variables.
This analysis is useful to address some of the issues discussed
in Sect. 3.2.
18
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
A.1. Notation
In the following we will often deal with ordered and unordered
n-tuples. The latter can be taken to be a set of n elements,
where typically each element is a random variable or an element
of another tuple; the former can be taken to be a list of n
elements, where thus each element is associated to a position
in the list. Hence, given a positive integer k ≤ n, it makes sense
to talk about the k-th element of an ordered n-tuple, while this
is meaningless for an unordered tuple.
We will denote an ordered n-tuple using brackets, as in
[x1, . . . , xn]; instead, we will reserve braces for the unordered
tuples, as in {x1, . . . , xn}. Note that, for consistency with the
definition, we identify unordered tuples if these differ just by a
permutation of the elements: thus {x1, x2, x3} is, for instance,
identical to {x2, x1, x3}.
A.2. P n
k and the median estimator
Let us call p(x) the probability distribution for the real ran-
dom variable x, and let us consider the generation of unordered
n-tuples {x1, x2, . . . , xn} of such variables. Suppose that, af-
ter the random generation of a tuple, we order the tuple such
that x1 ≤ x2 ≤ · · · ≤ xn. We consider now the probability
distribution pn
k (xk) for the k-th element of the ordered tuple
[x1, x2, . . . , xn], which can be written as
pn
k (xk) = n n − 1
k − 1!p(xk)(cid:2)P (xk)(cid:3)k−1(cid:2)1 − P (xk)(cid:3)n−k , (A.1)
where P (x) is the cumulative probability distribution of x:
P (x) =Z x
−∞
p(x′) dx′ .
(A.2)
The expression appearing in Eq. (A.1) can be explained as
follows. Let us consider an element (for example the first) of
the unordered n-tuple. The probability that this element is the
k-th in the ordered tuple and that it has a value in the range
[x, x + dx] is the product of three terms:
-- p(x) dx to takes into account the intrinsic probability dis-
tribution of x;
-- (cid:0)n−1
k−1(cid:1)(cid:2)P (xk)(cid:3)k−1(cid:2)1−P (xk)(cid:3)n−k, which is a simple binomial
distribution giving the probability that the value chosen
has k − 1 elements of the order n-tuple at its left, and n − k
elements at its right.
-- Finally, we have to multiply the whole result by n in order
to take into account the arbitrary choice of the k-th element
in the n-tuple.
The cumulative distribution P n
k (xk) is given by
1
0.8
0.6
0.4
0.2
0
)
k
x
(
nk
P
1
2
3
4
5
6
7
0
0.2
0.4
0.6
0.8
1
P (xk)
Fig. A.1. The cumulative probability distribution P n
k (xk)
as a function of P (x) for n = 7 and various values of k
(marked close to the relative curves).
1
0.8
0.6
0.4
0.2
0
)
k
x
(
nk
P
0
0.2
0.4
0.6
0.8
1
P (xk)
Fig. A.2. The cumulative probability distribution P n
k (xk)
as a function of P (x) for k ∈ {1, 2, 3, 4, 5, 6, 7} and
n = 2k − 1. Note that all curves pass through the point
(0.5, 0.5).
In the last step we changed the index variable into m = k + ℓ.
The final result obtained in Eq. (A.3) has the advantage to be
a simple (polynomial) expression in P (x).
Figures A.1 and A.2 show the polynomials P n
k (xk) as a
function of P (xk) for various values of n and k. These figures
suggest a number of properties for P n
k (xk) that can be verified
analytically with the help of the equations written above:
-- P n
k (x) = 0 if P (x) = 0, and P n
k (x) = 1 if P (x) = 1;
k is a monotonic increasing function of P . This
moreover, P n
implies that pn
k (x) vanishes where p(x) vanishes.
P n
k (xk) = Z xk
−∞
pn
k (x′
k) dx′
k
-- Using Eq. (A.1), it is possible to show that [see below
Eq. (A.10)]
=
=
n!
(k − 1)!(n − k)!
ℓ !
(−1)ℓ n − k
n−k
Xℓ=0
k) dx′
k
×Z xk
Xm=k
n
−∞(cid:2)P (x′
(−1)m+k(cid:18) n
k)(cid:3)k−1+ℓp(x′
m(cid:19)(cid:18)m − 1
k − 1(cid:19)(cid:2)P (xk)(cid:3)m .
1
n
n
Xk=1
P n
k (x) = P (x) ,
1
n
n
Xk=1
pn
k (x) = p(x) .
(A.4)
-- As suggested by Fig. A.1 and by Eq. (A.1), we have
P n
k (x) = 1 − P n
n+1−k(x′) ,
(A.5)
(A.3)
provided that P (x) = 1−P (x′). This, in particular, implies
that P 2k−1
(x) = 1/2 if P (x) = 1/2 (see Fig. A.2).
k
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
19
-- Equation (A.3) specialized to the cases k = 1 and k = n is
distributions. Let us then evaluate the sum
P n
1 (x1) =(cid:2)P (x1)(cid:3)n , P n
n (xn) = 1 −(cid:2)1 − P (xn)(cid:3)n .
(A.6)
The two last properties have a close relationship with the me-
dian estimator. We note, in fact, that for odd n the median
estimator is in our notation x(n+1)/2; as a result, the probabil-
ity distribution p2k−1
(xk) is just the probability distribution of
the median for n = 2k − 1. Hence, the property (A.5) written
above can be rephrased as
k
P n
≥k(x) =
=
1
1
n
n
P n
k (x)
Xk′=k
m(cid:19)(cid:2)P (x)(cid:3)m
Xm=k(cid:18) n
k′ − 1(cid:19) .
(−1)k+m(cid:18)m − 1
n − k + 1
n − k + 1
×
m
Xk′=k
(A.9)
The median of the probability distribution of the me-
dian estimator of the random variable x is the median
of the probability distribution of x.
We now consider two cases. If k = 1, then the last sum can
be taken to be the binomial expansion of (1 − 1)m−1, which
vanishes for all integers m > 1. Hence we are left just with the
case m = 1, for which we obtain
Note that here we use the term "median" to denote both the
usual median of an n-tuple {x1, . . . , xn}, and the median of a
distribution, defined as the value x such that the value cumu-
lative distribution is 1/2.
A.3. P n
≤k and P n
≥k
For our purposes, it is also useful to define and study two prob-
ability distributions closely related to P n
k (xk). Let us consider
again the ordered n-tuple [x1, . . . , xn], where each element is
drawn from a probability distribution p(x). Now let us retain
only the first k elements of the ordered n-tuple, and let us call
pn
≤k(x≤k) the probability distribution for each element of the
unordered k-tuple {x1, . . . , xk}; similarly, we call pn
≥k(x≥k) the
probability distribution for the tuple {xk, . . . , xn} where only
the last (n − k + 1) elements of the original ordered n-tuple are
retained.
The evaluation of the probability distribution pn
≥k(x) can
be broken into two parts. Consider the (k − 1)-th element xk−1
of the tuple [x1, . . . , xn] (i.e., the last element discarded); then,
clearly, each element of the (n − k + 1)-tuple {xk, . . . , xn} can-
not be smaller than xk−1. Moreover, each element of this un-
ordered tuple is distributed between xk−1 and ∞ according to
the (truncated) original probability distribution p(x). Finally,
by repeating this argument for each possible value of xk−1
(weighted by pn
k−1), we obtain the expression
pn
≥k(x) =Z x
−∞
pn
k−1(xk−1)
p(x)
1 − P (xk−1)
dxk−1 .
(A.7)
Note that the term 1 − P (xk−1) is introduced here to correctly
normalize the truncated probability distribution p(x).
As usual in this appendix, it is more convenient to consider
the cumulative probability distributions. We can thus integrate
pn
≥k(x) over x and obtain, after a few manipulations, a closed
expression for P n
≥k(x). Here, however, we prefer to follow a
different and simpler path.
Let us consider again the tuple [xk, . . . , xn]. Each element
of this ordered tuple follows the probability distribution pn
k (xk)
discussed in the previous section. Hence, if we consider the un-
ordered tuple, the elements {xk, . . . , xn} will follow the average
distribution
pn
≥k(x) =
1
n − k + 1
pn
k (x) .
n
Xk′=k
(A.8)
By integrating both sides of this equation on dx, we can verify
that the same relation holds for the cumulative probability
P n
≥1 =
1
n
n
Xk′=k
P n
k (x) = P (x) .
(A.10)
This shows that each element of the unordered n-tuple
{x1, . . . , xn} follows the original distribution p(x), a very nat-
ural result indeed.
If, instead, k > 1, then using the properties of the binomial
coefficient we find
P n
≥k(x) =
1
n − k + 1
n
Xm=k
(−1)m+k(cid:18) n
m(cid:19)(cid:18)m − 2
k − 2(cid:19)(cid:2)P (x)(cid:3)m .
(A.11)
The similarity of this result with the last line of Eq. (A.3) is
rather surprising.
Similarly, we wish to investigate the cumulative probability
≤k(x) for the k-tuple {x1, . . . , xk}. This quantity
distribution P n
is better evaluated using P n
≥k(x):
P n
≤k(x) =
=
=
=
k
1
k
n
1
P n
P n
k′ (x)
k′ (x) −
Xk′=1
k(cid:20) n
Xk′=k+1
Xk′=1
k(cid:20)nP (x) − (n − k)P n
Xm=k+1(cid:18) n
P (x) +
n
k
1
k
1
n
P n
k′ (x)(cid:21)
≥k+1(x)(cid:21)
m(cid:19)(cid:18)m − 2
×(cid:2)P (x)(cid:3)m .
k − 1(cid:19) (−1)m+k
(A.12)
As an application of the results obtained in this section,
we evaluate the bias introduced in the estimate of the column
density when using the technique described in Sect. 3.2 to re-
move foreground stars [see Eq. (54)]. Suppose that we observe
stars in a region with no significant extinction, so that both
foreground and background stars have the same distribution
in colors. For simplicity, we assume that the reddening esti-
mates AV for each individual star is a Gaussian distribution
with vanishing average and variance σ AV
. In this situation, if
we exclude the k bluer stars (because they are taken to be
foreground), we will bias the estimate of AV toward large col-
umn densities. In particular, the column densities for the stars
left will be distributed according to pn
≥k+1. The median of this
20
M. Lombardi: Optimal column density measurements from multiband near-infrared observations
distribution can be evaluated as follows. At AV = 0 we have
P (AV ) = 1/2, while
Kleinmann, S. G., Lysaght, M. G., Pughe, W. L., et al. 1994,
Experimental Astronomy, 3, 65
P n
≥k+1(0) =
1
n − k(cid:20) n
2
−
k
Xk′=1
P n
k′ (1/2)(cid:21) .
(A.13)
Assuming n ≫ k, then each term in the sum above is approx-
imately unity (cf. above Fig. A.1). Hence we obtain
P n
≥k+1(0) ≃
n − 2k
2n − 2k
.
(A.14)
The fact that this quantity is not exactly 1/2 is telling us that
there is a bias. The exact amount of this bias can be evalu-
ated by solving the equation P n
≥k+1(x) = 1/2. Using Newton's
method we obtain then the approximate solution
Lada, C. J., Alves, J., & Lada, E. A. 1999, ApJ, 512, 250
Lada, C. J., Lada, E. A., Clemens, D. P., & Bally, J. 1994,
ApJ, 429, 694
Larson, R. B. 1981, MNRAS, 194, 809
Li, A. & Greenberg, J. M. 2003, in Solid State Astrochemistry,
37, astro-ph/0204392
Lilley, A. E. 1955, ApJ, 121, 559
Lombardi, M. 2005, submitted to A&A
Lombardi, M. & Alves, J. 2001, A&A, 377, 1023
Mathis, J. S. 1990, ARA&A, 28, 37
Mattila, K. 1986, A&A, 160, 157
Miesch, M. S. & Bally, J. 1994, ApJ, 429, 645
Padoan, P., Jones, B. J. T., & Nordlund, A. P. 1997, ApJ, 474,
730
416, 157
Schultheis, M., Ganesh, S., Simon, G., et al. 1999, A&A, 349,
L69
Trumpler, R. J. 1930, Lick Observatory Bulletin, 14, 154
van de Hulst, H. C. 1946, Recherches Astronomiques de
l'Observatoire d'Utrecht, 11, 2
Wolf, M. 1923, Astronomische Nachrichten, 219, 109
x ≃(cid:20) dP n
≥k+1
dx
(cid:21)−1
x=0(cid:20) 1
2
− P n
≥k+1(0)(cid:21) ≃
k
2np(0)
.
(A.15)
Rieke, G. H. & Lebofsky, M. J. 1985, ApJ, 288, 618
Robin, A. C., Reyl´e, C., Derri`ere, S., & Picaud, S. 2004, A&A,
Note that in this equation we used the relationship between
cumulative and differential probability distributions; moreover,
we simplified pn
k (0) ≃ 0 at low k, and retained only terms linear
in k/n. The final result obtained is thus given by Eq. (54).
References
Alves, J., Lada, C. J., & Lada, E. A. 1999, ApJ, 515, 265
Alves, J., Lada, C. J., & Lada, E. A. 2001, Nature, 409, 159
Andreazza, C. M. & Vilas-Boas, J. W. S. 1996, A&AS, 116, 21
Barnard, E. E. 1919, ApJ, 49, 1
Bohlin, R. C., Savage, B. D., & Drake, J. F. 1978, ApJ, 224,
132
Bok, B. J. 1937, The distribution of the stars in space (Chicago:
University of Chicago Press, 1937)
Bok, B. J. 1956, AJ, 61, 309
Cambr´esy, L., Beichman, C. A., Jarrett, T. H., & Cutri, R. M.
2002, AJ, 123, 2559
Cambr´esy, L. 1999, A&A, 345, 965
Cambresy, L., Epchtein, N., Copet, E., et al. 1997, A&A, 324,
L5
Casali, M. M. 1986, MNRAS, 223, 341
Cressie, N. A. C. 1993, Statistics for Spatial Data (Wiley &
sons, New York)
Dickman, R. L. 1978, AJ, 83, 363
Eadie, W., Drijard, D., James, F., Roos, M., & Sadoulet,
B. 1971, Statistical Methods in Experimental Physics
(Amsterdam New-York Oxford: North-Holland Publishing
Company)
Epchtein, N., de Batz, B., Capoani, L., et al. 1997, The
Messenger, 87, 27
Frerking, M. A., Langer, W. D., & Wilson, R. W. 1982, ApJ,
262, 590
Harjunpaa, P., Lehtinen, K., & Haikala, L. K. 2004, A&A, 421,
1087
Heyer, M. H. & Brunt, C. M. 2004, ApJ, 615, L45
Horner, D. J., Lada, E. A., & Lada, C. J. 1997, AJ, 113, 1788
Indebetouw, R., Mathis, J. S., Babler, B. L., et al. 2004, ArXiv
Astrophysics e-prints
Jarrett, T. H., Dickman, R. L., & Herbst, W. 1994, ApJ, 424,
852
Jones, T. J., Hyland, A. R., & Bailey, J. 1984, ApJ, 282, 675
Jones, T. J., Hyland, A. R., Robinson, G., Smith, R., &
Thomas, J. 1980, ApJ, 242, 132
|
0810.4793 | 1 | 0810 | 2008-10-27T11:48:54 | On the Low Surface Magnetic Field Structure of Quark Stars | [
"astro-ph",
"cond-mat.supr-con"
] | Following some of the recent articles on hole super-conductivity and related phenomena by Hirsch \cite{H1,H2,H3}, a simple model is proposed to explain the observed low surface magnetic field of the expected quark stars. It is argued that the diamagnetic moments of the electrons circulating in the electro-sphere induce a magnetic field, which forces the existing quark star magnetic flux density to become dilute. We have also analysed the instability of normal-superconducting interface due to excess accumulation of magnetic flux lines, assuming an extremely slow growth of superconducting phase through a first order bubble nucleation type transition. | astro-ph | astro-ph | On the Low Surface Magnetic Field Structure of Quark Stars
Nandini Nag, Sutapa Ghosh, Roni Saha and Somenath Chakrabarty†
Department of Physics, Visva-Bharati, Santiniketan 731 235
West Bengal, India, ‡E-mail:[email protected]
Abstract
Following some of the recent articles on hole super-conductivity and related phenomena by
Hirsch [1, 2, 3], a simple model is proposed to explain the observed low surface magnetic field of
the expected quark stars. It is argued that the diamagnetic moments of the electrons circulating in
the electro-sphere induce a magnetic field, which forces the existing quark star magnetic flux density
to become dilute. We have also analysed the instability of normal-superconducting interface due to
excess accumulation of magnetic flux lines, assuming an extremely slow growth of superconducting
phase through a first order bubble nucleation type transition.
PACS numbers: 97.60.Jd, 97.60.-s, 75.25.+z
8
0
0
2
t
c
O
7
2
]
h
p
-
o
r
t
s
a
[
1
v
3
9
7
4
.
0
1
8
0
:
v
i
X
r
a
1
1.
INTRODUCTION
More than two decades ago, Witten had conjectured in an outstanding article [4] (see also
[5]) that a flavor symmetric quark matter at zero temperature and at zero pressure would be
energetically the most stable configuration. This exotic phase is known as the strange quark
matter (SQM). Since then a large number of articles have been published on the theoretical
studies of the properties of bulk SQM as well as droplets (known as strange-lets) of SQM
[6, 7, 8].
In those investigations various kinds of confinement models were used.
It has
further been shown that SQM phase can exist in compact stellar objects, known as strange
quark stars or strange stars [9, 10]. The finite size strange-lets are expected to be formed
both in ultra-relativistic heavy ion collisions and also during QCD phase transition in the
early universe [11, 12]. The strange-lets expected to have formed in the early universe, are
also called the strange nuggets, believed to be the relics from primordial quark-hadron phase
transition. These relics are also one of the viable candidates for baryonic dark matter. Inside
the strange stars, three light flavors: up (u), down (d), strange (s), along with small quantity
of electrons are in β-equilibrium. The presence of electrons make the system electrically
charge neutral. So far the gross properties of such compact objects are concerned, it is
almost impossible to distinguish strange stars from the conventional neutron stars. Except,
it is believed that these compact objects, which are expected to be strange stars, are fast
rotating (milli second or sub-milli second pulsars) and the surface magnetic field is quite
low ( ≤ 108G) [13]. It is further assumed that strange stars are very old and also extremely
cold objects, formed by the accretion of matter from the companion binary counter part.
However, for a strange star, the compositional structure in the microscopic scale is extremely
complex in nature. Since the density at the core region is very high (3 − 4 times normal
nuclear density), the corresponding quark matter is almost flavour symmetric. As one goes
radially outward, the density of matter decreases and since s-quarks are much more heavier
than u and d-quarks (current masses for u and d-quarks are 5−10MeV, whereas, for s-quark,
it is ≈ 150MeV), in this region, it is therefore energetically not favorable to produce enough
s-quarks through weak processes. The abundance of s-quark is therefore steadily decreases
as one moves towards the surface region. The charge neutrality of quark matter near the
outer core or the crustal region is therefore mainly maintained by electrons instead of strange
quarks. The density of electron is negligibly small at the core and is maximum at the crustal
2
region. Of course it is not very high (≤ 10−4 times normal nuclear density). These electrons
are bound to the positively charged quark matter by electromagnetic force (of which the
Coulomb force is the non-relativistic scenario) and are allowed to move freely across the
strong Coulomb field. Because of this strong electromagnetic force, they can not move far
away from the quark matter surface. The electron gas at the surface of strange stars, form a
very thin layer of width a few thousand Fermi, known as electro-sphere. Therefore, the gross
compositional structure of a strange star is very simple: a positively charged quark matter
region covered by a thin electro-sphere. Since the strange quark matter is energetically
more favorable, in the case of strange stars, it is unlikely to have a crust of dense iron like
matter, which is expected to be present in neutron stars. Further, depending on the internal
temperature and density of strange star, the quark matter may be in normal phase or in
super-conducting state or in color flavor locked phase (CFL) [14]. The CFL phase at the
core region is expected if the density is extremely high. Since the kinetic energies of the
electrons are much larger than the corresponding super-conducting gap energy, they never
show super-conductivity, either inside the star or at the electro-sphere. Which actually
means that the critical temperature for superconducting transition of electron is much less
than the corresponding critical temperature for quark matter.
In this article we shall propose a mechanism by which very low surface magnetic field of
strange stars can be obtained. We have assumed a type-I superconducting transition and
consider two possible flux expulsion processes. We shall show that extremely slow expulsion
mechanism leads to an instability at the normal-superconductor interface of quark matter.
In this article we have also analysed the interface instability. It is further noticed that by a
slow expulsion process, at several points on the interface, since the strength of magnetic field
becomes > Bc, the critical strength to destroy type-I superconductivity of quark matter,
the growth of superconducting quark matter phase will stop abruptly. On the other hand,
if it is an extremely fast process, we ultimately get a stable quark star of very low surface
magnetic field.
The paper is organized in the following manner: In the next section we shall compare a
strange star with a giant atom. In this section we shall discuss qualitatively the physical
processes which may take place at the surface of strange star and compare them with a giant
ultra-heavy atom. In section 3. we shall make a comparative study of superconducting quark
matter in strange star and the theory of hole superconductivity. The comparison is again
3
purely qualitative. The mathematical formalism will be given in section 6. We shall discuss
the mechanism of slow flux expulsion and give a mathematical formalism in sections 5 and
6 respectively. In the last section we have given the conclusion of the present work.
2. A GIANT ATOM MODEL
In this section we shall try to explain the possible low magnetic surface structure of quark
stars. In this model calculation, we assume a prompt phase transition to superconducting
phase and the strange star is treated as an equivalent giant atom with positively charged
quark matter (along with some admixture of electrons) as the nucleus and the electrons
in the electro-sphere are treated as orbital electrons. Unlike normal atoms, here size of
the nucleus is ≫ than the volume occupied by the electrons in the electro-sphere. Now
it is very easy to show from the β-equilibrium and charge neutrality conditions that the
net positive charge content of the quark matter nucleus is ≫ 172, the critical value of Z,
the atomic number of a typical super-heavy Dirac-atom. At the quark matter surface, the
vacuum will therefore become unstable, and spontaneous e+e−-pairs will be created if there
are unoccupied energy shell [15] (see also [16]).
Now for a many body fermion system, the microscopic theory of superconductivity sug-
gests that if the interaction favors the formation of pairs at low temperature, the system
may undergo a phase transition to a super-fluid state. In astrophysics, this is expected to
occur in the dense neutron matter of cold neutron stars. Whereas the small percentage of
protons (∼ 4%) inside neutron stars undergo a transition to type-II superconducting phase
[17, 18]. In the case of a many body system of fermions the well known BCS theory is gen-
erally used to study the super-conducting properties due to fermion pairing. One fermion
of momentum ~p and spin ~s combines with another one of momentum −~p and spin −~s and
form a Cooper pair. In the case of type-I electronic super-conductors the coupling is me-
diated by the electron-phonon interaction. In the case of quark matter, however, the basic
quark-quark interaction at large distance favors the formation of Cooper pairs. In the case
of quark matter, since the force is mediated via gluons, it gives rise to what is known as color
superconductivity. For a highly degenerate fermion system, which is true for strange star
matter, the pairing takes place near the Fermi surface. The other important condition that
must be satisfied for the formation of Cooper pairs is that the temperature (T ) of the system
4
should be much less than the super-conducting energy gap (∆). In the case of strange stars,
since the Fermi levels are not identical for u, d and s-quarks, only same type of quarks can
form Cooper pairs at the Fermi surface and give rise to color super-conductivity [19]. It is
interesting to note that the electrons, whose density is extremely low compared to quark
matter, may also be treated as highly degenerate relativistic plasma, but are unlikely to
form Cooper pairs.
In the next section we have assumed a type-I super-conducting phase of quark matter
within the strange stars and develop a mechanism by which the expelled magnetic field
is reduced at the electro-sphere. Since we are not investigating any of the properties of
super-conducting quark matter, rather, we are interested to study some of the important
features of normal electron gas layer at the surface region / inside the electro-sphere, which
are essential to have low surface magnetic field, then instead of standard relativistic version
of BCS theory [19], here, we have followed the interesting idea of "theory of hole super-
conductivity", proposed by Hirsch in a series of articles [1, 2, 3, 20, 21]. In the case of strange
stars the charge separation or the charge asymmetry takes place by the combined effect of
charge neutrality and β-equilibrium conditions; completely different physical mechanism.
Further, the properties of both positively charged quark matter inside the star and the
electrons in the electro-sphere are just opposite to the predicted nature of hole states and
the electrons for the hole super-conductor. In fact, we have noticed that the theory of hole
super-conductivity is not applicable for super-conducting quark matter of strange stars. The
reason is the strong interaction which bind the quarks within the strange stars.
3. THE HOLE SUPERCONDUCTIVITY AND COMPACT STRANGE STARS
To develop a mechanism of strange star magnetic field suppression at the electro-sphere,
we consider the charge asymmetric strange star as an equivalent hole super-conductor. We
have noticed, that there are a lot of similarities and also dissimilarities between the labo-
ratory samples and the strange stars within the theory of hole super-conductivity. In the
case of laboratory superconducting samples, according to the theory proposed by Hirsch, the
inner part is positively charged holes in normal phase. The electrons at the surface region
are super-conducting. The surface layer is bound to the inner region by strong electromag-
netic force. In the case of strange stars, the inner region is positively charged quark matter.
5
But unlike the laboratory superconductors, it is in super-conducting phase. Whereas, the
outer layer, the electro-sphere, which is a degenerate electron gas, is in normal phase. The
force between these two regions is again electromagnetic in nature. Going a step further,
we assume following the model by Hirsch, that the electrons in the electro-sphere gyrate
about the magnetic lines of force, expelled from the super-conducting interior of the strange
star. The origin of such orbital motion of these normal electrons in electro-sphere about the
magnetic flux lines is the well known Lorentz force. This orbital motion of electrons will
generate a magnetic field at the surface of the strange star. According to Lenz's law, the
induced field must be in the direction opposite to that of existing one. At the surface or
in the electro-sphere, therefore, the magnetic field from the tiny magnets produced by the
gyration of electrons about the existing lines of forces reduce the strange star magnetic field
by diamagnetic effect. Therefore, unlike the hole super-conductor, in the case of strange
stars, the magnetic field expelled from the super-conducting quark matter are suppressed
by the electron gas, which is in normal phase. By the repulsive diamagnetic action, there
will be an effective dilution of magnetic flux lines inside the electro-sphere. The dilution of
magnetic flux will naturally increase the size (width) of the electro-sphere. Analogous to
the phenomenon of frozen-in magnetic field, here, the "frozen-in electron gas" will be pulled
away by the flux lines. The increase in size of the electro-sphere will depend both on the
repulsive magnetic force and the attractive coulomb force and in equilibrium configuration,
these two will balance each other. The width of the electro-sphere of a typical quark star of
surface magnetic field ≤ 108G can be a few tens of km, which is at least an order of mag-
nitude less than the radius of the light cylinder, ∼ 100km, for a milli second pulsar. Now
this increase in size of the electro-sphere will reduce the electron density within the system.
The relation between the induced field and the existing field is given by eqn.(39). Further,
because of the motion of electrons around the quark matter nucleus, the actual trajectory of
an electron in the electro-sphere is more or less like a closed helical spring produced by the
motion of circular orbit along the lines of force. Since the magnetic field strengths at the
poles are very high, the electrons in the helical trajectory will be reflected back from these
region. The two ends will therefore behave like magnetic mirrors. It is also obvious from
eqn.(39) that the radius of gyration is larger at the equatorial region compared to the polar
values. Since the diamagnetism of the electrons inside the electro-sphere is caused by Lenz's
law, the sign of ~L.~p/~L.~p, which may be called as the effective "orbital-helicity" will change
6
sign after each reflection by the magnetic mirrors at the poles. Here ~p and ~L are respectively
the linear momentum and the angular momentum of the electrons. Further, when a balance
between the existing field and the induced field will be established, the electrons within each
half of the electro-sphere will under go steady helical motion.
4. SLOW EXPULSION OF MAGNETIC FLUX LINES
In this section we shall give a qualitative picture of very slow magnetic flux expulsion from
the super conducting region. It has been shown in ref.[19] in a relativistic version of BCS
theory, that if a normal quark matter system undergoes a superconducting phase transition,
the newly produced quark matter phase will be a type-I superconductor. They have also
shown in that the critical magnetic field to destroy such pairing is ∼ 1016G for n ∼ 3 − 4n0,
with n0 = 0.17fm−3, the normal nuclear density. This magnetic field strength is indeed much
larger than the typical pulsar magnetic field. The corresponding critical temperature is ∼
109−1010K, which is again high enough for strange stars, which are expected to be extremely
cold objects.
In this section, instead of investigating the superconducting properties of
quark matter inside strange stars, we shall give a possible mechanism of flux expulsion
by Meissner effect, assuming that the growth of superconducting phase is extremely slow.
Further, we assume that the magnetic field strength at the core region of a strange star
are much less than the corresponding critical value for the destruction of superconducting
property and the temperature is also low enough. Then during such a type-I superconducting
phase transition, the magnetic flux lines from the superconducting quark matter of the
strange star will be pushed out towards the normal crustal region. Now for a small type-
I superconducting laboratory sample placed in an external magnetic field less than the
corresponding critical value, the expulsion of magnetic field takes place instantaneously.
Whereas in the bulk strange star scenario, the picture may be completely different.
It
may take several thousands of years for the magnetic flux lines to get expelled from the
superconducting core. Which further means, that the growth of superconducting phase in
strange stars may not be instantaneous. A simple estimate shows that the expulsion time due
to ohmic diffusion is ∼ 104yrs [22]. It was shown by Chau using Ginzberg-Landau formalism
that the time for expulsion of magnetic lines of force accompanied by the enhancement of
magnetic field non-uniformities at the crustal region gets prolonged to 107yrs [23]. Alford
7
et. al. investigated the expulsion of magnetic lines of force from the colour superconducting
region by considering the pairing of like and unlike quarks and obtained the expulsion time
much larger than the age of the Universe [24].
In our investigation of magnetic flux expulsion from growing superconducting core of a
strange star, the idea of impurity diffusion in molten alloys or the transport of baryon num-
bers from hot quark matter soup to hadronic matter during quark-hadron phase transition
in the early universe, expected to occur micro-second after big bang (the first mechanism is
used by the material scientists and metallurgists [25], whereas the later one is used by cosmol-
ogists working in the field of big bang nucleo-synthesis [26, 27]) are assumed. In the present
section, we shall further show the possibility of Mullins-Sekerka normal-superconducting
interface instability [28, 29] in quark matter. This is generally observed in the case of solidi-
fication of pure molten metals at the solid-liquid interface, if there is a temperature gradient.
The interface will always be stable if the temperature gradient is positive, otherwise it will
be unstable. In alloys, the criteria for stable / unstable behaviour is more complicated. It is
seen that, during solidification of an alloy, there is a substantial change in the concentration
ahead of the interface. Here solute diffusion as well as the heat flow effects must be consid-
ered simultaneously. The particular problem we are going to investigate here is analogous
to solute diffusion during solidification of an alloy.
5. A FORMALISM FOR SLOW NUCLEATION
It has been assumed that the growth of superconducting quark bubble started from the
centre of the star and the nomenclature controlled growth for such phenomenon has been
used.
If the magnetic field strength and the temperature of the star are a few orders of
magnitude less than their critical values, the normal quark matter phase is thermodynami-
cally unstable relative to the corresponding superconducting one. Then due to fluctuation,
a droplet of superconducting quark matter bubble may be nucleated in metastable nor-
mal quark matter medium. If the size of this superconducting bubble is greater than the
corresponding critical value, it will act as the nucleating centre for the growth of supercon-
ducting quark core. The critical radius can be obtained by minimising the free energy. Then
8
following the work of Mullins and Sekerka, we have [28, 29]
,
(1)
rc =
16πα
B(c)2(cid:20)1 −(cid:16) B
B(c)(cid:17)2(cid:21)
where α is the surface tension or the surface energy per unit area of the critical supercon-
ducting bubbles, (from this expression it is possible to obtain the critical size of the quark
matter bubble by considering 10−3 ≤ α ≤ 1 (in MeV/fm−3) as the range of surface tension)
which is greater than zero for a type-I superconductor-normal interface, B(c) is the critical
magnetic field. In presence of a magnetic field B < B(c), the normal to superconducting
transition is first order in nature. As the superconducting phase grows continuously, the
magnetic field lines will be pushed out into the normal quark matter crust. This is the usual
Meissner effect observed in type-I superconductor. We compare this phenomenon of mag-
netic flux expulsion from a growing superconducting quark matter core with the diffusion of
impurities from the frozen phase of molten metal or the transport of baryon numbers from
hot quark matter soup during quark-hadron phase transition in the early universe. The
formation of superconducting zone is compared with the solidification of molten metal or
with the transition to hadronic phase with almost zero baryon number. It is known from
the simple thermodynamic calculations that if the free energy of molten phase decreases in
presence of impurity atoms, then during solidification they prefer to reside in the molten
phase otherwise they go to the solid phase. In this particular case the magnetic field lines
play the role of impurity atoms and because of less free energy, they prefer to remain in
normal quark matter phase. The normal quark matter phase plays the role of molten metal
or the hot quark soup. Whereas the superconducting phase can be compared with the frozen
solid phase or the hadronic phase. This idea was applied to baryon number transport dur-
ing first order quark-hadron phase transition in the early Universe, where baryon number
replaces impurity, quark phase replaces molten metal and hadronic matter replaces that of
solid metal [26, 27]. Of course the baryon number prefers to stay in the quark phase because
of Boltzmann suppression factor in the hadronic phase. Since the magnetic flux lines prefer
to reside in the normal phase, the well known Meissner effect can therefore be restated as the
solubility of magnetic flux lines in the superconducting phase is zero with a finite penetration
depth.
The dynamical equation for the flux expulsion can be obtained from the simplified model
of sharp normal-superconducting interface. The expulsion equation is given by the well
9
known diffusion equation [30]
∂B
∂t
= D∇2B
where B is the magnetic field intensity and D is the diffusion coefficient, given by
D =
c2
4πσn
,
(2)
(3)
σn is the electrical conductivity of the normal quark matter and c is the velocity of light.
The electrical conductivity of quark matter for B = 0 is given by [31, 32]
σn ∼ α−3/2
s
T −2
10
or [33]
σn ∼ (αsT10)−5/3
(4)
expressed in sec−1, where αs is the strong coupling constant and T10 = T /1010K, the nu-
merical value for this electrical conductivity in the case of quark matter relevant for quark
star density is ∼ 1026 sec−1. In the order of magnitude estimate, we have used this numer-
ical value for electrical conductivity. In this context we must mention that in presence of
strong quantizing magnetic field, σn may not be a scalar quantity, magnetic field destroys
the isotropy of the electromagnetic properties of the medium. In particular, for extremely
large B, the components of electric current vector orthogonal to B become extremely small.
Which indicates that the quarks can move only along the direction of magnetic field or in
other words, across the field the resistivity becomes extremely high. When the conditions
for charge neutrality and β-equilibrium are considered together, since the mass of s-quark
is assumed to be 150Mev, the electron density can not become exactly zero, but it is a few
orders of magnitude less than the s-quark density. Therefore, one can neglect the electron
contribution to electrical conductivity.
A solution of eqn.(2) with spherical symmetry can be obtained by Greens' function tech-
nique, and is given by (for a general topological structure, no analytical solution is possible)
B(r, t) =
B(0)(r′)
1
2r(πDt)1/2 Z ∞
0
[ exp(−u2
−) − exp(−u2
+)]r′dr′
(5)
where u± = (r ± r′)/2(Dt)1/2 and B(0)(r) is the magnetic field distribution within the star at
t = 0, which is of course an entirely unknown function of radial coordinate r. To obtain an
10
estimate of magnetic field diffusion time scale (τD), we assume B(0)(r) = B(0) = constant.
Then we have from eqn.(5)
B(r, t) = B(0)(cid:20)1 −
2
r
(πDt)1/2(cid:21)
(6)
Hence, if we put B(r, t) = 0 (field free condition), the estimated time scale for the expulsion
of magnetic flux lines is ∼ 105 − 106yrs. Which is of the same order of magnitude as
the Ohmic decay scale. From this simple estimate, this is the approximate time scale for
complete expulsion of magnetic field lines. Which is of course quite large. Unfortunately,
nothing else can be inferred about the growth of superconducting zone and the associated
expulsion of magnetic flux lines from this region. The reason behind such uncertainty is
our lack of knowledge or definite ideas on the numerical values of the parameters present in
eqn.(5).
To get some idea of the effect of magnetic field on the structure of growing superconduct-
ing zone, we shall now investigate the morphological instability of normal-superconducting
interface of quark matter. The motion of normal-superconducting interface is extremely im-
portant in this case and has to be taken into consideration. Then instead of eqn.(2) which is
valid in the rest frame, an equation expressed in a coordinate system which is moving with
an element of the boundary layer is the correct description of such superconducting growth,
known as Directional Growth. The equation is called Directional Growth Equation, and is
given by
∂B
∂t
− v
∂B
∂z
= D∇2B
(7)
where the motion of the plane interface is assumed to be along z-axis and v is the velocity of
the front. This diffusion equation must be supplemented by the boundary conditions at the
interface. The first boundary condition is obtained by combining Ampere's and Faraday's
laws at the interface, and is given by
Bv s= −D(∇B).n s
(8)
where n is the unit vector normal to the interface directed from the normal phase to the
superconducting phase. This is nothing but the continuity equation for magnetic flux diffu-
sion. For the normal growth of superconducting zone, the rate at which excess magnetic field
lines are rejected from the interior of the phase is balanced by the rate at which magnetic
flux lines diffuses ahead of the two-phase interface. Therefore the boundary layer between
11
superconducting-normal quark matter phases will become unstable if excess magnetic field
lines are present on the surface of the growing superconducting bubble, i.e., if the rate of
diffusion of magnetic flux lines is slow enough compared to the rate at which they are ex-
pelled from the superconducting zone. Local thermodynamic equilibrium at the interface
gives (Gibbs-Thompson criterion)
B s≈ B(c)(cid:18)1 −
4πα
RB(c)2(cid:19) = B(c) (1 − δC)
(9)
where δ is called capillary length with α the surface tension, C is the curvature = 1/R (for
a spherical surface), and B(c) is the thermodynamic critical field.
To investigate the stability of superconducting-normal interface, we shall follow the origi-
nal work by Mullins and Sekerka [28, 29], and consider a steady state growth of superconduct-
ing core. Then the time derivative in eqn.(7) will not appear. Introducing r⊥ = (x2 + y2)1/2
as the transverse coordinate at the interface, we have after rearranging eqn.(7)
" ∂2
∂r2
⊥
+
1
r⊥
∂
∂r⊥
+
∂2
∂z2 +
v
D
∂
∂z# B = 0
(10)
The approximation that the solidification is occurring under steady state condition used
in the freezing of molten material will be followed in the present case of normal to super-
conducting phase transition. Now if it is assumed that these two phenomena taking place
in two completely separate physical world are almost identical natural processes, then the
concentration of magnetic flux lines and normal-superconducting interface morphology will
be independent of time. The main disadvantage of this assumption is that there will be no
topological evolution of the interface shape. As a consequence of this constraint the solution
to the basic diffusion problem is indeterminate and a whole range of morphologies is per-
missible from the mathematical point of view. In order to distinguish the solution which is
the most likely to correspond to reality, it is necessary to find some additional criteria. The
examination of the stability of a slightly perturbed growth form is probably the most rea-
sonable manner in which this situation may be treated. In the following we shall investigate
the morphological instability of normal-superconducting interface from eqn.(10). Assuming
a solution of this equation expressed as the product of separate functions of r⊥ and z and
setting the separation constant equal to zero and using the boundary condition given by
eqn.(9), we have for an unperturbed boundary layer moving along z-axis
B = B(s) exp(−zv/D) = B(s) exp(−2z/l)
(11)
12
where l = 2D/v is the layer thickness, which is very small for small D. Mathematically, the
thickness of this layer is infinity. For practical purpose an effective value l can be taken.
The order of magnitude estimates or limiting values for the three quantities D, v and l can
be obtained from the stability condition of planer interface.
Due to excess magnetic flux lines at the interface, the form of the planer normal-
superconducting interface described by the equation z = 0 is assumed to be changed by
a small perturbation represented by the simple sine function
z = ǫ sin(~k.~r⊥)
(12)
where ǫ is very small amplitude and ~k is the wave vector of the perturbation. Then the
perturbed solution of the magnetic field distribution near the interface can be written as
B = B(s) exp(−vz/D) + Aǫ sin(~k.~r⊥) exp(−bz)
(13)
where A and b are two unknown constants. Since the solution should satisfy the diffusion
equation (eqn.(10)), we have
b =
v
2D
+"(cid:18) v
2D(cid:19)2
+ k2#1/2
(14)
To evaluate A, we utilise the assumption that both ǫ and ǫ sin(~k.~r⊥) are small enough so
that we can keep only the linear terms in the expansion of exponentials present in eqn.(13).
Then at the interface, we have after some straight forward algebraic manipulation
A =
v
D
B(s)
(15)
The expression describing the magnetic field distribution ahead of the slightly perturbed
interface then reduces to
B = B(s)(cid:20)exp(−vz/D) +
v
D
ǫ sin(~k.~r⊥) exp(−bz)(cid:21)
Now from the other boundary condition (eqn.(9)) we have
B(s) = B(c) −
4παB(c)
B(s)2 C
(16)
(17)
where C = z ′′/(1 + z ′2)3/2 is the curvature at z = ǫ sin(~k.~r⊥) and prime indicates derivative
with respect to r⊥.
13
Neglecting z ′2, which is small for small perturbation, we have
B(s) = B(c) + Γk2S
(18)
where Γ = 4παB(c)/B(s)2 and we have replaced ǫ sin(~k.~r⊥) by S. Since the amplitude of
perturbation ǫ is extremely small, the quantity S is also negligibly small.
Now eqn.(18) can also be written as
B(s) = B(c) + GS
where
G =
dB
dz
z=S= −
v
D (cid:18)1 −
vS
D (cid:19) B(s) − bAS(1 − bS)
Combining these two equations, we have
k2Γ +
v
D (cid:18)1 −
vS
D (cid:19) B(s) −
bv
D
B(s)S(1 − bS) = 0
(19)
(20)
(21)
This expression determines the form (values of k) which the perturbed interface must assume
in order to satisfy all of the conditions of the problem. To analyse the behaviour of the roots,
we replace right hand side of eqn.(21) by some parameter −P . (We have taken −P in order
to draw a close analogy with the method given in refs.[26, 27]). Then rearranging eqn.(21),
we have
− k2Γ −
v
D (cid:18)1 −
vS
D (cid:19) B(s) +
bvB(s)S
D
(1 − bS) = P
(22)
(in refs.[26, 27] the parameter P is related to the time derivative of ǫ, the amplitude of
small perturbation). If the parameter P is positive for any value of k, the distortion of the
interface will increase, whereas, if it is negative for all values of k, the perturbation will
disappear and the interface will be stable. In order to derive a stability criterion, it only
needs to know whether eqn.(20) has roots for positive values of k. If it has no roots, then
the interface is stable because the P − k curve never rises above the positive k-axis and P is
therefore negative for all wavelengths. We have used Decarte's theorem to check how many
positive roots are there. It is more convenient to express k in terms of b and then replacing
b by ω + v/D. Then we have from eqn.(22)
− ω2 Γ +
vB(s)S2
D ! − ω Γ +
2vB(s)S2
D
v
D
B(s)(cid:18)1 −
−
v
D
14
− B(s)S! v
S(cid:19)2
= P
D
(23)
This is a quadratic equation for ω. The first and the third terms are always negative. The
second term will also be negative if
Γ +
2vB(s)S2
D
− B(s)S > 0
(24)
Then it follows from Decart's rule that if the condition (24) is satisfied, there can not be
any positive root. Which implies that the small perturbation of the interface will disappear.
Since the amplitude of perturbation is assumed to be extremely small, the quantity S =
ǫ sin(~k.~r⊥) is also negligibly small. Under such circumstances the middle term of eqn.(23) is
much smaller than rest of the terms. The Decart's rule given by the condition (24) can be
re-written as
Γ > B(s)S
(25)
Which after some simplification gives the stability criterion for the plane unperturbed in-
terface, given by
α >
B(s)3S
4πB(c)
(26)
From the stability criterion, it follows that the normal-superconducting interface en-
ergy/area of quark matter has a lower bound, which depends on the interface magnetic field
strength, critical field strength and also on the perturbation term S. An order of magnitude
of this lower limit can be obtained by assuming B(s) = 10−3B(c). (Since the critical field
B(c) ∼ 1016G, and the neutron star magnetic field strength B ∼ 1013G, we may use this
equality). Then the lower limit is given by
αL ≈ 10−9 MeV/fm2 (cid:18) S
fm(cid:19)
On the other hand for B(s) = 0.1B(c), we have
αL ≈ 10−3 MeV/fm2 (cid:18) S
fm(cid:19)
The approximate general expression for the lower limit is given by
αL ≈ h3 MeV/fm2 (cid:18) S
fm(cid:19)
where h = B(s)/B(c). Therefore the maximum value of this lower limit is
αmax.
L ≈ 1 MeV/fm2 (cid:18) S
fm(cid:19)
15
(27)
(28)
(29)
(30)
when the two phase are in thermodynamic equilibrium. Of course for such a strong mag-
netic field, as we have seen [34, 35] that there can not be a first order quark-hadron phase
transition.
On the other hand if we do not have control on the interface energy, which can in principle
be obtained from Landau-Ginzberg model, we can re-write the stability criteria in terms of
interface concentration of magnetic field strength B(s), and is given by
1/3
B(s) <
4παB(c)
S(cid:16)1 − 2v
D S(cid:17)
(31)
This is more realistic than the condition imposed on the surface tension α. Now for a type-I
superconductor, the surface tension α > 0, which implies 1 − 2vS/D > 0. Therefore, we
have 2vS/D < 1. For the typical value of σn ∼ 1026 sec−1 for the electrical conductivity
of normal quark matter, the profile velocity v < D/2S ∼ 10−6/Scm/sec ∼ 1cm/sec for
S ∼ 10−6cm. Therefore the interface velocity < 1cm/sec for such typical values of σn and
S to make the planer interface stable under small perturbation. Now the thickness of the
layer at the interface is l = 2D/v > 10−6cm for such values of D (or σn) and v. Here S is
always greater than 0, otherwise, the magnetic field strength at the normal-superconductor
interface becomes unphysical. As before, if the second term of eqn.(23) is negligibly small
compared to other two terms, we have
B(s) < " 4παB(c)
S
#1/3
(32)
For the sake of illustration, we have shown in fig.(1), the distribution of magnetic field
in a very small portion of horizontal plane at the perturbed interface. We have solved
eqn.(17) numerically and use the typical parameter set as given above. Along z-axis, we
have plotted the ratio of the strength of surface magnetic field and the critical strength
for the destruction of type I quark matter superconductivity and x and y axes represent
the x-y coordinates of this tiny horizontal plane. The distribution is extremely chaotic in
nature and some of the magnetic field peaks are stronger that the critical strength to destroy
type-I superconductivity. Then analogous to the case of crystal growth in an impure molten
alloy, where the process is stopped because of high density of accumulated impurities at the
interface, here also, further growth of superconducting phase will be stopped. Therefore, it
is a kind of instability at the interface, which forces the growth process to stop abruptly.
16
Since the slow growth of superconducting phase does not give a stable quark star with very
low surface magnetic field, is therefore not a physically acceptable phenomenon.
6. THE PROMPT PROCESS OF FLUX EXPULSION
We have noticed that the slow expulsion process, in which diffusion mechanism of mag-
netic lines of force may be applicable, leads to a kind of instability at the interface.
It
ultimately stops abruptly the growth process of superconducting zone. As a consequence
we will not get a stable quark star with very low surface magnetic field.
In this section, we assume that the superconducting phase of quark matter grows by a
faster process; so that the diffusion model for the magnetic lines of force is not applicable.
Which actually means, that the expulsion also occurs with a faster rate. Now during this
process, since magnetic flux changes rapidly with time, an emf will be induced in the system.
Although, the superconducting quark Cooper pairs do not feel electromagnetic force, a
Lorentz force will be exerted on the electrons near the surface region, which are in normal
phase. These electrons will try to nullify the force following Lenz's law. As a consequence
they will start gyrating about the magnetic lines of force at the surface / in the electro-
sphere. We have already argued that these circulation of the electrons about the magnetic
field lines will produce a diamagnetic effect to oppose the Lorentz force by reducing the
strength of magnetic field at the surface.
To get an estimate for induced magnetic field strength at the strange star surface /
electro-sphere and the corresponding reduction in magnetic field strength, we follow the
recent article by Hirsch [20]. The magnetic dipole moment per electron in the electro-sphere,
due to the tiny orbital motion is given by
µ =
eu
2
a
(we have assumed that c = 1)
(33)
where u is the orbital velocity and a is the radius of the orbit. The induced emf in the
electronic orbit is then given by
E =
a
2
∂B
∂t
Then the corresponding change in orbital speed is given by
∆u =
ea
2me
B
17
(34)
(35)
Hence the change in magnetic moment per electron is
∆µ =
e2a2
4me
B,
(36)
and the associated average value of the induced magnetization per unit volume in the electro-
sphere is given by
M =
nee2a2
4me
B
(37)
where ne is the average electron density in the electro-sphere (in reality, ne must be maximum
near the quark matter surface and is minimum at the outer surface of the electro-sphere).
Hence one can obtain the strength of induced magnetic field at the surface region, which is
given by
Bind = 4πM
(38)
Therefore the ratio of induced magnetic field by the electrons in the electro-sphere to the
existing strange star magnetic field is given by
Bind
B
=
nee2
me
πa2
(39)
The strength of induced magnetic field is therefore a monotonically increasing function of
electron density. It is very easy to verify that for complete suppression of magnetic field of
a strange star, with a typical average electron density ne = 10−5nB, where nB = 0.17fm−3,
the normal nuclear density, the orbital radius a ∼ 7fm. Which is about three orders of
magnitude less than the width of the electro-sphere. Therefore it is quite obvious that the
electro-sphere of width ∼ 1000fm can accommodate a large number of tiny magnets.
It
may be argued that the actual form of trajectories for such tiny magnets in each half of the
electro-sphere are closed helical spring.
Now, if there is no reduction, the magnetic field strength at the surface could be as high
as 1030 times the magnetic field of a typical neutron star. One can obtain this number by
considering the ratio R2/d2, where R = 10Km, the radius of the star and d = 1000fm, the
width of the electro-sphere and assuming that the magnetic flux remains conserved during
Meissner expulsion from super-conducting quark matter. This value is far above the upper
limit predicted by Shabad and Usov [36].
Now it has been discussed in the literature that if the density of quark matter, particularly,
at the core region is sufficiently high, the color super-conducting quark matter undergoes a
18
phase transition to what we call the CFL phase [37, 38, 39]. It is therefore expected that
at the ultra-dense interior, there will be a color as well as charge neutral CFL phase. The
number of quarks in the CFL cluster is even and divisible by three. It is found that in this
new phase all the three flavors u, d and s can form pairs with the same flavor as well as
with other components, i.e., their Fermi levels are identical. In this case the excess u-quarks
will therefore go to the outer region. The gross structure as mentioned at the beginning of
this article may not therefore be correct for a strange star with extremely high core density;
at the interior, it could be the CFL phase, at the outer core or inner crust region, it is the
positively charged color neutral quark matter in non-CFL color super-conducting phase and
finally the thin outer surface, the electro-sphere, is the normal electron gas. Whether the
formation of CFL phase at the inner part of a super-conducting strange star is an equivalent
QCD picture of the so called Tau-effect [40, 41] needs further investigation.
Therefore we expect that depending on the density of quark matter, the inner part of a
strange star is either a type-I super-conductor or in CFL phase, the outer region is color
neutral type-I super-conducting non-CFL phase, mainly dominated by u-quarks and some
electrons, and finally the outer surface is a thin layer of electron gas in normal state. The
magnetic field lines are expelled from quark matter phase to the electron gas layer with a
small penetration at the outer surface of the quark matter. We have conjectured that the
classical diamagnetism of the normal electron gas suppresses the magnetic field of a strange
star by diluting the existing magnetic flux lines in the electro-sphere by the repulsive action
of the induced magnetic field.
If B ≈ Bind, we have almost total suppression.
In this
ideal situation, so far the emission due to magnetic activities are concerned, the object
becomes dark to the observers. However, because of vacuum instability, non-thermal surface
γ emission is possible. Since Z is very high, the Coulomb field at the quark matter surface
will be extremely strong, then e+e− pairs may be produced from vacuum by Schwinger
mechanism [42]. The created e+ gets annihilated with one of the electrons in the electro-
sphere and e− will occupy one of the empty energy levels. Therefore, γ emission will also
be forbidden if the star is extremely cold, when there will be no more vacant states for the
produced electrons [43].
An alternative explanation for low magnetic field of strange stars can also be obtained
from a very simple model of magnetic circuits. The combination of electro-sphere and the
magneto-sphere may be treated as an equivalent magnetic circuit with varying reluctance.
19
Of course, there are some open flux lines goes out after cutting the light cylinder. In that
case, we may assume that the whole universe is a complicated network of magnetic circuits
with varying reluctance and the strength of sources of magneto motive force. From the
definition, we have total flux
N =
MMF
R
(40)
where MMF is the effective magneto motive force and R is the reluctance, which is given by
R = I dl
αµ
(41)
where dl is the length of some flux carrying element of cross section α and magnetic perme-
ability µ. Let us consider the passage of expelled magnetic flux lines through surface region,
when there is no gyrating motion of electrons. Here we may take µ ≈ 1, the reluctance is
then given by.
+ Rms
(42)
where es indicates the electro-sphere and ms is the magneto-sphere. On the other hand,
with the gyrating electrons, since there is a diamagnetic effect at the surface / electro-sphere,
we must have the permeability µ < 1 in this region. In Gaussian unit, we have µ = 1 − 4πχ.
Where from eqn.(37) the susceptibility is given by
R1 = Z dl
α(cid:12)(cid:12)(cid:12)es
(43)
(44)
χ =
∂M
∂B
=
nee2a2
4me
The reluctance is then given by
R2 = Z dl
µα′(cid:12)(cid:12)(cid:12)es
+ Rms
where α′ is now the new cross sectional area of the flux carrying element. Since the MMF
source and the total flux does not change by the gyration of electrons at the surface, we
have effectively
which gives
R1 = R2
α′ = α
1
µ
(45)
(46)
Since µ ≪ 1, it is therefore quite obvious that the area of cross section of the electro-sphere
will increase by several orders of magnitude. As a result the magnetic flux density will
also be reduced by the same factor. Now it is well known that Larmor radius a ∼ B−1,
20
here B is the surface magnetic field. Therefore it is quite possible in the parameter space
(electron density and the corresponding Larmor radius) for the physically acceptable values
for electron density and Larmor radius, the surface magnetic field can be as low as 108G.
For the typical values of electron density ne ∼ 10−12n0 and Larmor radius a ∼ 10 A, the
magnetic permeability becomes ∼ 0.012. Which gives surface magnetic field a few times
107G. Since the Larmor radius increases with the decrease in magnetic field strength, it will
further increase the induced magnetic field (see eqn.(39)), if it dominates over the electron
density. The induced magnetic field, which is diamagnetic in nature will further reduce the
existing field. So the solution must be self-consistent in nature. We believe that ultimately
a steady state will be established in the electro-sphere. This is true for both the models.
7. CONCLUSION
From this investigation, we may conclude in a straight forward way that if strange stars
really exists with very low surface magnetic field, the type-I superconducting transition in
the quark matter phase must have occurred within the star. Further, the transition process
must be very prompt, so that diffusion model for the expulsion of magnetic flux lines is not
applicable, where the later gives rise to some kind of instability at the interface and finally
stops the growth of superconducting phase abruptly. Further, the slow transition model
does have any mechanism to reduce the surface field strength, which is ≤ 108G.
Our next conclusion is that the transition is quite fast. The classical diamagnetism of
gyrating electrons at the surface / electro sphere may be the possible source of opposing
field to reduce the expelled surface magnetic field by several orders of magnitude. It is quite
obvious that with this prompt transition model, using the idea of Hirsch, one can obtain a
surface magnetic field for the expected quark stars as low as 108G.
[1] J.E. Hirsch, Phys. Lett. A134, 451 (1989).
[2] J.E. Hirsch, Phys. Rev. B68, 184502 (2003).
[3] J.E. Hirsch, Phys. Rev. B71, 184521 (2005); Phys. Lett. A345, 453 (2005).
[4] E. Witten, Phys. Rev. D30, 272 (1984).
[5] A.R. Bodmer, Phys. Rev. D4, 1601 (1971).
21
[6] E. Farhi and R.L Jaffe, Phys. Rev. D30, 2379 (1984).
[7] S. Chakrabarty, S. Raha and B. Sinha, Phys. Lett. B229,112 (1989); S. Chakrabarty, Phys.
Rev. D43, 627 (1991).
[8] G. Baym, R.L. Jaffe, E.W. Kolb, L. Mclerran and T.P. Walker, Phys. Lett. B160, 181 (1985);
C. Alcock and E. Farhi, Phys. Rev. D32, 1273 (1985).
[9] P. Haensel, J.L. Zdunik and R. Schaeffer, Astr. and Astrophys. 160, 121 (1986).
[10] see for example the Proceeding of the Workshop on Strange Quark Matter in Physics and
Astrophysics, May 20-24, 1991, University of Aarhus, Denmark, Nucl. Phys. B24 (1991).
[11] J. Schaffner-Bielich, C. Greiner, A. Diener and H. Stoecker, Phys. Rev. C55, 3038 (1997).
[12] J. Madsen, Phys. Rev. Lett. 85, 4687 (2000).
[13] Observations released by the Chandra X-ray observatory predicted that compact objects
RXj1856.5-3754 and 3C58, where the later one is associated with the supernova SN1181 could
be the strange quark stars. Another such object XTEj1739-285, observed by a team led by
Philip Karret. It is a fast rotating compact object (1122 rotation per second). stars.
[14] M. Alford, K. Rajagopal, T. Schaefer and A. Schmitt, arXiv:0709.4632 (hep-ph) (to appear
in Rev. Mod. Phys.).
[15] B. Muller and J. Rafelski, Phys. Rev. Lett. 34, 349 (1975); B. Muller, Ann. Rev. Nucl. Sci.
26, 351 (1976).
[16] J. Madsen, Phys. Rev. Lett. 100, 151102 (2008).
[17] G. Baym, C. Pethick and D. Pines, Nature 224, 673 (1969); S.L. Shapiro and S.A. Teukolsky,
Black Holes, White Dwarfs and Neutron Stars, Wiley, New York (1983).
[18] M. Baldo, et. al., Nucl. Phys. A536, (1992) 349; S. Chakrabarty, Astrophysics & Space
Science, 314, (2008) 105.
[19] D. Bailin and A. Love, Phys. Rep. 107, 325 (1984).
[20] J. Hirsch, arXiv:cond-mat/0301611.
[21] J. Hirsch, arXiv:cond-mat/0508529; arXiv:0803.2054.
[22] Bailin B. and Love A., 1982, Nucl. Phys. B205, 119.
[23] Chau A.F., 1997, Ap. J. 479, 886.
[24] Alford M.G., Berges J. and Rajagopal K., 2000, Nucl. Phys. B571, 269.
[25] Losert W., Shi B.Q. and Cummins H.Z., 1998, Proc. Natl. Acad. Sci. USA (Applied Physical
Science) 95, 431.
22
[26] Adams F.C, Freese K. and Langer J.S., 1993, Phys. Rev. D47, 4303.
[27] Kamionkowski Marc and Freese K., 1992, Phys. Rev. Lett. 69, 2743.
[28] Mullins W.W. and Sekerka R.F., 1963, Jour. Appl. Phys. 34, 323.
[29] Mullins W.W. and Sekerka R.F., 1964, Jour. Appl. Phys. 1964; 35, 444.
[30] Langer J.S., 1980, Rev. Mod. Phys. 52, 1.
[31] Haensel P. and Jerzak A.J., 1989, Acta Phys. Pol. B20, 141.
[32] Haensel P., 1991, Nucl. Phys. B24 (proc. of the Int. Workshop on Strange quark Matter in
Physics and Astrophysics, Univ. of Aarhus, Denmark, May 20-24, 1991), 23.
[33] Heiselberg H and Pethick C.J, 1993, Phys. Rev. D48, 2916.
[34] Chakrabarty S., 1995, Phys. Rev. D51, 4591.
[35] Ghosh T. and Chakrabarty S., 2001, Phys. Rev. D63 043006.
[36] A.E. Shabad and V.V. Usov, Phys. Rev. Lett. 96, 180401-1 (2006); astro-ph/0601542.
[37] M. Alford, K. Rajagopal and F. Wilczek, Phys. Lett. B422, 247 (1998).
[38] M. Alford, K. Rajagopal, T. Schafer and A. Schmitt, arXiv:0709.4635.
[39] R. Rapp, T. Schafer, E. Shuryak and M. Velkovsky, Phys. Rev. Lett. 81, 53 (1998).
[40] J.E. Hirsch, Phys. Rev. Lett. 94, 187001 (2005).
[41] R. Tau, X. Zhang, X. Tang and P.W. Anderson, Phys. Rev. Lett. 83, 5575 (1999).
[42] J. Schwinger, Phys. Rev. 82, 664 (1951).
[43] V.V. Usov, Astrophys. Jour 550, L179 (2001).
23
4
3.5
3
2.5
2
1.5
1
0.5
0
0
10
20
30
40
50 0
20
10
50
40
30
FIG. 1: The variation of magnetic field along z-axis (plotted as a ratio of B/Bcr, where
Bcr ∼ 1016G, the critical strength to destroy the type-I quark matter superconductivity) with
the orthogonal coordinates x and y.
24
|
astro-ph/9710129 | 1 | 9710 | 1997-10-13T18:01:39 | Breaking the degeneracy between anisotropy and mass: The dark halo of the E0 galaxy NGC 6703 | [
"astro-ph"
] | (abridged) We have measured line-of-sight velocity profiles (VPs) in the E0 galaxy NGC 6703 out to 2.6 R_e. From these data we constrain the mass distribution and the anisotropy of the stellar orbits in this galaxy.
We have developed a non-parametric technique to determine the DF f(E,L^2) directly from the kinematic data. From Monte Carlo tests using the spatial extent, sampling, and error bars of the NGC 6703 data we find that smooth underlying DFs can be recovered to an rms accuracy of 12%, and the anisotropy parameter beta(r) to an accuracy of 0.1, in a given potential. An asymptotically constant halo circular velocity v_0 can be determined with an accuracy of +- \lta 50km/s.
For NGC 6703 we determine the true circular velocity at 2.6 R_e to be 250 +- 40km/s at 95% c.l., corresponding to a total mass in NGC 6703 inside 78'' (13.5 h_50^-1 kpc), of 1.6-2.6 x 10^11 h_50^-1 Msun. No model without dark matter will fit the data; however, a maximum stellar mass model in which the luminous component provides nearly all the mass in the centre does. In such a model, the total luminous mass inside 78'' is 9 x 10^10 Msun and the integrated M/L_B=5.3-10, corresponding to a rise from the center by at least a factor of 1.6.
The anisotropy of the stellar distribution function in NGC 6703 changes from near-isotropic at the centre to slightly radially anisotropic (beta=0.3-0.4 at 30'', beta=0.2-0.4 at 60'') and is not well-constrained at the outer edge of the data.
Our results suggest that also elliptical galaxies begin to be dominated by dark matter at radii of \sim 10kpc. | astro-ph | astro-ph | Mon. Not. R. Astron. Soc. 000, 000 -- 000 (0000)
Printed July 5, 2021
(MN LATEX style file v1.4)
Breaking the degeneracy between anisotropy and mass:
The dark halo of the E0 galaxy NGC 6703
Ortwin Gerhard1,2, Gunther Jeske2, R. P. Saglia3,⋆, Ralf Bender3,⋆
1 Astronomisches Institut, Universitat Basel, Venusstrasse 7, CH-4102 Binningen, Switzerland
2 Landessternwarte, Konigstuhl, D-69117 Heidelberg, Germany
3 Institut fur Astronomie und Astrophysik, Scheinerstr. 1, D-81679 Munchen, Germany
7
9
9
1
Accepted October 1997. Received May 1997, in original form June 1995.
t
c
O
3
1
1
v
9
2
1
0
1
7
9
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
ABSTRACT
We have measured line-of-sight velocity profiles (vps) in the E0 galaxy NGC 6703
out to 2.6Re. Comparing with the vps predicted from spherical distribution functions
(dfs), we constrain the mass distribution and the anisotropy of the stellar orbits in
this galaxy.
We have developed a non-parametric technique to determine the df f (E, L2)
directly from the kinematic data. We test this technique on Monte Carlo simulated
data with the spatial extent, sampling, and error bars of the NGC 6703 data. We find
that smooth underlying dfs can be recovered to an rms accuracy of 12% inside three
times the radius of the last kinematic data point, and the anisotropy parameter β(r)
to an accuracy of 0.1, in a known potential. These uncertainties can be reduced with
improved data.
By comparing such best-estimate, regularized models in different potentials,
we can derive constraints on the mass distribution and anisotropy. Tests show that
with presently available data, an asymptotically constant halo circular velocity v0 can
be determined with an accuracy of ± ∼< 50 km s−1. This formal range often includes
high -- v0 models with implausibly large gradients across the data boundary. However,
even with extremely high quality data some uncertainty on the detailed shape of the
underlying circular velocity curve remains.
50 kpc, where h50 ≡ H0/50km/s/Mpc) of 1.6 − 2.6 × 1011h−1
In the case of NGC 6703 we thus determine the true circular velocity at 2.6Re
to be 250 ± 40 km s−1 at 95% confidence, corresponding to a total mass in NGC 6703
inside 78′′ (13.5 h−1
50 M⊙. No
model without dark matter will fit the data; however, a maximum stellar mass model
in which the luminous component provides nearly all the mass in the centre does. In
such a model, the total luminous mass inside 78′′ is 9 × 1010 M⊙ and the integrated
B-band mass -- to -- light ratio out to this radius is ΥB = 5.3 − 10, corresponding to a
rise from the center by at least a factor of 1.6.
The anisotropy of the stellar distribution function in NGC 6703 changes from
near-isotropic at the centre to slightly radially anisotropic (β = 0.3 − 0.4 at 30",
β = 0.2 − 0.4 at 60") and is not well-constrained at the outer edge of the data, where
β = −0.5 − +0.4, depending on variations of the potential in the allowed range.
Our results suggest that also elliptical galaxies begin to be dominated by dark
matter at radii of ∼ 10 kpc.
Key words:
galaxies: kinematics and dynamics -- galaxies: individual -- line profiles
stellar dynamics -- dark matter -- galaxies: elliptical and lenticular --
⋆ Visiting Astronomer of the German-Spanish Astronomical Cen-
ter, Calar Alto, operated by the Max Plank Institut fur As-
tronomie, Heidelberg, jointly with the Spanish National Com-
mission for Astronomy
Current cosmological models predict that, similar to spiral
galaxies, elliptical galaxies should be surrounded by dark
matter haloes. The observational evidence for dark matter
1 INTRODUCTION
c(cid:13) 0000 RAS
2 O.E. Gerhard, G. Jeske, R.P. Saglia, R. Bender
in ellipticals is still weak, however. In a few cases masses
have been determined from X-ray observations (e.g., Awaki
et al. 1994, Kim & Fabbiano 1995) or HI ring velocities
(Franx, van Gorkom & de Zeeuw 1994). In others it has
been possible to rule out constant M/L from extended ve-
locity dispersion data (Saglia et al. 1993), from absorption
line profile measurements (Carollo et al. 1995, Rix et al.
1997), or from globular cluster or planetary nebula velocities
(e.g., Grillmair et al. 1994, Arnaboldi et al. 1994). Gravita-
tional lensing statistics (Maoz & Rix 1993) and individual
image lens separations (Kochanek & Keeton 1997) favour
models with extended dark matter haloes around ellipticals.
Despite of this, the detailed radial mass distribution in ellip-
tical galaxies remains largely unknown. Similarly, although
we know from the tensor virial theorem that giant ellipti-
cals are globally anisotropic (Binney 1978), their detailed
anisotropy structure is only poorly known.
The origin of this uncertainty is a fundamental de-
generacy -- in general, it is impossible to disentangle the
anisotropy in the velocity distribution and the gravitational
potential from velocity dispersion and rotation measure-
ments alone (Binney & Mamon 1982, Dejonghe & Merritt
1992). Tangential anisotropy, for example, can mimic the
presence of dark matter. Recent dynamical studies have in-
dicated, however, that the anisotropy of the stellar distribu-
tion function (df) is reflected in the shapes of the line-of-
sight velocity profiles (vps) in a way that depends on the
gravitational potential (Gerhard 1993, G93; Merritt 1993).
These papers argued that the extra constraints derived from
the vp measurements may be enough to break the degener-
acy and determine the mass distribution.
If this is correct, it provides a new method to investi-
gate the properties of the dark matter haloes around ellipti-
cal galaxies at intermediate radii: vps can now be estimated
from high-quality absorption line measurements out to ∼ 3
effective radii. Dynamical models are then used to disentan-
gle the effects of orbital anisotropy and potential gradient
on the vp shapes. In this paper, we implement these ideas
for the analysis of real data, analyzing the E0 galaxy NGC
6703. This study is part of an observational and theoretical
program aimed at understanding the mass distribution and
orbital structure in elliptical galaxies. Preliminary accounts
of this work have been given in Jeske et al. (1996) and Saglia
et al. (1997a).
We have obtained long -- slit spectroscopy for NGC 6703,
and have measured vps to ∼ 2.6Re with the method of Ben-
der (1990). The results are quantified by a Gauss-Hermite
decomposition (G93, van der Marel & Franx 1993) as de-
scribed by Bender, Saglia & Gerhard (1994; BSG). These
observations are described in Section 2. In Section 3 we use
simple dynamical models to describe the variation of the
vp shapes with anisotropy and potential, generalizing the
results of G93 for scale -- free models. These models, taken
from a systematic study of the relation between df and vps
in spherical potentials (Jeske 1995), are described in Ap-
pendix A. In Section 4 we develop a non-parametric method
for inferring the df and potential from absorption line pro-
file measurements. Tests on Monte Carlo generated data are
used to determine the degree of confidence with which the
df and potential can be inferred from real data. In Section
5 we analyse the kinematic data for NGC 6703. Compari-
son with the dynamical models from Jeske (1995) already
shows that no constant -- M/L model will fit the data. The
non -- parametric method developed in Section 4 is then used
to derive quantitative constraints on the mass distribution
and anisotropy of this galaxy. Finally, Section 6 presents a
discussion of the results and our conclusions.
2 OBSERVATIONS OF NGC 6703
NGC 6703 is an E0 galaxy at a distance D = 36 Mpc
(Faber et al. 1989) for H0 = 50 km/s/Mpc. From a B-
band CCD frame taken at the prime focus of the 3.5m
telescope on Calar Alto and kindly provided by U. Hopp
we have measured the inner surface brightness profile us-
ing the algorithm by Saglia et al. (1997b); this follows an
R1/4 law with Re = 30′′ = 5.2 h−1
50 kpc, or a Jaffe model with
rJ = 46.5′′ = 8.1 h−1
50 kpc, with small residuals (Fig. 1). A
3% increase of the sky value reduces the measured Jaffe ra-
dius to 35.5′′, a 1% decrease of the sky values increases it
to 54′′. Isophote shapes deviate little from circles (ǫ < 0.05,
a4/a < 0.005) and show small twisting (∆P A ≃ 10◦).
From the Jaffe profile fit we derive a fiducial (calibrated and
corrected for galactic absorption following Faber et al. 1989)
MB =−21.07, or luminosity LB = 4.16× 1010h−2
50 L⊙,B. Note
that the values of Re and MB derived here are slightly larger
than those (Re = 24′′, MB = −20.79) given by Faber et al.
(1989).
The spectroscopic observations were carried out in Oc-
tober 1994, May 1995 and August 1995 with the 3.5-m tele-
scope on Calar Alto, Spain. In all of the runs the same setup
was adopted. The Boller & Chivens longslit twin spectro-
graph was used with a 1200 line mm−1 grating giving 36
A/mm dispersion. The detector was a Tektronix CCD with
1024×1024 24µm pixels and the wavelength range 4760-5640
A. The instrumental resolution obtained using a 3.6 arcsec
wide slit was 85 km/s. We collected 1.5 hrs of observations
along the major axis of the galaxy, 4 hrs of observations per-
pendicular to the major axis and shifted to the North-East
of the center by 24 arcsec (0.8Re), and 13 hours of obser-
vations perpendicular to the major axis and shifted to the
North-East of the center by 36 arcsec (1.2Re). Spectra taken
parallel to the minor axis and shifted from the center allow
at the same time a good sky subtraction and the symmetry
check of the data points.
The analysis of the data was carried out following the
steps described by BSG. The logarithmic wavelength cali-
bration was performed at a smaller step (∆v = 30 km/s)
than the actual pixel size (≈ 50 km/s) to exploit the full ca-
pabilities of the Fourier Correlation Quotient method. A sky
subtraction better than 1% was achieved. The heliocentric
velocity difference between the May 1995 and the Septem-
ber 1994 - August 1995 frames was taken into account before
coadding the observed spectra. The spectra were rebinned
along the spatial direction to obtain a nearly constant sig-
nal to noise ratio larger than 50 per resolution element. The
effects of the continuum fitting and instrumental resolution
were extensively tested by Monte Carlo simulations. The
residual systematic effects on the values of the h3 and h4
parameters are expected to be less than 0.01. The resulting
fitted values for the folded velocity v , velocity dispersion
σ, h3 and h4 profiles are shown in Fig. 1 as a function of
the distance from the center, reaching ∼ 2.6Re. The v and
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
The dark halo of the E0 galaxy NGC 6703
3
points) is about 20 % (determined from the scatter of the
estimated signal to noise ratios).
Data points at the largest distances for the different
data sets have lower signal to noise ratio than the mean
and therefore have larger error bars. In addition, these data
points are expected to suffer more from the systematic ef-
fects due to the galaxy light contamination of the sky sub-
traction(see discussion in Saglia et al. 1993). They are in any
case consistent within the error bars with the more accurate
values derived from the other available spectra.
The observed scatter is sometimes slightly larger than
expected from the error estimates. This excess could be real
and due to the faint structures apparent in an unsharped
mask image of the galaxy. In particular this applies to the
asymmetries observed in the h4 profile in the central 5 arc-
sec. The negative h4 values detected for the first data points
of the 24 arcsec shifted spectrum are also real. They are de-
tected in the unbinned major axis spectra at R ≈ 22 arcsec
(dots in Fig. 1).
3 VELOCITY PROFILES IN SPHERICAL
GALAXIES
To better understand the relation between vp -- shape,
anisotropy, and gravitational potential, we have constructed
a large number of anisotropic models for spherical galaxies
in which the stars follow a Jaffe (1983) profile. The gravi-
tational potential was taken to be either that of the stars
(self-consistent case), or one with everywhere constant rota-
tion speed ('halo potential'). The latter case corresponds to
a mass distribution with a dark halo which has equal density
as the stars at r ≃ 0.4rJ , and equal interior mass at r ≃ rJ ,
where rJ is the scale radius of the Jaffe model. Anisotropic
quasi-separable distribution functions (dfs) g(E) h(E, L2)
were calculated by the method of Gerhard (1991; G91), but
contrary to G93 the circularity function (which specifies the
distribution of angular momenta on energy shells) was al-
lowed to vary with energy. These models include dfs in
which the anisotropy changes radially, from tangential to
radial or vice-versa, from isotropic to radial to tangential,
etc. (see Fig. 18). For comparison, we have also constructed
families of Osipkov (1979) - Merritt (1985) models which be-
come strongly radially anisotropic beyond a certain radius.
Details of this model data base and the properties of their
vps are given in Appendix A and in Jeske (1995); here we
only give a brief summary relevant for the comparison with
NGC 6703.
The shapes of the observable vps are most sensitive to
the anisotropy of the df, but depend also on the poten-
tial (G93). For rapidly falling luminosity profiles, the vps
are dominated by the stars at the tangent point. Then ra-
dially (tangentially) anisotropic dfs lead to more peaked
(more flat-topped) vps than in the isotropic case; in terms
of the Gauss-Hermite parameter h4, this corresponds to
h4 > (h4)iso and h4 < (h4)iso, respectively (Figs. 8,9 in G93).
Fig. 2 shows that these trends are also seen in the
present models in which both the luminosity density and the
anisotropy change with radius. An increase in radial (tan-
gential) anisotropy at intrinsic radius r is accompanied by
an increase (decrease) of h4 at projected radius R≃ r. The
correspondence is strongest in the models' outer parts, but
Figure 1. (a) Residuals of a Jaffe model fit to photometry for
NGC 6703. Full line: in surface brightness; dotted line: in the
curve of growth. The vertical dotted line marks the scaling Jaffe
radius rJ . (b) Folded mean velocity, (c) velocity dispersion, (d)
h3, and (e) h4 profiles. Crosses and filled circles refer to the two
sides of the galaxy and the major axis spectrum. The small dots
refer to the unrebinned spectrum (see text). Open and filled tri-
angles refer to the two sides of the galaxy and the spectrum taken
parallel to the minor axis and shifted 24 arcsec from the center.
Open and filled squares refer to the two sides of the galaxy and the
spectrum taken parallel to the minor axis and shifted 36 arcsec
from the center.
σ profiles are folded antisymetrically with respect to the
center for the major axis spectrum, symmetrically with re-
spect to the major axis for the spectra parallel to the minor
axis. Template mismatching was minimized by choosing the
template star which gave the minimal symmetric h3 profile
derived along the major axis of the galaxy. The systematic
effect due to the residual mismatching on the derived h4 val-
ues estimated from the remaining symmetric components is
less than 0.01.
The galaxy shows very little rotation (≈ 0 km/s for
R < Re, ≈ 20 − 30 km/s for R > Re). The (cylindrical) ro-
tation measured parallel to the minor axis is slightly larger
(≈ 35 km/s) along the 24 arcsec shifted spectrum, but con-
sistent with the peak velocity reached along the major axis
at R ≈ 22 arcsec. This is shown by the velocities derived
from the unbinned major axis spectra (dots in Fig. 1). The
velocity dispersion drops from the central ≈ 190 km/s to
≈ 140 km/s at Re/2, slowly declining to about 110 km/s in
the outer parts. The h3 and h4 values are everywhere close
to zero. The error bars are determined from Monte-Carlo
simulations. Noise is added to template stars (rebinned to
the original wavelength pixel size) broadened following the
observed values of σ and h4, matching the power spectrum
noise to peak ratio of the galaxy spectra. The accuracy of
the estimated error bars (the r.m.s. of 30 replica of the data
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
4 O.E. Gerhard, G. Jeske, R.P. Saglia, R. Bender
Jaffe, self
Jaffe, self
Jaffe, self
Jaffe, self
Jaffe, self
Jaffe, self
Jaffe, Halo
Jaffe, Halo
Jaffe, Halo
Jaffe, Halo
Jaffe, Halo
Jaffe, Halo
1
1
1
1
1
1
0.8
0.8
0.8
0.8
0.8
0.8
0.6
0.6
0.6
0.6
0.6
0.6
0.4
0.4
0.4
0.4
0.4
0.4
0.2
0.2
0.2
0.2
0.2
0.2
0
0
0
0
0
0
-0.2
-0.2
-0.2
-0.2
-0.2
-0.2
-0.4
-0.4
-0.4
-0.4
-0.4
-0.4
-0.6
-0.6
-0.6
-0.6
-0.6
-0.6
-0.8
-0.8
-0.8
-0.8
-0.8
-0.8
-1
-1
-1
-1
-1
-1
0.8
0.8
0.8
0.8
0.8
0.8
0.4
0.4
0.4
0.4
0.4
0.4
0
0
0
0
0
0
0.16
0.16
0.16
0.16
0.16
0.16
0.12
0.12
0.12
0.12
0.12
0.12
0.08
0.08
0.08
0.08
0.08
0.08
0.04
0.04
0.04
0.04
0.04
0.04
0
0
0
0
0
0
-0.04
-0.04
-0.04
-0.04
-0.04
-0.04
-0.08
-0.08
-0.08
-0.08
-0.08
-0.08
0.01
0.01
0.01
0.01
0.01
0.01
0.1
0.1
0.1
0.1
0.1
0.1
1
1
1
1
1
1
10
10
10
10
10
10
0.01
0.01
0.01
0.01
0.01
0.01
0.1
0.1
0.1
0.1
0.1
0.1
1
1
1
1
1
1
10
10
10
10
10
10
Figure 2. Fitted projected velocity dispersion σfit, anisotropy parameter β, and vp-parameter h4 for representative Jaffe models in
self-consistent (left) and halo potential (right). The models shown are radially and tangentially anisotropic models constructed with the
method of G91 (dashed lines), the isotropic model (solid), and two Osipkov-Merritt models (dotted lines). Note that while β is a function
of three-dimensional radius r and σfit, h4 are observed quantities depending on projected radius R, there is a close correspondence
between features in these profiles. See text.
is also seen to a lesser extent in the centre of a Jaffe model
where ρ(r) ∝ r−2 -- contrary to a homogeneous core where
radial orbits lead to broadened vps (Dejonghe 1987). Quan-
titatively the correspondence depends also on the anisotropy
gradient. Osipkov-Merritt-models show a reversal of this
trend near their anisotropy radius ra due to the large num-
ber of high-energy radial orbits all turning around near ra;
this leads to flat-topped vps in a small radius range near
ra. However, the properties of these models are extreme and
they are in general not very useful for modelling observed
vps.
Fig. 2 and Figs. 8,9 in G93 also show that as the mass
of the model at large r is increased at constant anisotropy,
both the projected dispersion and h4 increase. Increasing
β at constant potential, on the other hand, lowers σ and
increases h4. This suggests that by modelling σ and h4 both
mass M (r) and anisotropy β(r) can in principle be found.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
N (R, vk) = Z dzZZ dvxdvyf (E, L2)
= Z dzZZ dvxdvyf [ 1
(1)
2 v2
r + 1
2 v2
t + Φ(r), r2v2
t ].
4 MODELLING ABSORPTION LINE PROFILE
DATA
Having seen the effect of anisotropy and potential variations
on the line profile parameters, we now proceed to construct
an algorithm by which the distribution function and poten-
tial of a spherical galaxy can be constrained from its ob-
served σ and h4 -- profiles. Such absorption line profile data
contain a subset of the information given by the projected
distribution function N (R, vk), which in the spherical case
is related to the full df by
Here the position on the sky is specified by R = (x, y);
R = R. Velocities in the sky plane are denoted by (vx, vy),
z and vk are the line-of-sight position and velocity, and vr
and vt are the intrinsic radial and tangential velocities. In
spherical symmetry, the df f (E, L2) is a function of energy
and squared angular momentum only.
Notice from eq. (1) that the projected df depends lin-
early on f , but non-linearly on the potential Φ(r). Thus
f will generally be easier to determine from N (R, vk) than
Φ. Moreover, while considerations like those in the last sec-
tion do suggest that, in spherical symmetry and for positive
f (E, L2), both the df and the potential can be determined
from N (R, vk), there is no theoretical proof that this is in
fact true. We only know that in a fixed spherical potential
the df is uniquely determined from N (R, vk) (Dejonghe &
Merritt 1992). For these reasons we have found it useful to
split our problem into two parts (see also Merritt 1993):
(1) We fix the potential Φ, and from the photometric
and kinematic data determine the "best" df f for this po-
tential. Because in practice the surface brightness (sb) pro-
file is much better sampled than the kinematic observations
and also has smaller errors, we treat it separately and de-
termine the stellar luminosity density j(r) at the beginning.
The kinematic data are then used to determine the "best" f
for given j(r) and Φ(r), by approximately solving equation
(1) as a linear integral equation.
(2) We then vary Φ to find that potential which allows
the best fit overall. At present, it is not practical in step (2)
to attempt to determine the potential non-parametrically.
Rather, we choose a parametrized form for Φ, and find the
region in parameter space for which the "best" df as deter-
mined in step (1) reproduces the data adequately.
In view of the modelling of NGC 6703 in Section 5, we
have considered the following family of potentials, includ-
ing a luminous and a dark matter component: The stellar
component is approximated as a Jaffe (1983) sphere, with
scale-radius rJ and total mass MJ , so that
ΦL(r) =
GMJ
rJ
ln
r
r + rJ
.
The dark halo has an asymptotically flat rotation curve,
vc(r) = v0
r
√r2 + r2
c
,
so that its potential is
ΦH (r) = 1
2 v2
0 ln(r2 + r2
c ).
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
(2)
(3)
(4)
The dark halo of the E0 galaxy NGC 6703
5
This is specified by the asymptotic circular velocity v0 and
the core radius rc. Both the luminous and dark halo compo-
nents can be modified when needed, and need not be analytic
functions.
In testing our method below, we use parameters
adapted to NGC 6703. This galaxy is well-fit by a Jaffe
profile (Section 2), so rJ is known. This leaves three free
parameters, the mass MJ or mass -- to -- light ratio Υ of the
stellar component, and the halo parameters rc and v0. If
one assumes that the central kinematics is dominated by
the luminous matter, Υ can be determined. Then only the
two halo parameters rc and v0 are free. The assumption of
maximum stellar Υ is similar to the maximum disk assump-
tion in spiral galaxies.
In any of the potentials specified by eqns. (2) -- (4) we
determine the df by the algorithm described in Section 4.1
below. To assess the significance of the results obtained, we
test the algorithm on Monte Carlo -- generated pseudo data in
Section 4.2. For kinematic data with the spatial extent and
observational errors such as measured for NGC 6703, the
algorithm recovers a smooth spherical df ∼ 70% of the time
to an rms level of ∼ 12%, taken inside three times the radius
of the outermost kinematic data point. In Section 4.3, we
investigate the degree to which the gravitational potential
can be constrained from similar data.
4.1 Recovering f from σ and h4, given Φ
As discussed above, the projected distribution function
N (r, vk) suffices to determine df f (E, L2) uniquely. In prac-
tice, however, only incomplete and noisy data are available
in place of N (r, vk), and contrary to the two-dimensional
function N (r, vk), the observed σ(Ri) and h4(Ri) contain
only one-dimensional information. This suggests that we can
hope to recover only the gross features of f from such kine-
matic data. Indeed, the anisotropy parameter β(r) seems
to be essentially fixed from accurate h4 measurements (e.g.,
Figs. 8,9 in G93). Local fluctuations in the df will be inac-
cessible, but as we will show, smooth dfs can be recovered
with reasonable accuracy from presently available data.
To solve the inversion problem, we first compute a set of
self -- consistent models fk(E, x) for the stellar density j(r),
in the fixed potential Φ(r). The fk(E, x) are models of the
kind discussed in Section 2 and Appendix A; E and x are
the energy and circularity integrals of the motion. Then we
write the df as a sum over these "basis" functions:
akfk(E, x).
(5)
f = Xk=1,K
We do not need to use a doubly infinite, complete set of
basis functions because most of the high -- frequency struc-
ture represented by the higher -- order basis functions in such
a set will be swamped by noise in the observational data.
It is sufficient to choose the number of basis functions, K,
and the fk(E, x) themselves such that the data can be fit-
ted with a mean χ2 ≃ 1 per data point. We have found it
advantageous to use the isotropic model plus tangentially
anisotropic basis models, because with these the anisotropy
of the final composite df (5) can be varied in a more local
way than with radially anisotropic components. Since the
ak can be negative, it is of course no problem to generate a
6 O.E. Gerhard, G. Jeske, R.P. Saglia, R. Bender
(E = Ei, x = xj). Subject to these constraints the ak are
to be determined such that the kinematics predicted from
the df (5) match the observed kinematics in a minimum
χ2 -- sense.
The comparison between model and data is not entirely
straightforward, however. Since the measured (v, σ, h3, h4)
are obtained by fitting to the line profile (the observational
analogue of the projected df), they unfortunately depend
non-linearly on the galaxy's underlying df. Thus we cannot
use the observed v and σ in a linear least squares algorithm
to determine the ak from the data -- they cannot be writ-
ten as moments of f . In the fitting process, they have to be
replaced by quantities that do depend linearly on the df.
Moreover, the error bars for these new quantities generally
depend not only on the observed error bars of v and σ, but
also on the errors and the values of the line profile param-
eters h3 and h4. They must therefore be determined with
some care.
We have investigated several schemes along these lines.
The following seemed to do best in recovering a known un-
derlying spherical df from pseudo data. From the measured
(σ, h4), we compute an approximation to the true velocity
dispersion (second moment) σ2(Ri), by integrating over the
line profile (for negative h4, until it first becomes negative).
We also evaluate a new set of even Gauss-Hermite moments
The velocity profile moments of the basis function mo-
isotropic model with the galaxy's stellar density, in the cur-
rent potential Φ(r).
sn(Ri;eσ) from the data, using fixed, fiducial velocity scales
eσ(Ri) (G93, Appendix B). We have found it convenient to
take for theseeσ(Ri) the velocity dispersions σiso(Ri) of the
dels are transformed to the same velocity scaleseσ(Ri). The
moments σ2(Ri) and sn(Ri;eσ) of the composite df then
depend linearly on the corresponding moments of the basis
function models. For a regularized model they are smooth
functions of R, of which the (noisy) observational moments
are assumed to be a random realisation within the respective
errors.
To determine the best-fitting coefficient ak we minimize
the sum over data points i of all
(8)
(9)
σ,i ≡ w2
χ2
and
χ2
n,i ≡ w2
akσ2
σ(Ri)hσ2(Ri) − Xk=1,K
n(Ri)hsn(Ri;eσ) − Xk=1,K
k(Ri)i2
n (Ri;eσ)i2
aks(k)
for n = 2, 4. These equations make use of the fact that all
the fk are self-consistent models for the same j(r), so that
all surface density factors µk(Ri) = µ(Ri) cancel.
To determine the weights wσ and wn, we have done
Monte Carlo simulations studying the propagation of the
observational errors ∆σ and ∆h4. Based on the results of
these simulations, we have chosen
w−1
w−1
w−1
σ (Ri) = 2.4σ(Ri)[σ(Ri)/σ(Ri)]∆σ(Ri),
2 (Ri) = 0.7∆σ(Ri)/σ(Ri),
4 (Ri) = 0.95α∆h4.
(11)
(10)
(12)
The coefficients are representative in the range of values
taken by the observed error bars and the measured h4. In the
presently used fitting procedure, the additional dependence
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Figure 3. Anisotropy parameter β for a subset of basis functions
fk used in the self -- consistent Jaffe potential (top) and in a mixed
Jaffe plus halo model (bottom). The top full line in each panel
shows the isotropic model. All other basis models are tangentially
anisotropic. Two values of asymptotic anisotropy as r → ∞ are
used (full and dashed lines). For illustration, the long -- dashed lines
show the β -- profiles of the best -- fitting dfs derived with these
bases from the NGC 6703 data, in both potentials.
radially anisotropic df from the isotropic model plus a set
of tangential basis models.
In Fig. 3 we illustrate the basis models we have used,
by plotting radial profiles of the anisotropy parameter β(r)
for a subset of them. For these basis functions β(r) can be
regarded as a measure of the extent of the df in circularity
x, at the energy E = Φ(r). The figure shows that the basis
resolves three steps in x in the limits r → 0 and r → ∞;
at intermediate energies, it has much finer resolution. The
majority of the number of functions is spent on resolving
the energy dependence, i.e., on placing the main gradient
zones of the functions fk in the (E, x) plane on a relatively
dense grid in energy. The main difference to using power law
components (Fricke 1952, and, e.g., Dejonghe et al. 1996) is
that our basis models are already reasonable in the sense
that they are viable dynamical models for the density in
question, and that the superposition is used only to match
< 20 such func-
tions; this proved sufficient even for analysing pseudo data
of much better quality than we have for NGC 6703.
the kinematics. We have typically used K ∼
Each of the fk(E, x) reproduces the stellar density dis-
tribution j(r); so the ak satisfy
Moreover, the coefficients ak must take values such that
ak = 1.
Xk=1,K
Xk=1,K
akfk(E, x) ≥ 0
(6)
(7)
everywhere in phase-space. In practice, these positivity con-
straints are imposed on a grid in energy and circularity
on h4 and the respective "other" error bar is neglected -- the
tests below show that this leads to satisfactory results. Since
to first order the s2 -- moment thus measures the shift from
σ to σ, the parameter α in eq. (13) will be near α ≃ 2. We
have fixed α by requiring that the distributions of χ2
σ and
χ2
h4 have equal width (see below); this results in α = 1.7.
Finally, we assume that the df underlying the observed
kinematics is smooth to the degree that is compatible with
the measured data. Clearly, unless such an assumption is
made, it is impossible to determine a function of two vari-
ables, f (E, L2), from a small number of data points with
real error bars. One way to ensure that the df is smooth is
to use only a small number of terms in the expansion (5).
However, this is not a good way of smoothing as it biases
the recovered df towards the functional forms of the few
fk that are used in the sum. A better way of smoothing is
the method of regularization, as recently discussed by Mer-
ritt (1993) in a similar context. Regularization has tradition
in other branches of science, and different variants exist; for
references see Press et al. (1986) and Merritt's paper (1993).
In the algorithm here, regularization is implemented by tak-
ing the number of basis functions large enough (typically,
16 − 20) that the data can be modelled in some detail, and
then constraining the second derivatives of the composite
df to be small. That is we also seek to minimize
Λ(f )ij ≡ w2
∂2f
r (E) ×
∂2f
∂E∂x(cid:17)2
∂x2(cid:17)2(cid:21)E=Ei,x=xj
+(cid:16) ∂2f
on a grid of points (Ei, xj) in the energy and circularity in-
tegrals. Here we normalize to the isotropic df to ensure that
fluctuations in the composite df are penalized equally at all
energies; i.e., wr(E) = 1/fiso(E). The constant DE is pro-
portional to the range in potential energy in the respective
model.
To find a regularized spherical df for given kinematic
data in a specified potential, we thus minimize the quan-
tity
∆2 ≡
IXi=1(χ2
σ,i + Xn=2,4
χ2
n,i) + λXi,j
Λ(f )ij
(14)
for given regularization parameter λ, subject to the equality
and inequality constraints (6) and (7). For the actual nu-
merical solution we have used the NETLIB routine LSEI by
Hanson and Haskell (1981). Once the best model is found,
we redetermine its quality by evaluating its deviations from
the actually measured data:
(cid:20)(cid:16)D2
E
∂E2(cid:17)2
+ 2(cid:16)DE
(13)
χ2
σ = I −1
χ2
h4 = I −1
IXi=1hσ(Ri) − σ(f )(Ri)i2
4 (Ri)i2
IXi=1hh4(Ri) − h(f )
/(∆σ)2(Ri),
(15)
/(∆h4)2(Ri).
(16)
The parameters σ(f )(Ri) and h(f )
4 (Ri) are determined by
fitting a Gauss -- Hermite series to the velocity profiles of the
best fitting model; I is the number of kinematic data points.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
The dark halo of the E0 galaxy NGC 6703
7
4.2 Tests with model data
We have tested this method by applying it to kinematic
datasets generated in the following Monte Carlo -- like way.
First, velocity dispersions and h4-parameters are calculated
from a theoretical df of specified anisotropy in a known
potential, and are interpolated to the radii Ri/rJ at which
observed data points are available for NGC 6703. Error bars
at these Ri are taken to be either the measured error bars
for NGC 6703 (a realistic dataset), or are taken to be inde-
pendent of radius with ∆σ = 3 km s−1 and ∆h4 = 0.01 (an
idealized dataset). Then pseudo data are generated from the
model values of σ and h4 at Ri by adding Gaussian random
variates with 1-σ dispersion corresponding to the respec-
tive ∆σ or ∆h4 error bar at this point. Figs. 4, 5 show
datasets generated in this way from a radially anisotropic
model in a potential of a self-consistent Jaffe sphere with
rJ = 46.′′5, GMJ /rJ = 272 km s−1, and a dark halo with
v0 = 220 km s−1, rc = 56′′. This potential was chosen be-
cause it lies in the middle of the range of acceptable poten-
tials for NGC 6703 (see Sect. 5.2), so that any systematic
errors in our analysis will be similar for this model and for
the galaxy itself.
We have used the regularized inversion algorithm de-
scribed in the last subsection to analyse several such pseudo
data sets, and have determined composite dfs as a func-
tion of the regularization parameter λ. The algorithm was
given 16 basis function models. These included the isotropic
model and a variety of tangentially anisotropic models with
different anisotropy radii and circularity functions but not,
of course, the radially anisotropic model from which the data
are drawn (see Fig. 3 and Appendix A). For each composite
model returned by the algorithm we have determined two
diagnostic quantities. The first is the mean χ2 per σ and h4
data point, χ2
σ+h4, which measures the level at which this
model fits the data from which it was derived. The second
is the rms deviation between the returned df and the true
df of the model from which the data were drawn, in some
specified energy range.
The use of these diagnostic quantities requires some fur-
ther comments. As usual, the number of degrees of free-
dom in such a regularized inversion problem is not well --
determined. For near-zero λ, in the case at hand we can
adjust 16 coefficients ak and an overall mass scale. The to-
tal number of σ and h4 -- data points is 70; thus in this limit
the number of degrees of freedom is 53. For large λ, on the
other hand, the recovered df will be linear in both E and
x and the values of the ak are essentially fixed. Then the
number of degrees of freedom approaches 69. In both cases,
it is of the order of the number of data points. Hence the
use of χ2
σ+h4 per σ and h4 data point instead of a reduced
χ2.
The second diagnostic measuring the accuracy of the re-
covered df must clearly depend on the range in energy over
which it is calculated. Typically, a kinematic measurement
at projected radius Ri contains information about the df
in a range of energy above the energy of the circular orbit
at radius Ri. The precise upper end of this range is model --
specific; it depends on the potential, the kinematic proper-
ties of the df itself, and, through the projection process,
also on the stellar density profile. The use of the outermost
kinematic data points thus contains, explicitly or implic-
8 O.E. Gerhard, G. Jeske, R.P. Saglia, R. Bender
0.16
0.12
0.08
0.04
0
-0.04
-0.08
-0.12
-0.16
0.6
0.4
0.2
0
-0.2
0
10 20
40 50
30
R [arcsec]
60 70
80
0
10 20
40 50
30
r [arcsec]
60 70
80
0.16
0.12
0.08
0.04
0
-0.04
-0.08
-0.12
-0.16
0.6
0.4
0.2
0
-0.2
0
10 20
40 50
30
R [arcsec]
60 70
80
0
10 20
40 50
30
r [arcsec]
60 70
80
Figure 4. Model analysis of pseudo-data generated from a radi-
ally anisotropic model df with the sampling and the measured
error bars of NGC 6703 (see text). The full curves show the true
profiles of projected velocity dispersion, line-of-sight velocity dis-
tribution parameter h4, and anisotropy parameter β of the un-
derlying df. The dashed and dotted lines show the σ, h4, and β
profiles of five regularized composite dfs which were computed
by the method of Sect. 4.1 with λ = 10−4. One of the four curves
in each panel corresponds to the data points actually shown in
this figure; the other three curves derive from statistically identi-
cal kinematic data sets with randomly different values for σ and
h4 within the same (Gaussian) errors.
itly, assumptions on the radial smoothness of these quan-
tities. In a typical elliptical galaxy problem, the df must
be known fairly accurately at around 3Ri for the projected
kinematics at Ri to be securely predicted, and for radially
anisotropic models, even the values of the df near 10Ri can
make some difference. The rms residuals in the recovered
df given below have therefore been calculated in a range of
energies extending from Φ(r = 0) to Φ(r = 3Rm), where the
last kinematic data point in the data sets used is located at
Rm = 1.68rJ .
Figure 6 shows these quantities as a function of the regu-
larization parameter λ for the two pseudo data sets shown in
Figs. 4, 5. For small values of λ, the composite models fit the
Figure 5. Model analysis of pseudo-data generated from the
same radially anisotropic model df as in Fig. 4 but assuming
radially constant error bars ∆σ = 3 km s−1 and ∆h4 = 0.01. The
full curves again show the true σ, h4, and β profiles of the under-
lying df. The dashed lines show the projected σ and h4 profiles
derived by the method of Sect. 4.1 from the pseudo-data shown,
with λ = 6×10−5. The results obtained for these projected quanti-
ties from statistically identical data sets are now nearly identical.
The dotted curves show the uncertainties that remain in the de-
projected quantity β even with such small error bars.
data accurately, but the recovered df is not very accurate
because it contains large spurious oscillations depending on
the particular values of the data points. For large values of
λ, the models are so heavily smoothed that they neither fit
the data well, nor do they represent a good approximation
to the true df. The optimal regime is where the smooth-
ing is large enough to damp out the spurious oscillations,
but still permits resolving the important structures in the
underlying df. For values of λ in this regime the fit to the
data is still satisfactory, and the representation of the df is
optimal. Fig. 6 shows that the rms residuals of the df go
through minima at values of λ ≃ 2 × 10−4 for the pseudo
data in Fig. 4 and λ ≃ 5 × 10−5 for those in Fig. 5 [when
DE = 1; see below eq. (13)].
σ+h4(λ) curves were found to be al-
The shapes of the χ2
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
The dark halo of the E0 galaxy NGC 6703
9
Figure 6. Model results as a function of the regularization pa-
rameter λ, for the two pseudo-data sets shown in Figs. 4 and 5.
The two data sets are flagged by triangles (Fig. 4) and pentagons
(Fig. 5). The upper two curves marked by the open symbols show
the total χ2 per σ and h4 data point of regularized composite
models with 16 basis functions. The lower two curves marked by
the filled symbols show the rms deviation between the recovered
composite model dfs and the true df from which the data sets
were generated. This rms deviation was evaluated on a grid in
energy and angular momentum corresponding to radii < 3 times
the radius of the outermost data point.
ways similar to the upper curves in Fig. 6. The shapes of the
corresponding lower curves in Fig. 6 are more variable. The
resulting optimal values for λ can vary, depending on the
random realisation of the data within the assumed Gaussian
errors, as well as on the distribution function and potential
from which the model values are drawn. We have therefore
investigated 100 realisations of data generated from each of
several model dfs and potentials. Based on these experi-
ments we have fixed optimal values of λ = 1 × 10−4 and
λ = 6 × 10−5 for data with error bars like those in Figs. 4
and 5, respectively, for all models that include a dark halo
component.
We have done similar experiments with a self-consistent
model that is radially anisotropic near the center and tan-
gentially anisotropic in its outer parts, such as might be
relevant in tests for dark matter at large radii. In these tests
we have used 20 basis functions. To match the corresponding
pseudo-data with similar accuracy requires smaller optimal
values of λ (∼ 3 × 10−9), so as to compensate for the larger
derivatives in eq. (13).
σ+h4 and
of the rms residual in the df, as described, for the dynami-
cal models recovered from several such sets of Monte Carlo
data. From the top panel it is seen that our fitting algo-
rithm will match kinematic data in a known potential with
σ+h4 ≤ 1 about 60−70% of the time, and with χ2
χ2
σ+h4 > 1.28
only < 5% of the time. These numbers are similar to those
expected from a χ2 -- distribution with 70 degrees of freedom
Fig. 7 shows the cumulative distributions of χ2
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Figure 7. Top: The cumulative distribution of the total normal-
ized χ2
σ+h4, for dynamical models recovered from 100 random
Gaussian data sets derived from the underlying true df and the
observational errors in Figs. 4 (solid line) and 5 (dashed). Bot-
tom: Cumulative distribution of residuals between true and re-
covered distribution function, evaluated on a grid extending to
three times the radius of the last data point, for the same 100
datasets from both models. Also shown in the top panel are the
cumulative χ2
σ+h4 -- distributions for data drawn from radially an-
isotropic models in the two potentials corresponding to two of
the extreme solid lines in Fig. 11, with the same error bars as
in Fig. 4 (dot -- dashed lines), and for a self -- consistent model with
more complicated anisotropy structure (dotted line).
(for Gaussian data), which should describe the statistical
deviations of the Monte Carlo data points from the under-
lying true df. If in modelling the data for NGC 6703 a level
of χ2
σ+h4 < 1.28 cannot be reached, the assumed potential is
not correct with 95% confidence.
The bottom panel of Fig. 7 shows the distribution of
residuals in the recovered df for one model. There are three
factors which limit the degree to which a given underlying
df can be recovered. The first is determined by the data,
i.e., the size of the error bars, the sampling, the fact that
there are no measurements at large radii, etc. The second
is the level of detail that can be resolved by the modelling,
given the finite number and the particular form of the basis
functions. The third is the amount of small -- scale gradients
in the model itself. Figs. 6 and 7 show that, if the potential is
known, data like those for NGC 6703 (Fig. 4) allow recover-
< 12% inside
ing reasonably smooth dfs to an rms level of ∼
3Rm about 60 − 70% of the time. Data with much smaller
error bars but the same sampling (Fig. 5) would give an rms
< 8% for the other two halo models shown
level of ∼
< 10% (∼
10 O.E. Gerhard, G. Jeske, R.P. Saglia, R. Bender
in the top panel of Fig. 7). In the radial -- tangential self --
consistent model the df to 3Rm is recovered with an rms
accuracy of 16% and 35% for data with error bars like those
in Figs. 5 and 4, respectively. These comparisons and the
results of Fig. 7 show that the former values are dominated
by the measurement uncertainties rather than the resolu-
tion in the modelling. This conclusion would be different for
highly corrugated true dfs. However, we would not be able
to recover such dfs from realistic data in any case.
The kinematics of the regularized composite dfs de-
rived from pseudo-data with the chosen optimal λ values
are shown in Figs. 4 and 5. (For brevity, such model dfs
obtained with near -- optimal λ will henceforth be denoted as
"best -- estimate models".) The good match to the data points
is apparent. The differences in the intrinsic anisotropy pa-
rameter β between the recovered models and the true model
are larger. For the data set generated with the observed er-
ror bars of NGC 6703 the recovered β values are uncertain
by ∆β ≃ ±0.1, and slightly more at the largest radii where
the observational errors are large. For the pseudo-data with
the small error bars in Fig. 5, this uncertainty is reduced.
4.3 Constraining Φ
So far we have shown how the stellar df can be recovered
from vp -- data in a known spherical potential. In this Section
we return to the discussion of Section 3 and investigate the
degree to which the gravitational potential itself can be con-
strained. Eventually, it will clearly be important to answer
the theoretical question of whether, in principle, the gravita-
tional potential is nearly or even uniquely determined from
the projected df N (r, vk). However, we shall not attempt
to do this here; previous theoretical work suggests that the
range of potentials consistent with ideal data is small (G93,
Merritt 1993, see Section 3). More relevant at the moment
perhaps is the more practical question of how well the po-
tential can be constrained from real observational data with
realistic error bars, finite radial extent, and limited sam-
pling. What is the range of pairs (f, Φ) that correspond to
the same data? How does this range shrink as the data im-
prove?
Here we investigate these questions with a view to the
analysis of the NGC 6703 data below. We again use the
model underlying Figs. 4, 5; this is a radially anisotropic df
in the potential of a self -- consistent Jaffe sphere and a dark
halo with parameters GMJ /rJ = 272 km s−1, rc/rJ = 1.2
and v2
0 /(GMJ /rJ ) = 0.808. As before, random Gaussian
data sets were generated from this model, with the positions
and error bars of the data points (i) as in Fig. 4, (ii) as in
Fig. 5. The last data point is at Rm = 1.68rJ , as for the
NGC 6703 data, well beyond two effective radii. However,
contrary to Section 4.2, the data sets used in the following
were specially selected such that χ2
h4 ≃ 1. (It
turns out that the data points shown in Fig. 4 have less than
5% probability according to Fig. 7.).
σ ≃ 1 and χ2
For these pseudo data we have determined best --
estimate dfs in a number of assumed potentials, includ-
ing the underlying true potential. The potentials were cho-
sen such that (i) they correspond to approximately constant
mass -- to -- light ratio for r ≪ rJ , and (ii) they form a sequence
of varying true circular velocity vc(Rm) at the radius of the
last data point. The sequence, with the slightly different ve-
Figure 8. Goodness of fit χ2
σ+h4 for best -- estimate dfs derived
in a sequence of luminous plus dark matter potentials. The se-
quence of potentials, with rotation curves similar to those shown
in Fig. 11, is here parametrized by the total circular velocity at
the last observational radius, vc(Rm). The dfs were fitted to
pseudo data generated from a model with vc(Rm) = 242 km s−1.
Solid pentagons: data points with error bars as in Fig. 4. Open
pentagons: data points with error bars as in Fig. 5.
Fig. 8 shows the goodness -- of -- fit χ2
locity normalisations appropriate for NGC 6703, is shown in
Fig. 11. For each potential, specified by the selected values
of the parameters rc/rJ and v2
0 /(GMJ /rJ ), we determined
the goodness -- of -- fit χ2
σ+h4 as a function of the velocity scale
GMJ /rJ from the corresponding best -- estimate models. Fi-
nally, we computed the χ2
σ+h4 of the best -- estimate df for
the optimum velocity scale. This optimal velocity scale usu-
ally turns out slightly different for the two pseudo data sets.
σ+h4 of the opti-
mal sequence of best -- estimate models as a function of
vc(Rm), as determined for both model datasets. The un-
derlying true potential has vc(Rm) = 242 km s−1 with the
adopted parameters. Fig. 8 shows that only potentials with
vc(Rm) < 230 km s−1 can be ruled out (have less than 5%
probability according to Fig. 7) from the data with "re-
alistic" error bars. This conclusion is not surprising given
how large these error bars are. However, the surprise is that
also with "idealized" data there remains a range of poten-
tials with larger vc(Rm) than the true value, in which these
data can be fit no worse than in the true potential: formally,
235 km s−1 < vc(Rm) < 285 km s−1 with 95% confidence.
Fig. 9 shows the predicted kinematics and the intrinsic
anisotropy of the four models in Fig. 8 that match the "ideal-
ized" data set with χ2
σ+h4 ≃ 1. The best -- estimate dfs in the
potentials with higher vc(Rm) than that of the true potential
achieve their good fit to the data by the following means:
Inside rJ they compensate the potential's higher circular
speed by a larger radial anisotropy, which leads to slightly
larger h4 -- values (cf. Section 3). This effect is too small to
be detected even with the small error bars. Outside rJ , they
compensate by a radially increasing tangential anisotropy. In
this way, the velocity dispersions near the edge of the model
can be lowered, because the number of high -- energy orbits
coming in from outside is reduced. Such orbits would con-
tribute large line -- of -- sight velocities near their pericentres.
The h4 values in the region concerned are also only barely
affected, because the extra tangential orbits near radius r
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
The dark halo of the E0 galaxy NGC 6703
11
To test this, we have constructed a truly idealized data set
of two times seventy data points with "idealized" small error
bars as in Fig. 5, evenly spaced in radius, and extending to
6rJ . A model with potential differing only in the halo core
radius (66% of the true value) was found to fit even these
data with χ2
σ+h4 = 0.97, while for a model with different
halo core radius and different asymptotic circular velocity
(by 30 km s−1) a satisfactory fit could not be found.
We draw the following conclusions from these experi-
ments:
(i) Velocity profile data with presently achievable er-
ror bars contain useful information on the gravitational po-
tentials of elliptical galaxies. In particular, constant -- M/L
models are relatively easy to rule out once the data extend
beyond 2Re. The examples that we have studied in detail,
tuned to the NGC 6703 data, certainly belong to the less
favourable cases, because the dispersion profile is falling.
(ii) The detailed form of the true circular velocity curve
is much harder to determine. Conspiracies in the df are
possible that minimize the measurable changes in the line
profile parameters. A good way to parametrize the results is
in terms of the circular velocity at the radius of the outer-
most data point. With presently available data, this can be
determined to a precision of about ±50 km s−1.
(iii) Better results can be expected from higher -- quality
data, of the sort one could expect from the new class of 10m
telescopes. However, even with such data some uncertainty
on the detailed circular velocity curve will remain -- regard-
less of whether or not in theory the potential is uniquely
determined from the projected df N (r, vk). Therefore, the
combination of the type of analysis presented here with other
information (e.g., from X-ray data) will give the most pow-
erful results.
5 THE ANISOTROPY AND MASS
DISTRIBUTION OF NGC 6703
5.1 Constant -- M/L model fits
The sb-profile of NGC 6703 is well fitted by a Jaffe mo-
< 15% around R ≃ 25′′,
del. The largest local residuals are ∼
< 10% around R ≃ 40′′, and smaller elsewhere, in particular
∼
at R > 60′′. A non -- parametric inversion of the the surface
brightness profile showed that the deviations from a Jaffe
density law are not significantly larger than those quoted
in sb. In the curve of growth, measuring the total luminos-
ity inside R, the residuals are everywhere less than ∼ 2%
(Fig. 1). Since the potential Φ(r) is determined by the to-
tal mass M (< r), we can thus to good approximation use a
Jaffe model for the gravitational potential of the stars. This
enables us to also compare our kinematic data with a large
set of self-consistent dynamical models from Jeske (1995).
We have fitted all models from this database to the
observational data for NGC 6703, taking the fitted Jaffe
radius rJ = 46.′′5. All data points from Fig. 1 were included
and weighted equally with their individual error bars, except
for two adjustments. (i) The error bars of the three h4-points
near 25" have been set equal to their standard deviation, and
(ii) the error bars of the innermost eleven h4-points have
been set to twice the measured values. These modifications
prevent the total χ2
h4 to be dominated by these points which
0.16
0.12
0.08
0.04
0
-0.04
-0.08
-0.12
-0.16
0.6
0.4
0.2
0
-0.2
0 10 20 30 40 50 60 70 80 90 100
R [arcsec]
0 10 20 30 40 50 60 70 80 90 100
r [arcsec]
Figure 9. Model fits to pseudo velocity dispersion and h4 data
with "idealized" error bars (see text, and Fig. 5). The full line
shows the kinematics of the model from which these pseudo data
were generated. The dashed lines show the kinematics of best --
estimate dfs in a sequence of potentials with increasing halo
circular velocity, all of which fit the data perfectly with χ2
σ+h4 ≃
1. The intrinsic anisotropy of these models (dotted lines) and of
the true model are shown in the bottom panel.
and the lack of higher energy radial orbits from beyond r
nearly compensate.
Clearly, this mechanism will work less well both when
the difference between the true and attempted circular ve-
locity curves increases, and when the radial extent, sam-
pling, and quality of the data points improve. With data
yet better and more extended than those in Fig. 5, some
of the potentials consistent with the presently used model
data could probably be ruled out. Thus it appears that, if
the asymptotic circular velocity is constant, then the value of
that constant can be determined accurately with sufficiently
high quality data.
On the other hand, the potentials we have used are
very simply parametrized functions. There might well exist
more complicated potential functions, whose circular veloc-
ity curves differ in only a restricted range of radii, that would
be impossible to distinguish even with extremely good data.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
12 O.E. Gerhard, G. Jeske, R.P. Saglia, R. Bender
NGC 6703 Model Fits
5
5
5
5
5
5
5
5
5
5
0
0
0
0
0
0
0
0
0
0
0
5
10
15
-χ2
Figure 10. χ2
σ-diagram from fitting dynamical models to
h4
the kinematic data for NGC 6703. The crosses show fit results for
a variety of anisotropic, self -- consistent models of a Jaffe sphere,
whose density profile is a good approximation for the luminous
matter in NGC 6703. They fall upwards and to the right of a
curved envelope that separates them clearly from a perfect fit,
showing that no self -- consistent model can simultaneously fit both
the dispersion and line profile shape data. The heavy dot is the
best -- estimate self -- consistent model in the stars -- only potential,
obtained with the method of Section 4. The squares show a num-
ber of dynamical models with a dark halo; these are consistent
with the data.
are of no consequence for the halo of NGC 6703. Moreover,
systematic effects may play a role for the innermost data
points (Sect. 2).
h4 -χ2
The velocity scale of each model is matched to opti-
mize the fit to the σ-profile; that to the h4-profile is already
fixed with no extra assumption. The resulting values of χ2
h4 ,
χ2
σ are normalized by the number of fitted data points. In
Fig. 10 we plot a χ2
σ-diagram for all self-consistent mo-
dels from Jeske (1995). The χ2 values are again normalized
to the number of data points. Also plotted on the figure are
best-estimate models in a number of potentials, constructed
with the technique described in Section 4. The optimized
self-consistent model (the heavy dot) lies on the bound-
ing envelope of the self-consistent models; it has χ2
σ = 2.6,
χ2
h4 = 1.3. The squares show the normalized χ2-values for
several models with a dark halo whose rotation curves are
shown by the full lines in Fig. 9. For most of these χ2
σ ≃ 1.3,
χ2
h4 ≃ 0.8.
Fig. 10 shows that no self-consistent model will fit the
data: all self-consistent models are clearly separated by a
curved envelope from the lower left hand corner of the dia-
gram, which corresponds to a perfect fit. Either the velocity
dispersion profile is matched reasonably well, but then the
line profiles cannot be reproduced, or, when the h4-profile
is fitted accurately, the dispersion profile is poorly matched.
The cure for the discrepancy is to raise both σ and h4 at
large radii. Thus, according to Sect. 3 above, we require ex-
tra mass at large r. NGC 6703 must have a dark halo.
5.2 Dynamical models with dark halo
We will now derive constraints on the gravitational poten-
tial of NGC 6703 within the framework of the parametric
mass model of eqs. (2) -- (4). In doing this we have in mind
Figure 11. Rotation curves for a sequence of gravitational po-
tentials (stars plus dark halo) used in the analysis of NGC 6703.
The full lines show rotation curves that are consistent with the
NGC 6703 kinematic data inside the 95% confidence boundary at
λ = 10−5 (open symbols in Fig. 12). The other line styles show
rotation curves inconsistent with the data; among these is the
constant -- M/L model with no dark halo (short -- dashed).
the following working picture: The stellar component is as-
signed a constant mass-to-light ratio Υ, chosen maximally
such that the stars contribute as much mass in the center
as is consistent with the kinematic data. The model for the
halo incorporates a constant density core, and its parame-
ters are chosen such the halo adds mass mainly in the outer
parts of the galaxy if that is necessary. This is similar to
the maximum disk hypothesis in the analysis of disk galaxy
rotation curves. Within this framework we can determine
the maximum stellar mass-to-light ratio, ask whether it is
reasonable, and constrain the halo parameters.
Because determining the two potential parameters rc
and v0 together with the model velocity scale (equivalent
to the stellar mass -- to -- light ratio) is a three -- dimensional
problem, we will proceed in steps. Fig. 11 shows circular
rotation curves for the first sequence of mass models for
NGC 6703 that we have investigated. In all these models the
< Re.
halo contribution becomes significant only outside 20′′
∼
The corresponding halo core radii mostly lie between 1.2rJ
and 1.7rJ , i.e., 55′′ and 80′′. These values are relatively large
because of the falling dispersion curve in NGC 6703. This
implies that we can reliably determine only one of the halo
parameters for this galaxy.
We have chosen the circular velocity vc(Rm) at the ra-
dius Rm of the last kinematic data point as this parameter.
The sequence in Fig. 11 was constructed such as to vary
vc(Rm) and the rotation curve outside ∼ Re, while leaving
the central rotation curve nearly invariant. However, when
the model velocity scales are optimized in the determina-
tion of the best -- estimate dfs, the optimal velocity scale
is found to correlate inversely with vc(Rm). The rotation
curves in Fig. 11 are plotted with their optimal scaling; then
the central rotation curves are no longer identical, but be-
come scaled versions of each other.
In each of the corresponding potentials, we have con-
structed the best -- estimate df with optimal velocity scaling,
as described in Section 4. We first fit a composite df to the
velocity dispersion and line profile shape data for a series of
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
The dark halo of the E0 galaxy NGC 6703
13
Figure 13. The (v0, rc) halo parameter plane. Values of rc are
scaled with respect to rJ = 46′′.5. The luminous plus dark matter
models investigated are shown as plus signs. The contours show
lines of constant χ2
σ+h4 obtained by interpolating between the
model values. Acceptable potentials lie in a band extending from
low v0 and low rc to high v0 and high rc. Models at the upper
left are ruled out because they do not contain enough mass at
large radii. Models at the lower right are ruled out because no
satisfactory fit can be found for any constant value of the stellar
mass -- to -- light ratio.
The fitting results in these potentials are shown by the iso-
lated filled symbols in Fig. 12. Combining with the previous
results these allow us to investigate the range of accept-
able potentials for NGC 6703 more fully than with the one-
dimensional sequence in Fig. 11.
Fig. 13 shows contours of constant χ2
σ+h4 in the (v0, rc) --
plane. The most probable potentials lie in a band extending
from low v0 and low rc to high v0 and high rc. Thus, as al-
ready discussed above, it is not possible to determine both
halo parameters in the NGC 6703 case. However, potentials
in the band of most probable (v0, rc) all have circular veloc-
ities vc(Rm) in the same range of 250 ± 40 km s−1 as before.
In fact, the best -- fitting velocity scales of all models in
Fig. 13 turn out such that the resulting values of vc(Rm) are
in the range [189, 318] km s−1. The fitting procedure tends
to move vc(Rm) into the correct range even when no satis-
factory df can be found. E.g., some models in the lower
right of Fig. 13 with χ2
σ+h4 ≃ 1.5 appear in Fig. 12 at
> 2.4 (not
vc(Rm) = 250 − 300 km s−1, some with χ2
shown) at vc(Rm) = 280 − 340 km s−1. All models shown in
the upper left of Fig. 13 fall near the line defined by the se-
quence of models discussed at the beginning of this section.
The fact that vc(Rm) varies relatively little for a wide range
of luminous matter plus dark halo models suggests that our
results, when expressed in terms of this parameter, are not
sensitive to the choice of halo model in eqs. (3), (4).
σ+h4 ∼
From Fig. 12 we conclude that the true circular ve-
locity of NGC 6703 at 78" is vc(Rm) = 250 ± 40 km s−1
(the formal 95% confidence interval obtained from the filled
symbols, according to Fig. 7). In Section 4.3 we found,
however, that most of this indeterminacy is towards large
circular velocities, at least for a galaxy with a dispersion
curve like that of NGC 6703. By contrast, in potentials with
lower values of vc(Rm) than the true circular velocity, it
quickly becomes impossible to find a satisfactory df. Based
Figure 12. Quality with which the kinematics of NGC 6703
can be fitted in different potentials. The figure shows the av-
erage χ2 per σ -- and h4 data point, of the best -- estimate distri-
bution function fitted to the velocity dispersion and line profile
data, as a function of the assumed potential's circular rotation
velocity at the observed radius of the last kinematic data point.
Filled symbols show best -- estimate models derived with the opti-
mal λ = 10−4 determined in Section 4.2; open symbols represent
models derived with λ = 10−5. The self -- consistent (M/L = const.)
and the vc = const. models are marked separately. The horizon-
tal dashed line shows the 95% confidence boundary derived from
Fig. 7.
values of the unknown velocity scale. From this sequence of
models, we determine the optimal value of the model's ve-
locity scale in this potential, and then recompute the best --
estimate model with this velocity scale. In the following,
when speaking of the best -- estimate df for a given poten-
tial, we will always imply that the velocity scale has been
optimized in this way. In the fitting procedure we have used
a regularization parameter λ = 10−4; this was found appro-
priate in Section 4.2 for the error bars and sampling of the
NGC 6703 data. Compared with the tests in Section 4, we
have included a few extra basis functions (total K = 20) to
resolve the (possibly not real, cf. Section 2) high -- frequency
structure in the center of NGC 6703.
Fig. 12 shows, as a function of vc(Rm), the average χ2
per σ -- and h4 data point of the respective dfs so obtained.
The connected solid symbols represent the sequence of po-
tentials corresponding to Fig. 11. The potential with con-
stant mass -- to -- light ratio appears in the upper left -- hand
corner in Fig. 12; it is inconsistent with the data by a large
margin even for optimum velocity scale (see Fig. 7). The
best -- fitting potential with a completely flat rotation curve
has an optimal value of vc = vc(Rm) = 254 km s−1 and
χ2
σ+h4 = 1.17. Thus it does not provide the best possible fit
but is consistent with the data. Of the stars plus dark halo
models illustrated in Fig. 11, those models in the sequence
with vc(Rm) = 210 − 285 km s−1 have χ2
σ+h4 < 1.28. This
is consistent with the results of Section 4.3, from which we
would expect that the NGC 6703 data can be fit by a range
of gravitational potentials. Models in the sequence outside
this range of vc(Rm) are inconsistent with the kinematic
data at the 95% confidence level (cf. Fig. 7).
In a second step we have analyzed a more complete set
of potentials in a suitable part of the (v0, rc) -- plane (Fig. 13).
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
14 O.E. Gerhard, G. Jeske, R.P. Saglia, R. Bender
on these results, the lower values in the quoted range of
vc(Rm) = 250 ± 40 km s−1 appear to be the more probable
ones.
The open symbols in Fig. 12 show that the range of
potentials consistent with the data is enlarged only slightly
when the df is allowed to be less smooth. These points are
from best -- estimate models derived in the sequence of po-
tentials of Fig. 11 with λ = 10−5 instead of the optimal
λ = 10−4. The resulting curve which connects the λ = 10−5
models in Fig. 12 surrounds the corresponding λ = 10−4
curve. Generally it appeared that, in the models obtained
with λ = 10−5, the df came close to zero more easily and
more often. The last two facts, when taken together, sug-
gest that, in addition to the data themselves, the positivity
constraints on the df play an important role in determining
the boundary of the region in (f, Φ) space that is consistent
with given kinematic data.
Fig. 14 shows best -- estimate models in some of the
potentials consistent with the kinematic data (at λ =
10−4). The three full
lines are models with vc(Rm) =
(218, 231, 242) km s−1, the two dotted lines are models with
vc(Rm) = (253, 277) km s−1. Also shown is the best --
estimate model in the potential of only the stars with con-
stant M/L (long -- dashed), and that in the best potential
with vc = const. = 254 km s−1 (short -- dashed).
the best -- estimate
The dispersion profile of
self-
consistent model falls clearly below the data at both inter-
mediate and large radii, as does its h4-profile. From the dis-
cussion in Section 3, this is a clear sign of extra mass at large
radii. The models including dark halo contributions mainly
differ in the outermost parts of the velocity dispersion pro-
file. As expected, those with the highest velocity dispersions
at large radii correspond to the potentials with the largest
asymptotic circular velocities. This again suggests that with
smaller error bars at large radii and with spatially more ex-
tended data, we will be able to significantly narrow down the
uncertainties in the halo parameters. The model with every-
where constant rotation speed is constrained tightly by the
kinematic data in the central parts, where the ρ ∝∼ r−2 pro-
file is presumably dominated by the stars. It then has some
difficulties both with the h4 -- values at intermediate radii and
the velocity dispersions at large radii.
The lower part of Fig. 14 shows that, in order to match
the observed kinematics in a potential with large circular ve-
locity, the df must become rapidly tangentially anisotropic
at the radii of the last data points and beyond. As discussed
in Section 4, this is different from the better known effect of
increasing the projected dispersions in a potential with not
enough mass at large radii, by making the df more tangen-
tially anisotropic. Here, without the tangential anisotropy,
our models would predict too high values of the velocity
dispersion, because too much mass at large radii is implied.
The tangential anisotropy at radii outside those reached by
the observations reduces the number of orbits that come into
the observed range, orbits which would contribute relatively
large line-of-sight velocities near their turning points. In this
way the velocity dispersions can be reduced down to the ob-
served values. Clearly, more spatially extended data would
reduce this freedom.
From Fig. 14 we conclude that the stellar distribution
function in NGC 6703 is near-isotropic at the centre and
then changes to slightly radially anisotropic at intermedi-
200
160
120
0.16
0.12
0.08
0.04
0
-0.04
-0.08
-0.12
-0.16
0.6
0.4
0.2
0
-0.2
0 10 20 30 40 50 60 70 80 90 100
R [arcsec]
0 10 20 30 40 50 60 70 80 90 100
r [arcsec]
Figure 14. Dynamical models for the kinematics of NGC 6703
in several luminous plus dark matter potentials, compared to
projected velocity dispersion (top panel) and vp-shape param-
eter h4 (middle panel). The bottom panel shows the models' in-
trinsic anisotropy parameter β(r), with the same linestyles: self --
consistent model (stars only; long -- dashed), vc = const. model
(short -- dashed), three models with vc(78′′) in the lower part of
the acceptable range (full), and two models with vc(78′′) in the
upper part of this range (dotted lines).
ate radii (β = 0.3 − 0.4 at 30", β = 0.2 − 0.4 at 60"). It is
not well-constrained near the outer edge of the data, where
formally β =−0.5 − +0.4, depending on the correct poten-
tial in the allowed range. However, the models with large
asymptotic halo circular velocities shown in Fig. 14 appear
less plausible, because they are the models with the most
rapidly increasing velocity dispersions outside R ≃ 60′′. The
same models also show the most rapid increase in tangen-
tial anisotropy at and beyond Rm = 78′′, which again ap-
pears a priori implausible, because it implies rapid changes
in the df just outside the observed range. The combined
signature of both effects is strongly reminiscent of Fig. 9,
where it was clearly an artifact of the limited radial range
of the data. If this assumption is correct, one would again
conclude that lower -- vc(Rm) models in the formal range are
favoured. Fig. 15 shows the recovered df for the potential
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
The dark halo of the E0 galaxy NGC 6703
15
Figure 16. Luminous, dark, and total mass as a function of ra-
dius for the range of acceptable models of NGC 6703, according
to Fig. 12 (short dashed, dotted, and dash -- dashed or full lines, re-
spectively). Mass distributions in which a df with χ2
σ+h4 ≤ 1.12
(including 87% of the distribution in Fig. 7, 1.5σ) are coded by full
lines, those with χ2
σ+h4 ≤ 1.28 (including 95%, 2.0σ) by dash --
dashed lines. The vertical line denotes the position of the last
kinematic data point. At this radius, the luminous mass fails by
at least a factor of 1.6.
Figure 17. The B-band mass -- to -- light ratio of NGC 6703. The
solid and dash -- dashed lines (coding as in Fig. 16) are derived from
the dynamical models that span the range of acceptable vc(Rm)
in Fig. 12. The central mass -- to -- light ratio is Υ = 3.3, that at the
position of the last kinematic data point at 78" (vertical bar) is
in the range of Υ = 5.3 − 10.
5.3 Uncertainties
There are a number of possible sources of systematic error
which would affect the mass -- to -- light ratio derived for NGC
6703. Most of the errors so introduced are probably small
compared to the considerable uncertainty arising from the
kinematic measurement errors and limited radial sampling,
discussed above. One systematic error on the absolute mass-
to-light ratios in NGC 6703 comes from the uncertainty in
the distance, although this does not change the ratio of outer
to central values. A further systematic effect on this ratio
can be introduced by the sky brightness level: If this is in-
creased by 2− 3%, the fitted rJ decreases to 35". To the ex-
tent that the outermost kinematic data point at 78" (which
then moves to larger r/rJ ) is in the flat part of the circu-
Figure 15. Distribution function in the energy -- circularity plane
derived for NGC 6703, in the luminous plus dark matter potential
with vc(Rm) = 231 km s−1 shown as the middle solid line in the
upper panel of Fig. 14. Energy is specified by the value of r(E),
where E = Φ(r), in units of the fitted Jaffe radius rJ , and the
df is given in arbitrary logarithmic units. The last measured
kinematic data point is located at lg r(E)/rJ = 0.23.
with vc(Rm) = 231 km s−1 in Fig. 14. Note, however, that
all models shown in Fig. 14 except the self -- consistent one
are formally consistent with the presently available data.
50 M⊙,
(17)
From the constraints on the circular velocity vc(Rm) =
50 kpc
210−290 km s−1 the range in mass inside Rm = 13.5 h−1
is
M (< Rm) = 1.6 − 2.6 × 1011h−1
where h50 ≡ H0/50km/s/Mpc. The total mass in stars in-
side this radius is 8− 9× 1010 M⊙, assuming constant mass-
to-light ratio Υ and a maximum stellar mass model, and
taking an average value from the models consistent with the
kinematic data. The radial run of the luminous, dark, and
total mass is shown in Fig. 16 for the models that span the
allowed range according to Fig. 12. After dividing by the lu-
minosity LB(r) for the stars, the mass-to-light ratios shown
in Fig. 17 result. Between the centre and the last data point
r = 78′′ ≃ 2.6Re, the mass-to-light ratio of NGC 6703 rises
by a factor of 1.6 − 3.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
16 O.E. Gerhard, G. Jeske, R.P. Saglia, R. Bender
lar rotation curve, the inferred M/L changes only to second
order because the luminosity inside 78" remains essentially
unchanged. The same is true if the sky value is decreased by
1%, in which case the fitted rJ increases to 54′′.
In the previous analysis, we have ignored a possible
small rotation in the outer parts of NGC 6703 (perhaps
> 50′′, but the errors are large; cf.
∼ 20 − 30 km s−1 at R ∼
Fig. 1). The simplest possible estimate of the effect of this
rotation on the derived masses is to replace σ ≃ 105 km s−1
in this radial range by (σ2 + v2)1/2 ≃ 110 km s−1. This gives
a factor of 1.1, neglecting changes in the model structure
that would result because the central kinematics remain un-
changed.
Next we consider the possibility that NGC 6703 may
contain a face-on extended disk (see de Vaucouleurs, de Vau-
couleurs and Corwin 1976). From an R1/4-law plus disk de-
composition, we estimate that the contribution of such a
disk in the region where we model the kinematics could be
up to 10-20%. In this case we expect the velocity dispersion
to be decreased and the h4 coefficient to become more pos-
itive where the disk contributes significant light (Dehnen &
Gerhard 1994; see NGC 4660 as an example in BSG), most
likely in the outer parts.
Similarly, it is conceivable that NGC 6703 is in reality
slightly triaxial and is seen from a special direction so as to
appear E0. The likelihood of this is the smaller, the more
triaxial the intrinsic axial ratios are; thus slightly triaxial
shapes are the most plausible ones. Again this will imply
some extra loop orbits seen nearly face-on, similarly increa-
sing h4 and decreasing σ.
In both cases, we therefore expect the spherical com-
ponent in NGC 6703 to have lower h4 and larger σ than
the measured values. A similar analysis of such kinematics
would, according to the discussion in Section 3, lead to a mo-
del with greater tangential anisotropy at large radii, with
the mass distribution less affected. Recall that decreasing
the mass at large r in a spherical model lowers both σ and
h4.
6 DISCUSSION AND CONCLUSIONS
This study is part of an observational and theoretical pro-
gram aimed at understanding the mass distribution and or-
bital structure in elliptical galaxies. In the following we first
discuss general results on potential and anisotropy determi-
nation, and then proceed to the specific case of NGC 6703.
6.1 Velocity profiles and anisotropy and mass
The analysis of the vps of simple dynamical models in Sec-
tion 3 has broadly confirmed the conclusions of G93. At
large radii, where the luminosity profile falls rapidly, the vps
are dominated by the stars at tangent point. Then radially
(tangentially) anisotropic dfs can be recognized by more
peaked (more flat-topped) vps with more positive (nega-
tive) h4 than for the isotropic case. Increasing β at constant
potential thus lowers σ and increases h4. On the other hand,
increasing the mass of the system at large r at constant
anisotropy, increases both the projected dispersion and h4.
This suggests that by modelling σ and h4 both mass M (r)
and anisotropy β(r) can be constrained.
In practical applications, such an analysis is compli-
cated by a number of factors. Radial orbits at large radii
may lead to increased central velocity dispersions and flat --
topped central vps (already pointed out by Dejonghe 1987).
The former effect can be compensated by a decrease in the
stellar mass -- to -- light ratio. The latter is independent of this,
but can be compensated by changes to the distribution func-
tion in the inner parts of the galaxy (as in a number of cases
studied in Sections 4, 5). A more serious uncertainty is in-
troduced by the possibility of significant gradients in the
orbit population across the radii of interest. For example, a
population of high energy radial orbits with pericentres in a
limited radial range may mimic tangential anisotropy there.
In many cases it will be possible to exclude such a population
of orbits by its effects on the vps at exterior radii, i.e., by si-
multaneously analysing a number of observed vps. Yet this
is least possible precisely at the largest observed radii, where
mass determination is most interesting. Thus this chain of
argument suggests (correctly, see below) that the largest un-
certainty in determining masses and anisotropies in ellipti-
cals from vp -- data is the finite radial extent of these data.
To analyze realistic data we have constructed an algo-
rithm by which the distribution function and potential of
a spherical galaxy can be constrained directly from its ob-
served σ and h4 -- profiles. To assess the significance of the
results obtained, we have tested the algorithm on Monte
Carlo -- generated data sets tuned to the spatial extent, sam-
pling, and observational errors as measured for NGC 6703.
From such data, the present version of the algorithm recov-
ers a smooth spherical df, ∼ 70% of the time, to an rms
level of better than ∼ 12% inside three times the radius of
the outermost kinematic data point.
We have used this algorithm to study quantitatively
the degree to which the gravitational potential can be de-
termined from such data. Our main conclusion is that veloc-
ity profile data with presently achievable error bars already
constrain the gravitational potentials of elliptical galaxies
significantly. In particular, constant -- M/L models are rela-
tively easy to rule out once the data extend beyond 2Re.
The examples that we have studied in detail, tuned to the
NGC 6703 data, certainly belong to the less favourable cases,
because in this galaxy the dispersion profile is falling.
A good way to parametrize the results is in terms of the
true circular velocity vc(Rm) at the radius of the outermost
data point, Rm. With presently available data, vc(Rm) can
< 50 km s−1. This
be determined to a precision of about ± ∼
will improve when high -- quality data at several Re become
available, of the kind expected from the new class of 10m
telescopes. Apart from the fact that smaller error bars will
decrease the formally allowed range in vc(Rm), tests show
that this range often includes high -- vc(Rm) models which be-
come rapidly tangentially anisotropic just outside the data
boundary. These (not very plausible) models can be elimi-
nated with data extending to larger radii.
On the other hand, the detailed form of the true circu-
lar velocity curve is much harder to determine than vc(Rm).
Conspiracies in the df are possible that minimize the mea-
surable changes in the line profile parameters. Our tests
showed that two potentials differing by just the value of the
halo core radius could not be distinguished even with very
good data out to 6Re. Thus some uncertainty will remain in
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
practise, regardless of whether or not in theory the potential
is uniquely determined from the projected df N (r, vk).
A similar picture holds for the related determination
of the anisotropy of the df. For the present error bars in
the data, β(r) is relatively well -- determined out to about
half the limiting radius of the observations. Near the edge of
the data, uncertainties can be large depending on the grav-
itational potential (recall that in a fixed spherical potential,
the df is uniquely determined by the complete projected df
N (r, vk)). Again the unknown nature of the orbits beyond
the last data point has a large part in this uncertainty.
Because the largest uncertainties in determining masses
and anisotropies from vps occur near the outer radial limit
of these data, the combination of the type of analysis pre-
sented here with other information (e.g., from X-ray data)
will be particularly powerful.
6.2 The dark halo of NGC 6703
Fig. 10 shows that no self-consistent model will fit the
kinematic data for NGC 6703. Our non -- parametric best --
estimate self -- consistent model is inconsistent with the data
at the > 99% -- level (Figs. 12, 7). With self-consistent mo-
dels, either the velocity dispersion profile is matched reason-
ably well, but then the line profiles cannot be reproduced,
or, when the h4-profile is fitted accurately, the dispersion
profile is poorly matched. The cure for the discrepancy is to
raise both σ and h4 at large radii. Thus, as discussed above,
we require extra mass at large r. NGC 6703 must have a
dark halo.
We have next derived constraints on the parameters of
this halo as follows. The luminous component is assigned a
constant mass-to-light ratio Υ, chosen maximally such that
the stars contribute as much mass in the center as is consis-
tent with the kinematic data. Our parametric model for the
halo incorporates a constant density core, and its param-
eters (core radius rc and asymptotically constant circular
velocity v0) are chosen such the halo adds mass mainly in
the outer parts of the galaxy if that is necessary. We call
these models maximum stellar mass models (analogous to
the maximum disk hypothesis in the analysis of disk galaxy
rotation curves).
50 kpc is 9 × 1010h−1
We find that maximum stellar mass models in which
the luminous component provides nearly all the mass in the
centre fit the data well. In these models, the total luminous
mass inside the limiting observational radius Rm = 78′′ =
13.5 h−1
50 M⊙, corresponding to a central
B -- band mass -- to -- light ratio Υ = 3.3h50 M⊙/ L⊙. According
to Worthey's (1994) models, this is a rather low value for the
stellar population of an elliptical galaxy and would point to a
relatively low age (5 Gyrs) and/or low metallicity (less than
solar). However, the galaxy has a color (B − V )0 = 0.93 and
a central line index M g2 = 0.280 (Faber et al. 1989) which
are typical for ellipticals of similar velocity dispersion.
A larger value of H0 could increase the M/L value and
alleviate the demands on the stellar populations. However,
the distance used here (36 Mpc) includes a correction for
the large inferred peculiar velocity of the galaxy. If we had
used a distance based on the larger radial velocity in the
CMB frame, our derived M/L would be even lower. It is
also implausible that the low central value of M/L stems
from the contribution of a young stellar population in a disk
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
The dark halo of the E0 galaxy NGC 6703
17
component, which we estimate cannot be larger than 20% of
the total light (see above). Thus we conclude that the dark
halo in NGC 6703 is unlikely to have higher central densi-
ties than inferred from our maximum stellar mass models,
because otherwise the M/L of the stellar component would
be reduced to implausibly small values.
In a recent preprint, Rix et al. (1997) have analyzed
the velocity profiles of the E0 galaxy NGC 2434 with a
linear orbit superposition method. This galaxy provides an
interesting contrast to NGC 6703 because it has an essen-
tially flat dispersion profile. Its kinematics are likewise in-
consistent with a constant -- M/L potential, but are well -- fit
by a model with vc = const. This can be interpreted as a
maximum stellar mass model in the sense defined above, in
which the luminous component with maximal Υ contributes
most of the mass inside Re. The kinematics of NGC 2434
are also well -- fit by a range of specific, cosmologically moti-
vated mass models which, if applicable, would imply lower
Υ and significant dark mass inside Re. In NGC 6703, a mo-
del with vc = const is formally consistent with the present
data (within 2σ), but it is not a very plausible fit at large R
and requires large anisotropy gradients between 40" and 70".
It will be interesting to see whether future studies confirm
differences between the shapes of the true circular velocity
curves of elliptical galaxies.
Because of the falling dispersion curve in NGC 6703,
we can determine only one of the halo parameters (The halo
parameters of the most probable potentials lie in a band
extending from low v0 and low rc to high v0 and high rc.)
However, the circular velocity vc(Rm) at the data boundary
is relatively well -- determined for all these models. Thus we
find (Fig. 12) that the true circular velocity of NGC 6703 at
78" is vc(Rm) = 250± 40 km s−1 (formal 95% confidence in-
terval). Tests on pseudo data have shown that this range of-
ten includes high -- vc(Rm) models which become rapidly tan-
gentially anisotropic just outside the data boundary. Such
models may not be very plausible, so the lower values in the
quoted range of vc(Rm) = 250± 40 km s−1 may be the more
probable ones.
50 kpc the total mass en-
closed is M (< Rm) = 1.6 − 2.6 × 1011h−1
50 M⊙, and the inte-
grated mass -- to -- light ratio out to this radius is Υ = 5.3− 10,
corresponding to a rise from the center by at least a factor
of 1.6. We have already noted that NGC 6703 is an un-
favourable case because of its falling dispersion curve. The
fact that relatively small variations in Υ can nonetheless
be detected shows the power of the method. Note that a
scheme based on the analysis of the line of sight velocity
dispersions alone (Binney, Davies, Illingworth 1990, van der
Marel 1991) would conclude that constant mass -- to -- light ra-
tio models can provide good fits.
Thus, at Rm = 78′′ = 13.5 h−1
The stellar distribution function in NGC 6703 is near-
isotropic at the centre and then changes to slightly radi-
ally anisotropic at intermediate radii (β = 0.3 − 0.4 at 30",
β = 0.2−0.4 at 60"). It is not well-constrained near the outer
edge of the data, where formally β =−0.5− +0.4, depending
on the correct potential in the allowed range. Models near
the lower end of this range may be consistent with the data
only because of the limited radial extent of the measure-
ments.
18 O.E. Gerhard, G. Jeske, R.P. Saglia, R. Bender
6.3 Conclusions
In summary, we have shown that the mass distribution M (r)
and anisotropy structure β(r) for spherical galaxies can both
be constrained from vp and velocity dispersion measure-
ments. NGC 6703 must have a dark halo, contributing about
equal mass at 2.6Re as do the stars. The circular velocity at
the last kinematic data point (78") is constrained to lie in the
range 250 ± 40 km s−1 at 95% confidence. The anisotropy of
the stellar orbits changes from near-isotropic at the center to
slightly radially anisotropic at intermediate radii, and may
be either radially or tangentially anisotropic at 78". With
more extended and more accurate data it will be possible to
considerably narrow down these uncertainties.
If the results for this galaxy are typical, they suggest
that also in elliptical galaxies the stellar mass dominates at
small radii, and the dark matter begins to dominate at radii
around 10 kpc. It is important to obtain extended kinematic
data and do a similar analysis for a number of elliptical ga-
laxies. When we know the systematics and the spread in the
circular velocity curves and anisotropy profiles for a sam-
ple of ellipticals, we will have an important new means for
testing the currently popular formation theories.
ACKNOWLEDGMENTS
We thank U. Hopp for providing us with a CCD frame of
NGC 6703, and David Merritt for helpful discussions on reg-
ularization methods. We thank the referee for his rapid and
constructive comments, especially on the revised version,
and the editorial staff for their relentless efforts to secure his
reports. We acknowledge financial support by the Deutsche
Forschungsgemeinschaft under SFB 328 and SFB 375 and by
the Schweizerischer Nationalfonds under grants 21-40464.94
and 20-43218.95. OEG also acknowledges a Heisenberg fel-
lowship while at Heidelberg.
REFERENCES
Awaki H., et al. , 1994, PASJ 46, L65
Arnaboldi M., Freeman K.C., Hui X., Capaccioli M., Ford H.,
1994, Messenger 76, 40
Bender R., 1990, A&A 229, 441
Bender R., Saglia, R., Gerhard, O.E., 1994, MNRAS 269, 785
(BSG)
Binney J.J., 1978, MNRAS 183, 501
Binney, J.J., Davies, R.L., Illingworth, G.D., 1990, ApJ 361, 78
Binney J.J., Mamon G.A., 1982, MNRAS 200, 361
Carollo C.M., de Zeeuw P.T., van der Marel R.P., Danziger I.J.,
Qian E.E., 1995, ApJL 441, L25
Dehnen W., Gerhard O.E., 1994, MNRAS 268, 1019
Dejonghe H., 1987, MNRAS 224, 13
Dejonghe H., de Bruyne V., Vauterin P., Zeilinger W.W., 1996,
A&A 306, 363
Dejonghe H., Merritt D., 1992, ApJ 391, 531
de Vaucouleurs G., de Vaucouleurs A., Corwin H.G., 1976, Second
Reference Catalogue of Bright Galaxies, Univ. Texas Press,
Austin
Faber S.M., Wegner G., Burstein D., Davies R.L., Dressler A.,
Lynden-Bell D., Terlevich R.J., 1989, ApJS 69, 763
Franx M., van Gorkom J.H., de Zeeuw T., 1994, ApJ 436, 642
Fricke W., 1952, Astr. Nachr. 280, 193
Gerhard O.E., 1991, MNRAS 250, 812 (G91)
Gerhard O.E., 1993, MNRAS 265, 213 (G93)
Grillmair C.J., Freeman K.C., Bicknell G.V., Carter D., Couch
W.J., Sommer-Larsen J., Taylor K., 1994, ApJ 422, L9
Hanson R.J., Haskell K.H., 1981, Math. Programm. 21, 98.
Jaffe W., 1983, MNRAS 202, 995
Jeske, G., 1995, PhD Thesis, University of Heidelberg
Jeske, G., Gerhard, O.E., Bender, R., Saglia, R.P., 1996, Proc.
IAU 171, eds. Bender, R., Davies, R.L., p. 397
Kim D.-W., Fabbiano G., 1995, ApJ 441, 182
Kochanek C.S., Keeton C.R., 1997, in The Nature of Elliptical
Galaxies, 2nd Stromlo Symposium, eds. Arnaboldi M., Da
Costa G.S., Saha P., 1997, ASP 116, 21.
Maoz D., Rix H.-W., 1993, ApJ 416, 435
Merritt D., 1985, AJ 90, 1027
Merritt D., 1993, ApJ 413, 79
Osipkov L.P., 1979, Pis'ma Astr. Zh. 5, 77
Press, W.H., Flannery, B.P., Teukolsky, S.A., Vetterling, W.T.,
Numerical Recipes (Cambridge: Cambridge University Press)
Rix H.-W., de Zeeuw P.T., Carollo C.M., Cretton N., van der
Marel R.P., 1997, preprint
Saglia R.P. et al. , 1993, ApJ 403, 567
Saglia, R.P., Bender, R., Gerhard, O.E., Jeske, G., 1997a, in Dark
and Visible Matter in Galaxies and Cosmological Implica-
tions, eds. Persic M., Salucci P., ASP 117, 113.
Saglia R.P. et al. , 1997b, ApJS 109, 79
van der Marel, R., 1991, MNRAS 253, 710
van der Marel R.P., Franx M., 1993, ApJ 407, 525
Worthey, G. 1994, ApJS, 95, 107
APPENDIX A: LIBRARY OF ANISOTROPIC
SPHERICAL DISTRIBUTION FUNCTIONS
To understand the connection between anisotropy struc-
ture and observable line profile shapes we have constructed
a number of spherical distribution functions of the quasi --
separable form (Gerhard 1991, G91)
f (E, L) = g(E) h(x),
(18)
where the variable x depends on both energy and angular
momentum:
L
(19)
x =
L0 + Lc(E)
.
L0 is an angular momentum constant, or equivalently, an
anisotropy radius times a characteristic velocity; Lc(E) is
the angular momentum of the circular orbit at energy E. dfs
of the form (18) have the following properties: (i) the circu-
larity function h(x) has the effect of shifting stars between
orbits of different angular momenta on surfaces of constant
energy, while g(E) controls the distribution of stars between
energy surfaces. (ii) For the most bound stars L < Lc(E) ≪
L0; thus the model becomes isotropic in the centre unless
L0 = 0. (iii) For loosely bound stars L ∼ Lc(E) ≫ L0,
i.e., the angular momentum distribution becomes a func-
tion of circularity L/Lc(E) which is one-to-one related to
orbital eccentricity. Outside the anisotropy radius, the df
(18) therefore corresponds to an energy-independent orbit
distribution with constant anisotropy, radial or tangential.
In these models h(x) is an assigned function built into
the model to achieve the desired anisotropy (orbit distribu-
tion). Radially biased distribution functions correspond to
circularity functions h(x) decreasing with x; for example
h(x) = hα(x) ≡(cid:0)1 − x2(cid:1)α
.
(20)
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
The dark halo of the E0 galaxy NGC 6703
19
µ(E) =
1
2
+
1
π
arctan(cid:20) k
2
2
E2 − E
E2
(cid:21).
(23)
The parameters E and k determine the orbital energy near
which the anisotropy transition occurs, and the width of the
transition. A similar function µ(E) was used for models in
halo potentials. Figure 18 shows the intrinsic and projected
properties of a number of dfs of this kind, constructed in
the self -- consistent potential of a Jaffe sphere. Notice the
wide variety of kinematical profiles that can be constructed
in this way.
APPENDIX B: TRANSFORMING TO LINEAR
KINEMATIC DATA
In velocity line profile measurements, the depth of an ab-
sorption line in a spectral resolution element is assumed to
be proportional to the number of stars with line-of-sight ve-
locities corresponding to this wavelength interval. The line-
of-sight velocity distribution measured from the line profiles
is a discretized function linearly related to the underlying
df [cf. eq. (1)]. This linearity is lost when line profile mea-
surements are represented by the quantities v, σ, h3, h4 --
these quantities are obtained by a least-square fit of a Gauss-
Hermite series to the observed line profile.
To re-express the observed kinematics in terms of quan-
tities that depend linearly on f we proceed as follows. Con-
sistent with the assumption of spherical symmetry, we set
the mean streaming velocity v and all odd velocity profile
moments to zero. Next, we obtain an estimate for the veloc-
ity dispersion (second moment) σ2, by integrating over the
line profile specified by (σ, h4); for negative h4, until it first
becomes zero. For small h4, the linear correction formula
σ = σ(1 + √6h4) holds (van der Marel & Franx 1993), this
correction results in σ > σ for peaked profiles with h4 > 0.
The numerical correction from integrating over the velocity
profile also has this property (BSG).
From the measured h4(Ri), we compute new even
(24)
(25)
hj (Ri) Hj(x) exp(−x2/2)
(h0 = 1, h1 = h2 = h3 = 0) with x ≡ vk/σ(Ri) as
Nn sn Hn(ex) exp(−ex2/2),
Gauss-Hermite moments sn(Ri;eσ) by expanding the series
L(Ri, vk) = L0 Xj=0,4
L(Ri, vk) = Xn=0,2,4,...
where ex ≡ vk/eσ(Ri). Here Hn are Hermite polynomials,
the Nn are normalization constants (G93), and eσ(Ri) are
practice, we have found it convenient to take foreσ(Ri) the
galaxy being analyzed. The sn(Ri;eσ) are estimates for the
sn(Ri;eσ) = (cid:0)2n−1n!(cid:1)−1/2
Gauss-Hermite moments related to the true velocity profile
by
velocity dispersions σiso(Ri) of the isotropic model in the
given potential Φ(r) with the same stellar density as the
fiducial scaling velocities generally different from σ(Ri). In
(26)
×Z ∞
−∞
For a theoretical model, we obtain corresponding mo-
dvk Hn(vk/eσ) exp(−v2
k/2eσ2)L(Ri, vk).
Figure 18. Anisotropy parameter β(r), projected velocity disper-
sion σp, and line profile shape parameter h4 for several families
of anisotropic dfs. Left: scale -- free radially anisotropic (dotted)
and tangentially anisotropic (dashed lines) in the potential of a
self -- consistent Jaffe sphere. For these models, circularity func-
tions of the type (20) and (21) were used with different α and
c, respectively. The isotropic model is shown for reference (solid
line). Right: Families of models with anisotropy changing from
radial to tangential (dotted) or from tangential to radial (dashed
lines). These were constructed from the same circularity functions
and a weight factor as in eq. (23).
The family can also be used to construct tangentially aniso-
tropic models, such as
(21)
hc,α(x) = c + (1 − c)(cid:2)1 − (1 − x2)α(cid:3) .
In these tangentially anisotropic models one cannot choose
h(0) = 0 unless L0 = 0, otherwise the density at r = 0 would
be zero. Of course, other forms for h(x) are possible, such
as Gaussians.
Given the assigned function h(x), the integral equation
for ρ(r) in terms of f (E, L2) is solved for the derived func-
tion g(E); see G91 and Jeske (1995). Fig. 3 shows line profile
shape parameters for representative dfs constructed in this
way. Fig. 3 shows the anisotropy profiles of two sets of tan-
gentially anisotropic models. Here the density of stars has
been taken to be that of a Jaffe sphere, and the potential
in which the stars orbit is either the self -- consistent poten-
tial or one of the mixed stars plus halo potentials used in
Sections 4, 5. Sequences like that in Fig. 3 are used as basis
functions in the non -- parametric analysis in Section 5.
Models whose anisotropy changes from radial to tan-
gential or vice versa were constructed by linearly combining
the above circularity functions with energy -- dependent co-
efficients. In this way one obtains dfs of the more general
form
f (E, L) = g(E) h(E, x).
(22)
For self -- consistent Jaffe models we used energy -- dependent
coefficients µ(E) of the following form:
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
20 O.E. Gerhard, G. Jeske, R.P. Saglia, R. Bender
ments by inserting the projected df N (Ri, vk) from (1) into
the new sn-moments of the composite model are linear in
the ak, i.e.,
eq. (26) instead of L(Ri, vk), using the sameeσ(Ri). Clearly,
sn(Ri;eσ) = Xk=1,K
n (Ri;eσ)
with the s(k)
n corresponding to the respective fk.
aks(k)
(27)
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
|
astro-ph/9608098 | 1 | 9608 | 1996-08-15T11:22:04 | Depolarization of the cosmic microwave background by a primordial magnetic field and its effect upon temperature anisotropy | [
"astro-ph"
] | We estimate the depolarizing effect of a primordial magnetic field upon the cosmic microwave background radiation due to differential Faraday rotation across the last scattering surface. The degree of linear polarization of the CMB is significantly reduced at frequencies around and below 30 GHz $(B_* /10^{-2}{\rm Gauss})^{1/2}$, where $B_*$ is the value of the primordial field at recombination. The depolarizing mechanism reduces the damping of anisotropies due to photon diffusion on small angular scales. The $l\approx 1000$ multipoles of the CMB temperature anisotropy correlation function in a standard cold dark matter cosmology increase by up to 7.5\% at frequencies where depolarization is significant. | astro-ph | astro-ph |
Depolarization of the cosmic microwave background
by a primordial magnetic field
and its effect upon temperature anisotropy
Diego D. Harari,a,b,1 Justin D. Haywarda,c,2 and Matias Zaldarriagad,3
a D.A.R.C., Observatoire de Paris - Meudon
5 Place Jules Janssen, 92195 Meudon, France
b Departamento de F´ısica, Facultad de Ciencias Exactas y Naturales
Universidad de Buenos Aires, Ciudad Universitaria - Pab. 1
1428 Buenos Aires, Argentina
c Department of Applied Mathematics and Theoretical Physics
University of Cambridge, Cambridge CB3 9EW, UK
d Department of Physics, MIT, Cambridge, MA 02139 USA
Abstract
We estimate the depolarizing effect of a primordial magnetic field upon
the cosmic microwave background radiation due to differential Faraday rota-
tion across the last scattering surface. The degree of linear polarization of
the CMB is significantly reduced at frequencies around and below 30 GHz
(B∗/10−2Gauss)1/2, where B∗ is the value of the primordial field at recom-
bination. The depolarizing mechanism reduces the damping of anisotropies
due to photon diffusion on small angular scales. The l ≈ 1000 multipoles of
the CMB temperature anisotropy correlation function in a standard cold dark
matter cosmology increase by up to 7.5% at frequencies where depolarization
is significant.
PACS numbers: 98.80.-k, 98.70.Vc, 98.80.Es
1Email address: [email protected]
2Email address: [email protected]
3Email address: [email protected]
1
Introduction
The cosmic microwave background radiation (CMB) is expected to have acquired a
small degree of linear polarization through Thomson scattering [1], which polarizes the
radiation if there is a quadrupole anisotropy in its distribution function [2]. Typically,
the CMB degree of linear polarization is expected to be more than ten times smaller
than the relative temperature anisotropy on comparable angular scales, at least within
a standard ionization history of the Universe. The CMB has not yet been observed
to be polarized, the upper limit on its degree of linear polarization on large angular
scales being P < 6 × 10−5 [3]. When measured, the CMB polarization will provide
a wealth of information about the early Universe, additional to that revealed by the
CMB anisotropy.
The polarization properties of the CMB may prove particularly valuable to either
constrain or detect an hypothetical primordial magnetic field [4, 5]. A cosmological
magnetic field could leave significant imprints upon the CMB polarization through
the effect of Faraday rotation. After traversing a distance L in a direction q within
an homogeneous magnetic field ~B, linearly polarized radiation has its plane of polar-
ization rotated an angle
ϕ =
e3nexe ~B · q
8π2m2c2 λ2L .
(1)
ne is the total number-density of electrons and xe its ionized fraction. λ is the wave-
length of the radiation, m is the electron mass, and c is the speed of light. We work
in Heaviside-Lorentz electromagnetic units (α = e2/4π ≈ 1/137 is the fine structure
constant if we take ¯h = c = 1).
Faraday rotation of synchrotron emission by distant galaxies serves, for instance,
to estimate the value of galactic and extragalactic magnetic fields [6]. Faraday ro-
tation acts also as a depolarizing mechanism. If an extended source emits polarized
radiation, the total outcome may become significantly depolarized by a magnetic
field, after the radiation emanating from points at different depths within the source
experience different amounts of Faraday rotation. This process affects significantly
the radio emission of galaxies and quasars [7].
In this paper we analyze the depolarizing effect exerted by a primordial magnetic
field upon the CMB across the last scattering surface. We consider a Robertson-
Walker universe with scalar, energy-density fluctuations, and assume a standard
thermal history. We make use of an analytic approach [8], based on a recent re-
finement and extension [9] of the tight-coupling approximation [10], that highlights
the physical process responsible for the CMB polarization and its dependence upon
various cosmological parameters, while still yielding reasonably accurate results. The
polarization of the CMB is proportional to the width of the last scattering surface
(LSS), the interval of time during which most of the CMB photons that we observe
today last-scattered off free electrons. A primordial magnetic field could prevent the
1
polarization from growing across the full width of the LSS. We shall see that the effect
is controlled by the dimensionless and time-independent parameter
F ≡
3
2πe
B
ν2 ≈ 0.7(
B∗
10−3Gauss(cid:17)(cid:16)10GHz
ν0
(cid:17)2
.
(2)
The coefficient F represents the average Faraday rotation (in radians) between Thom-
son scatterings [4]. ν0 is the CMB frequency observed today. B∗ = B(t∗) is the
strength of the primordial magnetic field at a redshift z∗ = 1000, around the time
of decoupling of matter and radiation. Current bounds suggest that a magnetic field
pervading cosmological distances, if it exists, should have a present strength below
B0 ≈ 10−9 Gauss [6]. It is conceivable that the large scale magnetic fields observed
in galaxies and clusters have their origin in a primordial field, and several theoretical
speculations exist about its possible origin [11]. A primordial magnetic field is ex-
pected to scale as B(t) = B(t0)a2(t0)/a2(t), where a(t) is the Robertson-Walker scale
factor. Thus, a primordial field with strength B∗ = 10−3 Gauss at recombination
would have a present strength
B0 =
B∗
(1 + z∗)2 ≈ 10−9 Gauss (
B∗
10−3 Gauss
)
.
(3)
A primordial magnetic field may significantly depolarize the CMB right before
its decoupling from matter. The effect is sensitive to the strength of the magnetic
field at recombination, not to its present strength. A value of B∗ somewhat larger
than 10−3 Gauss is not ruled out. Compatibility with big-bang nucleosynthesis, for
instance, places an upper bound that, extrapolated to the time of recombination, is
at most B∗ = 0.1 Gauss [12]. Recent proposals for either detecting or constraining a
primordial field at recombination were suggested in [5, 13]. In [5], Faraday rotation of
the CMB polarization was analyzed in the limit of small rotation angles, concluding
that a measurement of the effect could provide evidence for magnetic fields of order
B∗ ≈ 10−3 Gauss at recombination. In [13] the change in the photon-baryon sound
speed in the presence of a magnetic field of order B∗ = 0.2 Gauss was claimed to
distort the structure of the acoustic peaks in the CMB anisotropy power spectrum at
a level detectable by currently planned CMB experiments.
We shall entertain in our discussions the possibility that the strength of the pri-
mordial magnetic field at recombination be somewhat larger than B∗ = 10−3 Gauss.
We will show that currently planned CMB experiments might be sensitive to the effect
of depolarization upon the temperature anisotropy power spectrum on small angu-
lar scales if B∗ is around or larger than 0.01 Gauss, while experiments at somewhat
lower frequencies would be sensitive to primordial fields of strength around B∗ ≈ 10−3
Gauss.
The impact of depolarization upon anisotropy comes about as a consequence of
the polarization-dependence of Thomson scattering, which feeds back polarization
2
into anisotropy [8, 14, 15]. The dominant effect is a reduction in the exponential
damping due to photon diffusion, which results in an increase of the anisotropy at
those frequencies for which depolarization is significant. We shall perform an analytic
estimate of the effect, based on the tight coupling approximation. In order to make
more quantitative and specific predictions about the impact and potential measura-
bility of the effect of depolarization upon temperature anisotropy, we shall also use
a recently developed numerical code [16] to integrate the Boltzmann equations in
a standard cold dark matter model. We shall see that the temperature anisotropy
correlation function multipoles at l ≈ 1000 increase by up to 7.5% at frequencies
where depolarization is significant. We conclude that a primordial magnetic field of
strength around 10−2 Gauss at recombination is worth of membership in the list of
multiple cosmological parameters that one may attempt to determine through CMB
anisotropy measurements on small angular scales [17].
The paper is structured as follows. In section II, we write down and describe the
radiative transfer equations for the total and polarized photon-distribution function
in the presence of a single Fourier mode of the scalar metric fluctuations. We include
the term describing Faraday rotation by a primordial magnetic field. We solve these
equations in the tight coupling approximation, and find the dependence of the degree
of polarization upon the frequency of the CMB photons in the presence of a primordial
magnetic field. In section III we discuss the effects of the depolarizing mechanism
upon the anisotropy of the CMB on small angular scales, both analytically as well
as numerically. We discuss the possibility that the effect be detected by currently
planned CMB experiments. Section IV is the discussion and conclusion.
2 Depolarization by a magnetic field
2.1 Boltzmann equations
We begin by considering the radiative transfer equations for a single Fourier mode
of the temperature and polarization fluctuations in a Robertson-Walker spatially-flat
Universe with scalar (energy-density) metric fluctuations, described in terms of the
gauge-invariant gravitational potentials Ψ and Φ. We follow the notation and for-
malism of Ref.
[8]. The total temperature fluctuation is denoted by ∆T , while the
fluctuation in the Stokes parameters Q and U are denoted by ∆Q and ∆U respec-
tively. The degree of linear polarization is given by ∆P = (∆2
U )1/2. All three
quantities are expanded in Legendre polynomials as ∆X = Pl(2l + 1)∆XlPl(µ), where
µ = cos θ = ~k · q/~k is the cosine of the angle between the wave vector of a given
Fourier mode ~k, and the direction of photon propagation q. The evolution equations
for the Fourier mode of wave vector ~k of the gauge-invariant temperature and polar-
ization fluctuations [8, 9, 18, 19], including the Faraday rotation effect of a primordial
Q + ∆2
3
magnetic field [5], read
∆T + ikµ(∆T + Ψ) = − Φ − κ[∆T − ∆T0 − µVb +
1
2
P2(µ)SP ]
∆Q + ikµ∆Q = − κ[∆Q −
1
2
(1 − P2(µ))SP ] + 2ωB∆U
∆U + ikµ∆U = − κ∆U − 2ωB∆Q .
We have defined
SP ≡ −∆T2 − ∆Q2 + ∆Q0
(4)
(5)
(6)
(7)
which acts as the effective source term for the polarization. Vb is the bulk velocity of
the baryons, which verifies the continuity equation
Vb = −
a
a
Vb − ikΨ +
κ
R
(3∆T1 − Vb)
.
(8)
An overdot means derivative with respect to the conformal time τ = R dta0/a, with
a(t) the scale factor of the spatially flat Robertson-Walker metric, and a0 = a(t0) its
value at the present time. R ≡ 3ρb/4ργ coincides with the scale factor a(t) normalized
to 3/4 at the time of equal baryon and radiation densities.
κ = xeneσT a/a0 is the
Thomson scattering rate, or differential optical depth, with ne the electron number
density, xe its ionized fraction, and σT the Thomson scattering cross-section. Finally,
ωB is the Faraday rotation rate [5]
dϕ
dτ
=
e3nexe ~B · q
8π2m2ν2
a
a0
ωB ≡
(9)
If there were axial symmetry around ~k and no Faraday rotation, one could always
choose a basis for the Stokes parameters such that U = 0. A magnetic field with
arbitrary orientation breaks the axial symmetry, and Faraday rotation mixes Q and
U.
2.2 Tight coupling approximation
We now solve the equations (4,5,6) in the tight - coupling approximation, which
amounts to an expansion in powers of kτC, where τC ≡ κ−1 is the average conformal
time between collisions.
At times earlier than decoupling, Thomson scattering is very efficient, and the
mean free path of the photons is very short. The lowest order tight-coupling expres-
sion constitutes in that case an excellent approximation. It implies that the photon
distribution function is isotropic in the baryon's rest frame, and hence the polariza-
tion vanishes [8]. To first order in kτC there is a small quadrupole anisotropy, and
4
thus a small polarization. As decoupling of matter and radiation proceeds, the tight-
coupling approximation breaks down. Still, for wavelengths longer than the width
of the last scattering surface, it provides a very accurate approximation to the exact
result.
In the absence of a magnetic field (ωB = 0), the tight - coupling solutions, to first
order in kτC, are such that [8]
∆U = 0 ; ∆Q =
3
4
SP sin2 θ
SP = −
5
2
∆T2 =
4
3
ikτC∆T1 = −
4
3
τC ∆0
,
(10)
(11)
where we defined ∆0 ≡ ∆T0 + Φ. Notice that ∆Q0 = −5∆Q2 = − 5
2SP , while
all multipoles with l ≥ 3 vanish to first order in kτC. All quantities of interest can be
expressed, in the tight-coupling approximation, in terms of ∆0, which in turn verifies
the equation of a forced and damped harmonic oscillator [9]
4 ∆T2 = 1
∆0 + h
R
1 + R
+
16
45
k2τC
(1 + R)i ∆0 +
k2
3(1 + R)
∆0 =
k2
3(1 + R)
[Φ − (1 + R)Ψ]
,
(12)
where we have neglected O(R2) corrections.
Now consider the effect of the magnetic field (ωB 6= 0), assumed spatially homo-
geneous over the scale of a perturbation with wave-vector ~k. Faraday rotation breaks
the axial symmetry around the direction of the wave-vector. The depolarizing effect
of Faraday rotation depends not only upon the angle between the magnetic field and
the direction in which the radiation propagates, but also upon the angle between the
magnetic field and the wavevector ~k. Nevertheless, we shall only be interested in
the stochastic superposition of all Fourier modes of the density fluctuations, with a
Gaussian spectrum that has no privileged direction. Average quantities thus depend
only upon the angle between the line of sight and the direction of the magnetic field,
but not upon the angle between the magnetic field and the wavevector ~k, which is
integrated away. For simplicity of the calculation, when computing the evolution of
perturbations with wave-vector ~k we shall consider a magnetic field with no compo-
nent perpendicular to ~k. This choice also satisfies the condition of axial symmetry
around ~k, under which eqs. (4,5,6) for ∆T , ∆Q and ∆U were derived. We shall later
use the result of this calculation for the stochastic superposition of all Fourier modes
with arbitrary orientation relative to the magnetic field. This simplification will result
at most in an underestimate of the net depolarizing effect, since the case ~B k ~k is
that for which depolarization is less effective, the magnetic field being perpendicular
to the direction in which polarization is maximum.
To first order in kτC the tight-coupling solutions in the presence of a homogeneous
5
magnetic field ~B k ~k are such that:
∆U = −F cos θ∆Q ; ∆Q =
3
4
SP sin2 θ
(1 + F 2 cos2 θ)
where we have defined the coefficient F as
F cos θ ≡ 2ωBτC
and so
F =
e3
4π2m2σT
B
ν2 ≈ 0.7(cid:16)
B∗
10−3Gauss(cid:17)(cid:16)10 GHz
ν0
(cid:17)2
.
(13)
(14)
(15)
The coefficient F represents the average Faraday rotation between collisions, since
2ωB is the Faraday rotation rate and τC = κ−1 is the photons mean free path (in con-
formal time units). When calculating the evolution of each mode ~k we have assumed
that the strength of the primordial magnetic field scales as B(t) = B(t∗)a2(t∗)/a2(t),
which is justified by flux conservation and because the Universe behaves as a good
conductor [6]. Since the frequency also redshifts as ν = ν0a(t0)/a(t), the parameter F
is time-independent. ν0 is the frequency of the CMB photons at present time, while
B∗ is the strength of the magnetic field at a redshift z∗ = 1000, around recombina-
tion. Within a standard thermal history, with no early reionization, depolarization
is only significant across the LSS, and it thus depends only upon the value of the
primordial magnetic field around the time of recombination. Notice that Faraday
rotation between collisions becomes considerably large, paving the way to an efficient
depolarizing mechanism, at frequencies around and below νd defined such that
F ≡ (cid:16)νd
ν0(cid:17)2
so that
νd ≈ 8.4 GHz (cid:16)
B∗
10−3Gauss(cid:17)1/2
≈ 27 GHz (cid:16)
B∗
0.01 Gauss(cid:17)1/2
.
(16)
(17)
From eqs. (13) we can read the values of ∆Q0 and ∆Q2. They reduce to eqs. (10)
with O(F 2) corrections for small F , while they vanish as F −1 for large F . We write
them as:
∆Q0 =
d0(F )SP
; ∆Q2 = −
1
2
1
10
d2(F )SP
.
(18)
The coefficients d0, d2 are defined so that di ≈ 1 + O(F 2) for small F , while di →
O(1/F ) as F → ∞, and represent the effect of depolarization. They read:
d0(F ) =
3
2harctan(F )
F
(1 +
d2(F ) =
15
4 harctan(F )
F
(1 +
4
F 2 +
6
1
F 2 ) −
3
F 4 ) −
1
F 2i
3
F 2 −
(19)
(20)
3
F 4i
.
In terms of the combination
5
6
d ≡
(d0 +
d2
5
) =
15
8 harctan(F )
F
(1 +
2
F 2 +
1
F 4 ) −
5
3F 2 −
1
F 4i
and using the definition of SP , we find the relation
∆T2 = −SP (1 −
3
5
d)
and from the equation for ∆T in the tight coupling limit we get
SP =
4
3(3 − 2d)
ikτC∆T1 = −
4
3(3 − 2d)
τC ∆0
.
(21)
(22)
(23)
Notice that d ≈ 1 − F 2/7 if F << 1 while d → 15
16πF −1 for large F . We stress
here again that in a general case the depolarizing coefficient d depends upon the
angle between ~k and ~B. The net anisotropy and polarization being the outcome of
the stochastic superposition of all Fourier modes of the density-fluctuations, with
a spectrum that has no privileged direction, the average depolarizing factor, after
superposition of all wavevectors in arbitrary orientations with respect to the magnetic
field, depends only upon F . The average depolarizing factor might slightly differ from
that calculated with ~k k ~B, which at most underestimates the average effect.
Equations (13,22,23) condense the main effects of a magnetic field upon polariza-
8 ∆T2 sin2 θ.
tion. When there is no magnetic field (F = 0, d = 1) ∆U = 0 and ∆Q = − 15
A magnetic field generates ∆U , through Faraday rotation, and reduces ∆Q. In the
limit of very large F (large Faraday rotation between collisions) the polarization van-
ishes. The quadrupole anisotropy ∆T2 is also reduced by the depolarizing effect of
the magnetic field, by a factor 5/6 in the large F limit, because of the feedback of
∆Q upon the anisotropy or, in other words, because of the polarization dependence
of Thomson scattering. The dipole ∆T1 and monopole ∆T0 are affected by the mag-
netic field only through its incidence upon the damping mechanism due to photon
diffusion for small wavelengths, that we shall discuss in detail in section III. Indeed,
the equation for ∆0 = ∆T0 + Φ, neglecting O(R2) contributions, now reads:
∆0 + h
R
1 + R
+
16
90
(5 − 3d)
(3 − 2d)
k2τC
(1 + R)i ∆0 +
k2
3(1 + R)
∆0 =
k2
3(1 + R)
[Φ − (1 + R)Ψ]
.
(24)
The damping term is reduced by a factor 5/6 at frequencies such that d << 1, for
which depolarization is significant.
We have assumed that the magnetic field is spatially homogeneous. We can expect
corrections to our result if the field is inhomogeneous over scales smaller than τC
at any given time around decoupling.
Indeed, if the field reverses its direction N
times along a photon path during a time τC, Faraday rotation will not accumulate
7
as assumed above. In that case depolarization would start to be significant only at
those frequencies such that Faraday rotation is large over the scale on which the
magnetic field reverses its direction. The frequencies at which depolarization starts
to be significant would thus be reduced by a factor 1/√N .
2.3 Frequency-dependence of the degree of polarization
The anisotropy and polarization observed at present time can be evaluated using the
formal solutions of eqs. (4,5,6)
∆T (τ0) = R τ0
∆Q(τ0) = Z τ0
0
2 P2(µ)SP (τ )]
0 dτ eikµ(τ −τ0)g(τ )[∆T0(τ ) + µVb(τ ) − 1
0 dτ eikµ(τ −τ0)e−κ(τ0,τ )( Ψ − Φ)
+R τ0
dτ eikµ(τ −τ0)g(τ ){
1
2
[1 − P2(µ)]SP (τ ) + F ∆U (τ )}
where
∆U (τ0) = −Z τ0
0
dτ eikµ(τ −τ0)g(τ )F ∆Q(τ )
g(τ ) ≡ κe−κ(τ0,τ )
(25)
(26)
(27)
(28)
is the visibility function. It represents the probability that a photon observed at τ0
last-scattered within dτ of a given τ . For a standard thermal history, with no signifi-
cant early reionization after recombination, g(z) is well approximated by a Gaussian
centered at a redshift of about z ≈ 1000 and width ∆z ≈ 80 [20]. In conformal time,
we shall denote the center and width of the Gaussian which approximately describes
the process of decoupling by τD and ∆τD respectively.
The visibility function being strongly peaked around the time of decoupling, the
first integral in eq. (25) for the anisotropy is well approximated, at least for wave-
lengths longer than the width of the last scattering surface, by its instantaneous
recombination limit. In that case it reduces to the tight-coupling expression of its
integrand evaluated at time τ = τD [9]. This first integral is dominated by its first
two terms, proportional to the monopole ∆T0 and the baryon velocity Vb (in turn
proportional to ∆T1) respectively. The quadrupole term SP gives a negligible contri-
bution for long wavelengths, but becomes relatively significant on small scales. The
second integral in eq. (25) corresponds to the anisotropies induced by time-dependent
potentials after the time of last-scattering.
Eqs. (26,27) for the polarization can be approximated replacing the integrand by
its tight-coupling expression. Then
∆Q(τ0) =
3
4
sin2 θ
(1 + F 2 cos2 θ) Z τ0
0
dτ eik cos θ(τ −τ0)g(τ )SP (τ )
(29)
8
while the total polarization, ∆P = (∆2
∆U (τ0) = −F cos θ∆Q(τ0)
U )1/2 reads
Q + ∆2
∆P (τ0) = √1 + F 2 cos2 θ∆Q(τ0)
(30)
(31)
Evaluation of the time integral in eq. (29) requires a more detailed knowledge of
the time-dependence of the integrand than in the case of the anisotropy.
Indeed,
the tight-coupling expression (23) for the quadrupole term SP being proportional to
the mean free path τC, which varies rapidly during decoupling, the instantaneous
recombination approximation becomes inappropriate. The induced polarization is,
indeed, proportional to the width of the last scattering surface. Adapting the method
of [8] to include also the effect of the primordial magnetic field, we write down the
equation satisfied by SP when all other quantities are already approximated by their
first-order tight-coupling expressions:
SP +
3
10
(3 − 2d) κSP =
2
5
ik∆T1
(32)
Neglect of SP returns the tight-coupling result of eq. (23). Instead, the formal solution
to equation (32)
SP (τ ) =
2
5
ik Z τ ′
0
dτ ′∆T1e−
3
10 κ(τ,τ ′)(3−2d)
(33)
tracks down the time-dependence of SP through the decoupling process with better
accuracy.
For wavelengths longer than the width of the LSS we can neglect the time variation
of ∆T1 and that of eik cos θ(τ −τ0) around decoupling. We also approximate the visibility
function by a Gaussian, which justifies the approximation κ(τ0, τ ) ≈ −κ(τ0,τ )
[21].
Then
∆τD
3
10 xκ(3−2d)
e−
,
(34)
SP (τ ) ≈
2
5
ik∆T1(τD)∆τDe
3
10 κ(τ0,τ )(3−2d) Z ∞
1
dx
x
where the integration variable has been changed to x = κ(τ0,τ )
approximations,
κ(τ0,τ ′). Thus, within these
R τ0
0 dτ g(τ )SP (τ ) = − 2
= 4
5 ik∆T1(τD)∆τD R ∞
1+6d ik∆T1(τD)∆τD[ln( 10
0 dκe−
1+6d
10 κEi(− 3
3 ) − ln(3 − 2d)]
10 (3 − 2d)κ)
(35)
Finally, the total polarization induced at an angle θ with respect to the wavector
~k, reads
∆P (τ0) =
3
(1 + 6d)
[ln(
10
3
) − ln(3 − 2d)]
sin2 θeik cos θ(τD −τ0)
√1 + F 2 cos2 θ
ik∆T1(τD)∆τD .
(36)
9
It can also be written as follows, in terms of the polarization that would be induced
if there were no magnetic field (or equivalently, in terms of the polarization at fre-
quencies large enough such that the depolarizing effect is negligible):
where we have defined the depolarizing factor as
∆P (θ, F ) = D(θ, F )∆P (B = 0)
with
D(θ, F ) =
1
√1 + F 2 cos2 θ
f (F )
f (F ) =
7
1 + 6dh1 −
ln(3 − 2d)
ln(10/3) i
(37)
(38)
(39)
Eq. (38), together with the defining eqs. (15,16) and (21) for F and d, summarize
the main result of this section. Notice that
f → 1
as F → 0 ;
f → 0.61 as F → ∞
(40)
The polarization observed at present times depends upon the angle between the
line of sight and the orientation of the magnetic field at the time of decoupling. There
will be no depolarization if the magnetic field is perpendicular to the line of sight.
The magnetic field is likely to change orientation randomly over scales longer than
the Hubble radius at the time of decoupling, which subtends an angle of order one
degree in the sky, so that after averaging over many regions separated by more than
a few degrees, we can always expect a net average depolarizing effect. To roughly
estimate its order of magnitude we could assume an average component of ~B parallel
to the observation direction of order B/√2 and define an average ¯D as
¯D =
1
q1 + F 2/2
f (F )
.
(41)
Fig. 1 displays the depolarizing factor ¯D as a function of the CMB frequency ν0. We
have plotted it for three different values of the magnetic field B∗ to help visualize
the relevant frequency range, but notice that since depolarization depends only upon
F = (νd/ν0)2, the plot for an arbitrary value of B∗ is identical to that corresponding
to another value of the magnetic field after an appropriate scaling of the frequency
units, proportional to the square root of the magnetic field.
At low frequencies, those for which the effect is large, the average depolarizing
factor scales as
√2
F ≈ 0.85(cid:16) ν0
νd(cid:17)2
¯D ≈ 0.6
if
ν0 << νd
.
At comparatively large frequencies instead
¯D ≈ 1 − 0.36F 2 = 1 − 0.36(cid:16)νd
ν0(cid:17)4
if
ν0 >> νd
.
(42)
(43)
10
3 Effects upon the anisotropy
Depolarization by a primordial magnetic field has significant and potentially measur-
able effects upon the anisotropy of the CMB on small angular scales.
Indeed, the
polarization properties of the CMB feed back into its anisotropy, as evidenced in eq.
(4), due to the polarization dependence of Thomson scattering. The dominant effect
of polarization upon anisotropy derives from its impact upon the photon diffusion
length[8, 14, 15], which damps anisotropies on small angular scales [9, 22, 23]. It was
shown in [14], through numerical integration of the Boltzmann equations, that neglect
of the polarization properties of the CMB leads to an overestimate of its anisotropy
on small angular scales as large as 10%. We thus expect depolarization by a primor-
dial magnetic field to introduce a significant frequency-dependent distortion of the
CMB anisotropy power spectrum. Notice that a different (frequency-independent)
distortion of the CMB anisotropy power spectrum by a primordial magnetic field,
due to its impact upon the photon-baryon fluid sound speed, was recently discussed
in [13].
3.1 Reduced diffusion-damping
Photon diffusion damps anisotropies on small angular scales [9, 22, 23]. The effect
is described by the term proportional to k2τC ∆0 in eq. (24). The photon diffusion
length depends upon the degree of polarization of the CMB [8, 14, 15]. Thus, the
photon-diffusion length is different at frequencies where the depolarizing effect is
significant.
The damping of anisotropies on small angular scales due to photon diffusion can be
found, now including the full R-dependence, by solving the tight-coupling equations
to second order, assuming solutions of the form
∆X(τ ) = ∆X eiωτ
(44)
for X = T , Q, and U, and similarly for the baryon velocity Vb. One then finds that
ω =
k
q3(1 + R)
+ iγ
with the photon-diffusion damping length-scale determined by
γ(d) ≡
k2
k2
D
=
k2τC
6(1 + R)(cid:16) 8
15
(5 − 3d)
(3 − 2d)
+
R2
1 + R(cid:17)
(45)
(46)
The depolarizing effect of the magnetic field reduces the viscous damping of anisotro-
pies. In the case of small R, such that R2 terms can be neglected, the damping factor
11
γ is smaller by a factor 5/6 at those frequencies for which the depolarizing effect is
large.
We make now an analytic estimate of the effect of the frequency-dependence of
the photon diffusion length, in the presence of a primordial magnetic field, upon the
CMB anisotropy power spectrum. The temperature anisotropy correlation function
is typically expanded in Legendre polynomials as
C(θ) =< ∆T (n1)∆T (n2) >n1·n2=cos θ=
1
4π
∞
Xl=0
(2l + 1)ClPl(cos θ)
.
(47)
The multipole coefficients of the anisotropy power spectrum are given by
Cl = (4π)2 Z k2dkP (k)∆Tl(k, τ0)2
(48)
with P (k) the power-spectrum of the scalar fluctuations, assumed scale-invariant in
the sCDM model. The largest contribution to a given multipole Cl comes from those
wavelenghts such that l = k(τ0−τD), where τ0 is the conformal time at present and τD
the conformal time at decoupling. The average damping factor due to photon diffusion
upon the Cl's is given by an integral of e−2γ times the visibility function across the
last scattering surface [9]. It depends upon cosmological parameters, notably R, and
upon the recombination history. Approximately, and for a standard cold dark matter
model, we can take 2γ(d = 1) ≈ (l/1500)2. The relative change in the Cl's due to
the change in the photon-diffusion length, as we move down from frequencies where
depolarization is insignificant (d = 1) to lower frequencies (d << 1), is then given by
∆Cl =
Cl(d)
Cl(d = 1) − 1 ≈ exp(cid:16)(l/1500)2(1 − d)
(6 − 4d)
(cid:17) − 1
.
(49)
In figure 2 we plot ∆Cl (expressed as a percentage) at l = 1000 as a function of
frequency, for three different values of the magnetic field B∗ = 0.001, 0.01 and 0.1
Gauss. Once again, the graph for an arbitrary value of B∗ can be read from any of
these with an appropriate scaling of the frequency units. We have chosen to display
the effect at l = 1000, that will be accessible by the recently funded CMB satellite
experiments, MAP [24] and COBRAS/SAMBA [25].
3.2 Reduced quadrupole contribution
The depolarizing effect of a primordial magnetic field also changes the strength of the
quadrupole term SP around decoupling, and thus its incidence upon the anisotropy
of the CMB on small angular scales.
Indeed, the quadrupole anisotropy and the
polarization of the CMB at the time of recombination contribute to the presently
observed anisotropy through the following term of eq. (25)
∆SP (τ0) ≡ −
1
2
P2(cos θ)Z τ0
0
dτ eik cos θ(τ −τ0)g(τ )SP (τ )
.
(50)
12
This term is negligible for long wavelengths, those that dominate the lowest multipoles
of the present anisotropy, but becomes non-negligible on small angular scales (large
multipoles). Indeed, in the tight coupling approximation SP ∝ τC ∆0 and thus, barring
a very strong time-dependence of the scalar potential, the contribution of SP is well
below that of the monopole term, except for small wavelengths.
The depolarizing effect of a magnetic field modifies the value of SP around de-
coupling, compared to what it would have been if there were no magnetic field, and
then
∆SP (F ) = f (F )∆SP (B = 0)
(51)
with f as defined in eq.
(39). Recall that f ≈ 1 if ν >> νd while f ≈ 0.6 if
ν << νd. Thus, at frequencies such that the depolarizing effect of the magnetic field is
significant, the partial contribution of the quadrupole term SP to the total anisotropy
is reduced by a factor 0.6 compared to that at frequencies where depolarization is
unimportant. On small angular scales this could represent a decrease of the anisotropy
by a few per cent. The effect is opposite to that of the change in diffusion damping,
but is likely to be less significant on small angular scales.
3.3 Numerical estimate of the effect upon the anisotropy
In order to accurately ascertain the net effect of the depolarizing mechanism upon
the CMB anisotropy and to make definite quantitative predictions within a standard
cosmological model, we turn now to the numerical integration of the Boltzmann equa-
tions (4,5,6). We use the recently developed code CMBFAST[16], that integrates the
sources over the photon past light cone. Its starting point are the formal solutions
(25,26,27), where the geometrical and dynamical contributions are separately handled
to improve efficiency. As in our analytic estimates, when computing the evolution of
each Fourier mode we introduce in the code the Faraday rotation term with the angu-
lar dependence corresponding to the case where ~B has no component perpendicular
to ~k.
Figures 3 and 4 summarize the numerical calculation of the effect of depolarization
upon temperature anisotropy, in a standard cold dark matter model (sCDM).
The quantity plotted in Fig. 3 is l(l + 1)Cl, for the sCDM model without a
magnetic field and with a magnetic field and at frequencies such that F = 1, 4, 9,
corresponding to ν0 = νd, νd/2 and νd/3 respectively. Fig. 3 clearly shows that the
CMB anisotropy on small angular scales increases at frequencies where depolarization
is significant. This result indicates that the reduction in diffusion damping due to
depolarization is the dominant effect among the two opposite effects discussed in the
previous sections.
Fig. 4 displays the same results but expressed in terms of ∆Cl, the percentual
increase in Cl relative to the case without magnetic field. The monotonic curves
13
in the same figure, included for comparison purposes, correspond to the analytic
estimate of the effect of reduced diffusion damping, eq.(49). As expected, the effect is
larger on smaller angular scales (larger l). The numerical result approximately follows
the analytic estimate of the effect of the reduction in diffusion damping. The total
effect, however, does not increase monotonically with l. This can be understood as
a consequence of the nature of the subdominant quadrupole contribution SP , which
oscillates out of phase with the Cl's [8] (remember that SP ∝ ∆0), and is reduced by
depolarization through a factor f (F ).
It is also clear from Fig. 4 that the analytic result for the change in diffusion
damping due to depolarization overestimates the total effect at high l. This is because
the actual damping in the Cl spectra has two contributions, one from Silk damping
and the other due to cancellations in the integral across the last scattering surface
produced by the oscillations in the exponential and sources in equation (25). Only
Silk damping is reduced by the magnetic field, and that is why equation (49) slightly
overestimates the net effect.
The analytic and numeric calculations are in very good agreement around l ≈
1000. The frequency-dependence of ∆Cl at l = 1000 plotted in Fig. 2 fits very well
the analogous result after the full numerical integration of the Boltzmann equations.
We conclude that the depolarizing effect of the magnetic field results in an increase
of the anisotropy correlation function multipoles of up to 7.5% (for sufficiently low
frequencies) on small angular scales (l ≈ 1000), those that will be accessible by future
CMB satellite experiments such as MAP and COBRAS/SAMBA. The frequencies at
which the effect is significant, however, depend on the strength and coherence length
of the primordial magnetic field at the time of recombination.
Depolarization depending upon frequency, the effect might be difficult to separate
from foreground contamination. The relative change of the Cl's at l = 1000 is larger
than 2% on frequencies below 30 GHz (accessible to the first two channels in MAP),
if B∗ = 0.02 Gauss or larger. The first two channels in COBRAS/SAMBA being
at 31.5 and 53 GHz, the signal would reach a 2% level within this range only if B∗
is around or larger than 0.1 Gauss. However, COBRAS/SAMBA might reach out
to larger values of l, where ∆Cl is larger, and might thus have a sensitivity to the
depolarizing effect of B∗ comparable to MAP. In any case, both experiments will be
sensitive to a magnetic field around B∗=0.1 Gauss, and would thus at least be able
to place a direct constraint on B∗ comparable or better than the one obtained from
extrapolation of the nucleosynthesis bound.
Experiments searching CMB anisotropy and polarization at smaller frequencies,
which currently operate down to 5 GHz [26, 27], may play a significant role to detect
the depolarizing effect of a primordial magnetic field.
14
4 Conclusion
The CMB is expected to have a small degree of linear polarization. Several esti-
mates were made for the predicted polarization, both in the context of anisotropic
cosmological models [1, 28], as well as in isotropic and homogeneous cosmologies per-
turbed with either energy-density fluctuations or gravitational waves [18, 21, 29]. The
polarization of the CMB remains undetected, its upper limit being P < 6 × 10−5 [3].
A primordial magnetic field depolarizes the CMB radiation on those frequencies
that experience a significant amount of Faraday rotation around the time of decou-
pling.
[8]
to estimate the depolarizing effect of a primordial magnetic field across the last-
scattering surface, assuming a standard ionization history. The result is expressed by
eqs. (38,42,43) and is represented in Fig. 1. The CMB becomes significantly depo-
larized at frequencies around and below 30 GHz (B∗/0.01 Gauss)1/2, below which the
degree of polarization decreases quadratically with frequency. B∗ is the value of the
primordial field at a redshift z∗ = 1000, around recombination, likely to be 106 times
larger than an hypothetical cosmological magnetic field at present times.
In this paper, we have applied the analytic method developed in Ref.
The average depolarizing factor depends only upon the parameter F , as defined by
eq. (15), which represents the average Faraday rotation between collisions. We have
calculated the depolarizing factor d(F ), as given by equation (21), in the particular
case of a wavevector ~k k ~B. In a general case, the factor d depends upon the angle
between ~k and ~B. This dependence integrates away in average quantities, after the
stochastic superposition of all Fourier modes of the density fluctuations. The value
derived here for d is at most an underestimate of the average depolarizing effect,
which would eventually start to be significant at slightly larger frequencies. Our
derivation also assumed that Faraday rotation accumulates over the width of the last
scattering surface. If the primordial magnetic field is very entangled over that scale,
the depolarizing effect starts to be significant at smaller frequencies.
The depolarizing mechanism has a significant effect upon the anisotropy of the
CMB on small angular scales. On those angular scales and at frequencies such that
the depolarizing effect is large, the damping of anisotropies by photon diffusion is
reduced, which results in a significant increase of the anisotropy at a fixed angular
scale. Besides, depolarization reduces the contribution of the intrinsic quadrupole
anisotropy. Figure 2 displays the estimate for the percentual change of the anisotropy
power spectrum at l = 1000 due to the reduction in diffusion damping, as a function
of frequency and for different values of the primordial magnetic field at recombination.
We conclude that a primordial magnetic field increases the anisotropy of the CMB
by up to 7.5% at l ≈ 1000 in a standard CDM cosmology. The asymptotic strength of
the effect is independent of the intensity of the magnetic field, but the frequencies at
which it starts to be significant are those around and below 30 GHz (B∗/0.01Gauss)1/2.
15
Measurements of anisotropy and polarization at sufficiently low frequencies could
probe primordial magnetic fields in an interesting range.
Acknowledgements
DH and JH are grateful to Nathalie Deruelle and the DARC at Meudon for hospitality
while working on this project. The work of DH was partially supported by an EEC,
DG-XII Grant No. CT94-0004, and by CONICET. JH is grateful to the Royal Society
for a grant.
References
[1] M. Rees, Astrophys. J. 153,L1 (1968).
[2] S. Chandrasekhar, Radiative Transfer, Oxford University Press, Oxford, Eng-
land (1950).
[3] P. M. Lubin and G.F. Smoot, Astrophys. J. 273, L1 (1983).
[4] E. Milaneschi and R. Fabbri, Astron. Astrophys. 151, 7 (1985).
[5] A. Kosowsky and A. Loeb, "Faraday rotation of microwave background polar-
ization by a primordial magnetic field", astro-ph/9601055.
[6] For reviews on extragalactic magnetic fields see H. Asseo and H. Sol, Phys. Rep.
148, 307 (1987); P.P. Kronberg, Rep. Prog. Phys. 325, 382 (1994).
[7] B. J. Burn, Mon. Not. R. Astr. Soc. 133, 67 (1966).
[8] M. Zaldarriaga and D. Harari, Phys. Rev. D52, 3276 (1995).
[9] W. Hu and N. Sugiyama, Phys. Rev. D 51, 2599 (1995); Astrophys. J. 445, 521
(1995).
[10] P.J.E. Peebles and J.T. Yu, Astrophys. J. 162, 815 (1970).
[11] M. Turner and L. Widrow, Phys. Rev. D37, 2743 (1988); T. Vachaspati, Phys.
Lett. B265, 258 (1991); B. Cheng and A. Olinto, Phys. Rev. D50, 2421 (1994);
M. Gasperini, M. Giovannini and G. Veneziano, Phys. Rev. Lett. 75, 3796
(1995); D. Lemoine and M. Lemoine, Phys. Rev. D52, 1955 (1995); F.D. Mazz-
itelli and F.M. Spedalieri, Phys. Rev. D52, 6694 (1995).
[12] D. Grasso and H.R. Rubinstein, Astroparticle Phys. 3, 95 (1995); B.L. Cheng,
A. Olinto, D.N. Schramm and J.W. Truran, preprint astro-ph/9602055.
16
[13] J. Adams, U.H. Danielson, D. Grasso and H. Rubinstein, "Distortion of the
acoustic peaks in the CMBR due to a primordial magnetic field", astro-
ph/9607043.
[14] W. Hu, D. Scott, N. Sugiyama and M. White, Phys. Rev. D52, 5498 (1995).
[15] N. Kaiser, Mon. Not. R. Astron. Soc. 202, 1169 (1983).
[16] U. Seljak and M. Zaldarriaga, "A line of sight integration approach to CMB
anisotropies", astro-ph/9603033.
[17] G. Jungman, M. Kamionkowski, A. Kosowsky and D.N. Spergel, Phys. Rev.
Lett. 76, 1007 (1996). See also astro-ph/9605147 for a discussion in terms of the
parameters of the MAP and COBRAS/SAMBA planned experiments.
[18] J.R. Bond and G. Efstathiou, Astrophys. J. 285, L45 (1984).
[19] A. Kosowsky, Ann. Phys. 246, 49 (1996).
[20] B.T.J. Jones and R.F.G. Wyse, Astron. Astrophys., 149, 144 (1985).
[21] A. Polnarev, Sov. Astron. 29, 607 (1985).
[22] J. Silk, Astrophys. J. 151, 459 (1968).
[23] P.J.E. Peebles, "The Large Scale Structure of the Universe", Princeton Univer-
sity Press (1980).
[24] Visit the MAP home page at http://map.gsfc.nasa.gov.
[25] Visit the COBRAS/SAMBA homepage at http://astro.estec.esa.nl:80/SA-
general/Projects/Cobras/cobras.html.
[26] E. Fomalont, B. Partridge et al., Astrophys. J. 404, 8 (1993).
[27] Visit the CMB Jodrell Bank experiment home page at http://www.jb.man.ac.-
uk/sjm/cmb jb.html.
[28] G.P. Nanos, Astrophys. J. 232,341 (1979); M. Basko and A. Polnarev, Sov.
Astron. 24, 3 (1979); J. Negroponte and J. Silk, Phys. Rev. Lett. 44,1433 (1980);
B.W. Tolman and R.A. Matzner, Proc. R. Soc. London Ser. A 392, 1803 (1984);
B.W. Tolman, Astrophys. J. 290,1 (1985).
[29] J.R. Bond and G. Efstathiou, Monthly Not. R. Astron. Soc. 226, 655 (1987); E.
Milaneschi and R. Valdarnini, Astron. Astrophys. 162,5 (1986);R. Crittenden,
R.L. Davis and P.J. Steinhardt, Astrophys. J. 417, L13 (1993); D.D Harari and
M. Zaldarriaga, Phys. Lett. B 319, 96 (1993); R.A. Frewin, A.G. Polnarev, P.
17
Coles, Monthly Not. R. Astron. Soc. 266, L21 (1994); K. L. Ng and K.W. Ng,
Phys. Rev. D 51, 364 (1995) and Astrophys. J. 445, 521 (1995); D. Coulson,
R.G. Crittenden and N.G. Turok, Phys. Rev. Lett. 73, 2390 (1994) and Phys.
Rev. D52, 5402 (1995).
18
B
=0.001G
*
B*B*
=0.01G
=0.1G
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
_
D
0
0
10
20
40
50
60
30
0 (GHz)
ν
Figure 1: The average depolarizing factor ¯D as a function of the CMB frequency ν0.
The corresponding figure for an arbitrary value of B∗ is identical to any of these after
a scaling of the frequency units, proportional to B1/2
.
∗
19
8
7
6
5
∆ C
(%)
l
4
3
2
1
0
B*=0.001G
B*=0.01G
B*
=0.1G
(l = 1000)
0
10
20
40
50
60
30
0 (GHz)
ν
Figure 2: Analytic estimate of the percentual change due to reduction in diffusion
damping of the l = 1000 anisotropy correlation function multipoles as a function
of the CMB frequency for different strengths of the primordial magnetic field at
recombination. The corresponding figure for arbitrary B∗ can be obtained from any
of these after a scaling of the frequency units, proportional to B1/2
∗
20
π
2
/
0
1
0
1
x
l
C
)
1
+
l
(
l
4
3.5
3
2.5
2
1.5
1
0.5
0
0
sCDM
F=1
F=4
F=9
200
400
600
800
1000
1200
1400
1600
l
Figure 3: Numerical integration for the multipoles of the anisotropy correlation func-
tion in a standard CDM model without a primordial magnetic field (F = 0), and
with F = 1, 4, 9, which correspond to ν0 = νd, νd/2, νd/3 respectively, with νd ≈
27 GHz (B∗/0.01Gauss)1/2.
21
F=1
F=4
F=9
16
14
12
10
8
6
4
2
0
∆ C
(%)
l
-2
0
200
400
600
1000
1200
1400
1600
800
l
Figure 4: Numerical result for the percentual change of Cl as a function of l relative
to its value without magnetic field in a standard CDM model. The monotonic curves
also shown for comparison purposes correspond to the analytic estimate of the effect
of reduced diffusion damping.
22
|
astro-ph/9408020 | 1 | 9408 | 1994-08-05T18:02:42 | Are BL Lacertae Objects Beamed QSO Remnants? | [
"astro-ph"
] | This paper considers the hypothesis that BL Lacertae objects (BLLs) are the beamed remnants of Quasi Stellar Objects. The hypothesis explains why BLLs do not undergo the strong evolution seen in other active galactic nuclei since it naturally predicts that the space density of BLLs should increase with cosmic time, as shown by recent observations. Numerical models reproduce, with reasonable parameters, the known redshift and magnitude counts of BL Lac objects. It is assumed that radio-quiet as well as radio-loud quasars are capable of generating jets but that jets are snuffed in young radio-quiet objects and only emerge in aged ones. I argue that the observations allow this assumption. | astro-ph | astro-ph | Are BL Lacertae Objects Beamed QSO Remnants?
E.F. Borra
Centre d'Optique, Photonique et Laser, Observatoire Astronomique du Mont
Département de Physique, Université Laval, Québec, Canada G1K 7P4
Mégantic,
SUBJECT HEADINGS: BL Lacertae objects -- galaxies: jets -- galaxies: nuclei
-- radio sources: galaxies
To appear in: COMMENTS ON ASTROPHYSICS Vol. 17 No 5
4
9
g
u
A
5
0
2
0
8
0
4
9
/
h
p
-
o
r
t
s
a
ABSTRACT
2
This paper considers the hypothesis that BL Lacertae objects (BLLs) are the beamed
remnants of Quasi Stellar Objects. The hypothesis explains why BLLs do not undergo the
strong evolution seen in other active galactic nuclei since it naturally predicts that the space
density of BLLs should increase with cosmic time, as shown by recent observations.
Numerical models reproduce, with reasonable parameters, the known redshift and
magnitude counts of BL Lac objects. It is assumed that radio-quiet as well as radio-loud
quasars are capable of generating jets but that jets are snuffed in young radio-quiet objects
and only emerge in aged ones. I argue that the observations allow this assumption.
3
1. INTRODUCTION
It is increasingly accepted that the objects that populate the high-energy extragalactic
"zoo" are Active Galactic Nuclei (AGNs), that they are manifestations of the same basic
phenomenon and that they all will eventually be unified within energetic, geometrical or
evolutionary schemes. Among AGNs, the BL Lacertae objects (hereafter referred to as
BLLs) are particularly puzzling "beasts", since they exhibit extreme characteristics among
the AGNs. A consensus has emerged that these characteristics are due to relativistic
beaming1. Over the past several years, evidence has gradually built up showing a puzzling
behavior that perhaps differentiates them the most from other AGNs: They do not show
the large decrease in comoving density with cosmic time exhibited by other AGNs. On the
contrary, Morris et al.2 recently found evidence of negative evolution, namely that the
space density of BLLs actually increases with cosmic time.
Borra 3 has proposed that BL Lacertae objects are the beamed remnants of Quasi
Stellar Objects (QSOs).This paper was probably not noticed at that time by workers active
in AGN research because the hypothesis was used in a cosmological test rather than
presented as the theme of the paper. What prompted this hypothesis was that it naturally
explains why BLLs do not undergo the strong evolution seen in other AGNs; indeed
negative evolution is a natural consequence of it. Borra 3 made the assumption that all
Quasi Stellar Objects (QSOs), radio-quiet (RQQ) as well as radio-loud (RLQ), are capable
of generating jets but that jets are snuffed in young RQQs and only emerge in aged
objects. This assumption was necessary to obtain good agreement between the space
densities of the objects concerned. Prima facie, this assumption runs counter the prevailing
belief that radio-quiet and radio-loud quasars are distinct objects hosted in different types
of galaxies (spirals versus ellipticals) located in different environments (poor versus rich
clusters of galaxies). On the other hand, radio-loud and radio-quiet QSOs have remarkably
4
similar spectra and powers, so that it is natural to assume that the same basic mechanisms
are at work and that, for some reason, energy in the radio region is either absorbed or not
generated in the RQQs. Section III further addresses this issue.
In this paper, I reexamine the hypothesis in the light of 10 years of advances
in our knowledge of AGNs. I assume that all QSOs have jets but that the jet is quenched in
the young RQQs. As the QSO ages, the quenching mechanism weakens allowing the jet to
emerge. I shall also argue that the observational evidence on the environmental statistics of
the RLQs and RQQs does not rule out the hypothesis.
2. NUMERICAL MODELS
The basic hypothesis assumes snuffing that decreases in strength with age but does
not identify a specific snuffing model. Let us simply assume that all AGNs are capable of
generating a jet and that material surrounding the nucleus is somehow responsible for
snuffing it in radio-quiet QSOs. This material may be in a leaky shell, or torus,
surrounding the central powerhouse, with holes or weak spots that allow the emergence of
a jet only if it is aimed at one of them. With a toroidal geometry, the jet can be directed
along directions not necessarily aligned with the axis of symmetry of the torus. We shall
not make any assumption on the snuffing mechanism itself but assume that the material
interferes somehow with the mechanism that either accelerates the material or delivers it to
the outer extended lobes. It seems reasonable to assume that the quantity of material
surrounding the nucleus eventually decreases as the QSO ages, so that it becomes easier
for a jet to emerge from an older quasar.
One could in principle evolve the observed luminosity function of QSOs from high
to low redshifts, compute the rate of formation of remnants and compare the predictions to
the observed luminosity function of the local BL Lac objects. However, in the absence of
a quantitative theory, this is not a particularly fruitful approach, and we shall instead
5
simply find a local luminosity of the QRs that is compatible with the known evolution of
the luminosity function of QSOs and the local luminosity function of BLLs. We then shall
see whether relativistic beaming can give a luminosity function that reproduces the
observations.
Urry & Shafer 4 have studied the effects of relativistic beaming on a luminosity
function. Following 4, we shall assume a power law for the isotropic unbeamed
component of the luminosity function of the local QSO remnants (QR):
F QR(L,z=0) dL = KL-B dL for L1<L< L2
= 0
for L<L1 or L<L2
.
(1)
As Urry & Shafer 4 point out, the upper and lower luminosity cutoffs, although
unrealistic, approximate the fact the luminosity function must turn down at some
luminosity. They also show that relativistic beaming changes the original power law
luminosity function in a way that can be approximated by two power laws, the high
luminosity end having index ~ B, and the low luminosity end having a flattened exponent
~ (p+1)/p, where p depends on the shape of the spectrum, the structure of the jet and the
frequencies being compared; it usually is in the range 3<p<5. We keep the notation of 4
and the reader is referred to that paper for more details.
If QSOs are long lived, the local density of their remnants should be essentially
equal to the density of quasars at z = 2 and
M urF QR(M,z=0.0)dM =
M lr
M uqF QSO(M,z=2.0)dM ,
M lq
(2)
6
where F QR(M,z) and F QSO(M,z) are the luminosity functions of QSO remnants and
QSOs. This approximates the actual situation since QSOs are probably continuously
formed rather than in a burst at z= 2.0 as implied by Equation 2. The Mu and Ml upper
and lower magnitude cut-offs take into account the fact that the luminosity functions must
turn down at some magnitudes and that very faint QSOs probably do not evolve
significantly to our epoch. The unbeamed luminosity function is further restricted by the
requirement that local QSO remnants are inconspicuous and therefore much less luminous
than QSOs. Therefore we must have, at least for MV <-22
F QR(M,z=0) << F QSO(M,z=0).
(3)
We now feel free to experiment with luminosity functions of the form given by Equation 7
in 4 subject only to the conditions set by Equations 2 and 3.
The redshift distribution given in 2 has two peculiarities: It shows a very sharp
drop at z < 0.2 and a more gradual, but still rapid, decrease for z > 0.2. A narrow
luminosity function having the appropriate mean luminosity, combined with negative
evolution of the space density of the objects can reproduce this. Such a luminosity
function naturally arises from a beamed population of low-luminosity galactic nuclei
having a power-law luminosity function with a steep index and small luminosity range.
The small luminosity range may be real and approximate a sharply peaked luminosity
function (e.g., Gaussian-like). One also may justify the low-luminosity cut-off by the fact
that a faint object no longer looks like a BL Lac object since there is a strong contribution
to the spectrum from the host galaxy; while the high luminosity drop-off approximates the
fact that the luminosity function must turn down at some luminosity. It is also important
7
that the unbeamed objects be faint since too high a mean luminosity would shift the peak
of the redshift distribution to high redshifts.
We obtain a good fit to the observations with a model having an unbeamed power-
law luminosity function with index B= 3.0 and -11 < MV > -13. This luminosity function
respects the requirements set by Equation 3, that the unbeamed component be
inconspicuous. We obtain the normalization factor K from Equation 2, where we use the
luminosity function F QSO(M,z) determined in 5 for q0=0.5 and H0=50 Km/sec/Mpc. We
use Mlq=-22 and Muq=-30, thus assuming that QSOs fainter than MB=-22 do not evolve
enough to z =0 to contribute significantly to the local population of QRs. Although the
luminosity functions in 5 are truncated at MB=-23 they show strong evolution at MB = -23,
and are compatible with significant evolution at MB=-22. There are sources of uncertainty
in K, arising from the determination of the luminosity function of QSOs, from
uncertainties in the cosmological parameters and from the lower magnitude cutoff. We use
a relativistic Lorentz factor g = 10, which is a reasonable average value 6, and p= 4 for the
exponent in Equation 1, a value appropriate for our problem. The beamed luminosity
function is computed assuming viewing angles 0.0º < q
< 5º. The beamed luminosity
function has a steep power law from -26<MV<-24 joined to a shallow one for -24<MV<-
21.5, compatible with the known luminosities of BL Lac objects. The MV<-21.5 low-
luminosity cutoff justifies truncating the viewing angle at 5º for the luminosity would then
be less than the luminosity of the host galaxy (BLLs appear in bright host galaxies) and the
object would no longer be identified as a BLL. We compute the redshift distribution from
the usual cosmological relation given by
dN
dW dz
=
m1F BL(M,z)dV
dz
m0
dm ,
(4)
where F BL(M,z) is the beamed differential luminosity function (per unit magnitude), W
the
surface area and dV(H0,q0,z) the cosmological volume element. The z dependence of the
luminosity function is given by
8
F BL (M,z) = F BL (M,z=0)(1+z)b ,
(5)
following the density evolution law determined in 2. We shall use b = -5.5,
the best fit
values determined by them for density evolution. The sample in 2 is X-ray selected and,
in principle, it would be desirable to base our computations fully on the X-ray luminosity
function of QSOs. In practice, because one needs the luminosity functions of QSOs at z =
0 and z = 2.0 (Equations 2 and 3) and because they are much better known in the optical,
especially at low fluxes and high redshifts, it is preferable to use optical luminosity
functions. For our computations, we shall use the mV distribution given in 2 for their
sample and a value of ml = 20 in Equation 4 as indicated from their mV magnitude
distribution. We use H0=50 Km/sec/Mpc and q0= 0.5. The model described gives the
redshift distribution of Figure 1 (dashed line), where the histogram gives the
observations in 2. We can see that the model fits well the main features of the redshift
distribution in 2 and it is reasonable to assume that the observations sample the computed
parent distribution of Figure 1. The computed z-distribution rises sharply, peaks at the
same redshift and rapidly drops to undetectability for z>0.8.
We have plotted in Figure 2 the integrated counts as a function of magnitude
obtained from
dN
dW dm
=
z1F BL(M,z)dV
dz
z0
dz ,
(6)
9
where the integrand is the same as in Equation 6 and all symbols and constants are the
same. We have plotted in the same figure the estimates of the optical integrated counts of
BLLs obtained in 6; the arrows indicate either lower or upper limits. We can see that the
theoretical counts also are in agreement with the observations.
Our model is a simple one since it considers the contribution of the beam alone,
assumes that all sources have the same value of g , and uses a power-law luminosity
function. It does not take into account that, in the optical, there also is a contribution from
the light of the host galaxy. Padovani and Urry 6 have considered this case and find that it
does not change the basic conclusions of the simple model. Their modified model, that
uses two additional free parameters, still predicts a double power law with slope ~B above
the break but with slope steeper than (p+1)/p below the break. They have also extended
their basic model 7 to more general luminosity functions and the case where there is a
distribution of Lorentz factors g . The uncertainties in the observations and some of the
assumptions that are made in the computations do not justify the introduction of additional
free parameters into our models nor the consideration of other types of luminosity
functions.
1 0
3. DO RADIO-QUIET QUASARS HAVE SNUFFED JETS?
a) Snuffing and energy distributions
RLQs and RQQs have remarkably similar spectral energy distributions and powers
at all wavelengths except radio wavelengths; it is therefore logical to assume that they
actually are similar objects and that the remarkable differences in their radio fluxes are due
to absorption of the radio power or snuffing of the radio emission mechanism. Another
attractive model invokes relativistic beaming and different viewing angles to explain the
difference but fails (8, 9, 10). A number of absorption processes have been considered
(e.g., synchrotron self-absorption, free-free absorption) but fail to account for the bimodal
distribution of radiopower in quasars 8. The remaining viable hypothesis, also suggested
by 9, is that some mechanism prevents a relativistic jet from fully developing and
generating the relativistic electrons responsible for the radio emission. Snuffing naturally
explains the remarkable similarity of the 2 types of quasars at all wavelengths but the radio
region, as well as the bimodality of the radio emission. It accounts for the observations of
11 who find that lobe dominated radio quasars have spectra that show deep minima in the
millimeter region, indicating thus that their radio emission is not a continuation of the
optical-infrared spectrum but that they contain normal RQQ cores with additional emission
from a morphologically unrelated radio source.
b) Changes in the number ratios (RLQ/RQQ) with z and magnitude
Under our hypothesis radio-jets emerge as the RQQs age, leading to the prediction
that the fraction of radio-loud objects should increase with age, hence decreasing z.
Therefore one could, in principle, use radio observations of optically selected QSOs to
1 1
verify these predictions. However, an observational test is difficult because we simply do
not know how the snuffing factor varies with age; hence we cannot quantitatively predict
how the ratio RLQ/RQQ varies with redshift. Furthermore, if the jets emerge when the
objects are faint, they will not longer be identified as quasars, rendering the comparison
meaningless. The redshift distribution of the X-ray selected BLLs gives some clue, since
the counts peak at z~0.3, first indicating the cosmic time at which most radio jets emerge
and second that jets emerge mostly among very evolved and very faint objects that would
not be recognized as quasars because of their very low luminosities. Comparison to the
observations must therefore be tempered by these realizations. On the other hand, it is
legitimate to assume that, although jets emerge very late in the evolution of most RQQs,
the emergence of the jets may occur earlier in some objects and there may therefore be a
tendency for the fraction of RLQs to increase with cosmic time. Whether this should be
detectable and is detected with the surveys presently available is another matter that we
shall now consider. This preamble should therefore make it clear that our hypothesis
would stand even if the test of this subsection fails.
There are very few surveys of optically selected QSOs. The most recent and largest
ones are by 10 and 12. They confirm the bimodality of the radio emission in QSOs.
Visnovsky et al. 12, who also use the data in 10 find strong evidence that in optically
selected QSOs the radio-loud fraction increases with decreasing redshift, in agreement
with our prediction but also that the ratio decreases with fainter limiting magnitude (albeit
with weaker statistical significance). Under our hypothesis, the ratio loud/quiet should
increase with decreasing redshift in optically selected samples, because at a given
magnitude, as the universe ages, one collects an increasing number of older, originally
brighter QSOs.
Because quasars get fainter as they age, one would expect, prima facie, that the
ratio should also increase with faintness; however, this prediction may be too simple-
minded. If brighter sources have a higher probability of developing a jet either simply
1 2
because the jet is stronger and can get through a snuffing shield more easily or because
the snuffing shield is ablated faster by the stronger radiation from the brighter core, this
will cause the RLQ/RQQ ratio to increase with brightness for middle-aged objects. This
picture finds some support from 12 where correlation between radio and optical
luminosity for the radio-weak sources is found, as though more luminous objects have a
stronger embryonic radio-jet trying to break through and giving some radio flux prior to
snuffing. Optically selected QSOs are found in surveys that are magnitude limited, the
absolute flux decreasing with z. They therefore include only relatively bright (typically MB
<-23), presumably middle-aged objects, missing thus the very aged objects that have all
developed jets, because they are too faint. Visnovsky et al. 12, mostly discuss QSOs with
MB<-23 and 0.1<z< 3. It is possible that in the range of brightness and redshifts
discussed in 12, the brightness effect dominates over the aging effect thus explaining why
the ratio RLQ/RQQ increases with brightness.
I do not claim that this discussion proves that RQQs have snuffed jet but
only that the observations do not rule this hypothesis. Section 3b is particularly speculative
but , on the other hand, it could be dropped without affecting our discussion. The stronger
argument in the favor of the snuffing hypothesis is by absentia of an alternative.
4. DISCUSSION
Prima Facie, the weakness of this work comes from the assumption that RLQs and
RQQs are basically the same objects, which runs counter the presently held belief that
these are different objects, residing in different environments (RLQs in elliptical galaxies
and rich clusters, RQQs in disk galaxies and poor clusters). However, first, this is not a
firmly established fact since the statistics are poor and the earthbound observations have
too marginal a resolution to unambiguously reveal the morphologies of the host galaxies.
The uncertainties in the properties of the host galaxies have been studied 13, reaching the
1 3
conclusion that, although they favor the hypothesis that RLQs are in elliptical hosts and
RQQs in disks, the uncertainties are considerable. A second consideration comes from the
fact that the meager observational data are obtained at z < 0.6 and one has no information
about the morphologies of the host galaxies of RLQs and RQQs at higher redshifts.
Third, recent spectroscopic and morphological observations coming from an HST key
project carried out by Butcher, Dressler, Oemler and Gunn (reported in 14) of 2 clusters at
z = 0.4 find a much greater proportion of disk galaxies in those clusters than there are at
low z. They also find that they often appear a bit disturbed or irregular in morphology.
Dressler concludes that some mechanism, presumably interactions and mergers, is
responsible for the disappearance of the spiral galaxies in clusters since z = 0.6. This is
relevant to this work since it allows us to assume that disk galaxies hosting RQQs at high z
have changed morphology to ellipticals at low z. In conclusion, the observational evidence
therefore allows us two assumptions: First that at z 2 any Hubble type can host RQQs as
well as RLQs; even though RLQs may favor ellipticals and RQQs disk galaxies at low
redshifts. One may of course wonder why it should be so; one can only speculate that this
is due to a younger universe and younger environments. Second, the HST observations
also allow another plausible assumption; that the host spirals at high z have now become
ellipticals. The HST observations makes this assumption the most plausible one.
Until recently, it was generally believed that the hosts of BLLs were exclusively
elliptical galaxies. However, there is now conclusive evidence that at least some BLLs are
in disk galaxies. For example, Abraham, McHardy, & Crawford 15 have carried out an
optical imaging survey of BLL host galaxies, concluding that out of 6 classifiable hosts 2
are disk systems. One may be skeptical of this classification since there have been
previous controversies regarding the nature of the hosts of some BLL (e.g. 16 , and
references therein). However, an unpublished high-quality HST image of the BLL
PKS1413+135 (McHardy, private communication) unambiguously shows it to be a in a
disk galaxy. As 15 points out, the discovery of disk hosts is in disagreement with the
1 4
standard model that assumes that all BLLs are hosted in elliptical galaxies . This is
however allowed by our hypothesis.
There have been proposals to relate BLLs to RLQs (2, 17 ). With respect to our
hypothesis, these have the advantage of not having to assume a snuffing mechanism since
they relate only radio-loud objects. However, because the ratio RQQ/RLQ is about 10 at z
=2, it is more difficult to reproduce the observed space density of BLLs at low z and
obtain the sort of good agreement with observations seen in Figures 1 and 2. It is of
course possible to reduce the gap by assuming parent quasars having Mv >-22 , by
decreasing the relativistic g factor or by taking the lower limits to the space densities of
BLLs from 2. However, this fine tuning must be contrasted with the agreement that we
obtained with more reasonable estimates. Finally, the assumption that RLQs are
exclusively in elliptical galaxies and are the progenitors of BLLs is in conflict with the
presence of some BLLs in disk galaxies.
There obviously are details that our basic hypothesis does not explain. For
example, one may want to explain the relation between OVV quasars (radio quasars with
flat radio spectra and highly polarized QSOs) and BLLs and their spectral differences (e.g.
18, 19 ). This is beyond the scope of this work, but it is reasonable to assume that the
spectral differences are due to different evolutionary ages or environments . Assuming that
BL Lacs are more evolved objects than quasars, it is not surprising that there are spectral
and morphological differences between them and similar objects they may be evolutionary
related to. It is the reverse that would be surprising. In particular, our hypothesis is not in
conflict with the assumption that the parent population of BL Lacs is a subset of FR I. We
can assume that although strong and young RLQs are FR II, the weaker and more evolved
BL Lacs are FRI. Perhaps it is power that decides whether an object is FR I or FR II.
Finally, it is interesting to notice that the leaky-shell, leaky-torus model may also
explain why BLLs have very weak broad emission lines. If the shell contains enough
absorbing material (e.g., dust), and if it surrounds the broad-line region, it can greatly
1 5
absorb the flux from the region that forms the broad lines. Although the jet punches a hole
through the shell, the hole may be large enough to let the jet through but too small to allow
a significant flux from the emission-line region to give strong broad emission lines. In a
classical QSO, the viewing angle would be such that the observer sees the nucleus and
broad-line region through a large hole or at a small angle with respect to the axis of
symmetry of the torus. Old age may explain why narrow lines are weak. While a young
QSO should have a healthy narrow-line region, since it is in a young galactic environment
with plenty of gas, a BLL does not have a strong narrow line region since gas is depleted
in its older environment.
5. CONCLUSION
We have reexamined, after ten years of advances in AGN research, the hypothesis
advanced by Borra 3 that BL Lacertae Objects are the beamed remnants of Quasi Stellar
Objects. This hypothesis explains why BLLs do not undergo the strong evolution seen in
other AGNs. It naturally accounts for the observations of 2 that show that the space
density of BL Lacertae objects increases with cosmic time; in the opposite sense of the
evolution of other Active Galactic Nuclei. Constrained by the luminosity function of QSOs
at z= 2 and the requirement that their remnants at z < 1 be faint, we find that a power-law
luminosity function that is relativistically beamed can give redshift and magnitude counts
compatible with the observations. Our successful model has free parameters that were
adjusted to yield a good fit to the observations, and the solution obtained may not be
unique. On the other hand, the values used for these free parameters are all reasonable and
compatible with the observations.
To obtain agreement between the observed space densities, we make the additional
assumption that all QSOs, radio-quiet as well as radio-loud, are capable of generating jets,
1 6
but that jets are snuffed in the young RQQs and only emerge in old age, thus implicitly
assuming a unification of RLQs and RQQs.
The meager existing statistics of the classification of the host galaxies of BLLs seem
to indicate that they favor elliptical (60%) rather than disk galaxies (30%). Prima facie, one
would have to conclude that this is only allowed by our hypothesis either if elliptical can
form RQQs as well as RLQs at z 2 or if a large fraction of QSO-hosting spirals can
evolve into elliptical. Recent HST observations seem to favor the later.
Although this paper does not prove that BLLs are quasar remnants, it shows that
the hypothesis is compatible with our present knowledge of the statistics of QSOs and BL
Lac objects, its outstanding success being its natural prediction that the space density of
BLLs should increase with cosmic time, and that it should be taken seriously. It does not
say anything regarding the unbeamed component of BL Lac objects, except that they
should be optically inconspicuous objects and have FR I morphology.
Aknowledgments
1 7
This research was supported by the Natural Sciences and Engineering Research
council of Canada and the FCAR program of the province of Québec. I wish to thank Dr.
I. McHardy for sending me an unpublished HST image of PKS1413+135.
1 8
REFERENCES
1 Blandford, R., & Rees, M.J., 1978, in Proc. Pittsburgh Conf. on BL Lac Objects, ed.
A.N. Wolfe (Pittsburgh: University of Pittsburgh Press), p. 328
2 Morris, S.L., Stocke, J.T., Gioia, I.M, Schild, R.E., Wolter, A., Maccacaro, T., &
Della Ceca, R. 1991, ApJ 389, 49
3 Borra E.F, 1983 ApJ 273, L55
4 Urry, C. M., & Shafer, R.A. 1984, ApJ 280,569
5 Hartwick, F.D.A, & Schade, D. 1990, ARA&A 28, 437
6 Padovani, P., & Urry, C.M. 1991 ApJ 368, 373
7 Urry, C. M., & Padovani, P.1991, ApJ 371, 60
8 Condon, J.J., O'Dell, S.L., Puschell, J.J., & Stein, W.A.. 1981, ApJ. 246, 624
9 Strittmatter, P.A., Hill, P., Pauliny-Toth, I.I.K., Stepe, H., & Witzel, A. 1980, A&A
88, L12
10 Kellerman, K.I., Sramek, R., Schmidt, M., Shaffer, D.B., & Green, R. 1989, AJ 98,
1195
11 Antonucci, R., Barvainis, R. & Alloin, D. 1990, ApJ 353,416
12 Visnovsky, K.L., Impey, C.D., Foltz, C.B., Hewett, P.C., Weymann, R.J., &
Morris, S.L. 1992, ApJ 391,560
13 Abraham, R.G., Crawford, C.S., & McHardy, I.M. 1992, ApJ 401, 474
14 Dressler, A. 1992 preprint of an invited talk at the Milan conference on observational
cosmology.
15 Abraham, R.G., McHardy, I.M., &Crawford, C.S. 1991, MNRAS 252, 482
16 Romanishin, W. 1992, ApJ 401, L65
17 Vagnetti, F., Giallongo, E. &Cavaliere, A. 1991, ApJ 368, 366
18 Guilbert, P.W. Fabian, A.C. & McCray, R. 1983, ApJ 266, 466
19 Worrall, D.M. & Wilkes, B.J. 1990, ApJ 360, 396
1 9
FIGURE CAPTIONS
2 0
r
e
b
m
u
N
1 0
8
6
4
2
0
0
0 . 2
0 . 4
0 . 6
0 . 8
1
Redshift
Figure 1: The dashed line gives the redshift distribution obtained from the relativistically
beamed model described in the text. The counts are in bins of D z = 0.05. The histogram
shows the observations of 2.
2 1
1
-
g
a
m
2
-
g
e
d
)
V
<
N
(
g
o
L
- 1
- 1 . 5
- 2
- 2 . 5
- 3
- 3 . 5
- 4
- 4 . 5
1 2
1 4
1 6
1 8
2 0
2 2
2 4
2 6
V
Figure 2: Integrated counts (per square degree) from the relativistically beamed model
described in the text. The observed counts are from 6; and the symbols have the same
meaning as in that reference. The arrows indicate upper or lower limits.
|
astro-ph/9704202 | 1 | 9704 | 1997-04-21T15:17:14 | Using the Fundamental Plane to Estimate the Total Binding Mass in A2626 | [
"astro-ph"
] | We use fundamental plane (FP) distance estimates to the components of the double cluster A2626 (cz~17,500 km/s) to constrain cluster kinematics and estimate total binding mass. The FP coefficients for a sample of 24 early type and S0 cluster members (alpha=1.30+/-0.36 and beta=0.31+/-0.06) are consistent with others reported in the literature. We examine the Mg_b distributions within both subclusters and find them to be indistinguishable. Lacking evidence for stellar population differences, we interpret the FP zeropoint offset (\log(D_B/D_A)=-0.037+/-0.046, where D_{cl} is distance to subcluster cl) as a measure of the distance difference. This measurement is consistent with the subclusters being at the same distance, and it rules out the Hubble flow hypothesis (distances proportional to velocity) with 99% confidence; analysis of the subcluster galaxy magnitude distributions rules out Hubble flow at 93% confidence. Both results favor a kinematic model where the subclusters are bound and infalling. We estimate the total cluster binding mass by modelling the subcluster merger as radial infall. The minimum possible total binding mass is 1.65 times higher than the sum of the standard virial masses, a difference statistically significant at the ~3sigma level. We discuss explanations for the inconsistency including (1) biases in the standard virial mass estimator, (2) biases in our radial infall mass estimate, and (3) mass beyond the virialized cluster region; if the standard virial mass is significantly in error, the cluster has an unusually high mass to light ratio (~1000h). Because observational signatures of departures from radial infall are absent, we explore the implications of mass beyond the virialized, core regions. (abridged) | astro-ph | astro-ph |
USING FUNDAMENTAL PLANE DISTANCES to ESTIMATE
the TOTAL BINDING MASS in ABELL 26261
Departments of Physics and Astronomy, University of Michigan, Ann Arbor, MI 48109
Joseph J. Mohr
Department of Physics and Astronomy, Dartmouth College, Hanover, NH 03755
Gary Wegner
Accepted for publication in The Astronomical Journal
ABSTRACT
We use fundamental plane (FP) distance estimates to the components of the double
cluster A2626 (cz ∼ 17, 500 km/s) to constrain cluster kinematics and estimate total
binding mass. We employ deep R band CCD photometry, multi–object spectroscopy,
and software designed to account for seeing effects to measure the FP parameters Re,
σ, and hµei for 24 known early type and S0 cluster members. The FP coefficients from
this sample (α = 1.30 ± 0.36 and β = 0.31 ± 0.06) are consistent with others reported
in the literature.
We examine the Mgb equivalent width distributions within both subclusters and
find them to be indistinguishable. Lacking evidence for stellar population differences,
we interpret the FP zeropoint offset between the two subclusters as a measure of the
distance difference. We find log(DB/DA) = −0.037 ± 0.046, where Dcl is the distance
to subcluster cl. This measurement is consistent with the subclusters being at the
same distance, and it rules out the Hubble flow hypothesis (distances proportional to
velocity) with 99% confidence; analysis of the subcluster galaxy magnitude distributions
rules out Hubble flow at 93% confidence. Both results favor a kinematic model where
the subclusters are bound and infalling.
We estimate the total cluster binding mass by modelling the subcluster merger as
radial infall. The projected separation, the line of sight velocity difference and the line
of sight separation constrain the cluster mass; the minimum possible total binding mass
is 1.65 times higher than the sum of the standard virial masses, a difference statistically
significant at the ∼ 3σ level. We discuss explanations for the inconsistency including
(1) biases in the standard virial mass estimator, (2) biases in our radial infall mass
estimate, and (3) mass beyond the virialized cluster region; if the standard virial mass
is significantly in error, the cluster has an unusually high mass–to–light ratio (∼ 1000h).
Because observational signatures of departures from radial infall are absent, we explore
the implications of mass beyond the virialized, core regions.
1Observations reported here were obtained at the MDM Observatory, a facility jointly operated by the University of Michigan, Dartmouth
College, and the Massachusetts Institute of Technology.
1
1.
Introduction
Galaxy cluster masses are clearly of cosmological
significance, and examples abound: (1) the differences
in cluster evolution as a function of mass provide con-
straints on the power spectrum (Lacey & Cole 1993,
1994), (2) comparisons of the amounts of luminous
and non–luminous matter in clusters provide clues
to the nature of dark matter and the efficiency of
galaxy formation (David, Jones & Forman 1995), and
(3) the cluster mass–to–light ratio and baryon frac-
tion constrain the cosmological density parameter Ω0
(e.g. Ramella, Geller & Huchra 1989, White et al.
1993, Mohr et al. 1996).
Cluster mass studies tend to focus on the central
regions where virial equilibrium is a more accurate
approximation, and the weak lensing signals are eas-
iest to detect. Here we apply a method to measure
the mass beyond the virialized region in the double
cluster A2626. We utilize recent improvements in the
understanding of the fundamental plane (FP) to con-
strain the total cluster binding mass by revisiting the
subcluster radial infall model (Beers, Geller & Huchra
1982). Specifically, measuring FP distances (with
an accuracy of 15% to 25%) to a reasonably large
sample of cluster galaxies can yield a very accurate
(sub)cluster distance (Lucey & Carter 1988, Lynden–
Bell et al. 1988). For two bound subclusters, the
line of sight separation, the line of sight velocity dif-
ference, and the projected separation provide enough
information to estimate the total cluster binding mass
in cases where a radial infall model is appropriate.
Although A2626 is rather distant, it is well suited
for this analysis because simply demonstrating that
the two subclusters are bound and infalling has inter-
esting implications. A straightforward application of
the radial infall model, using the (rest frame) line of
sight velocity difference and the virial mass estimates
within each subcluster, rules out bound and infalling
merger models (Mohr, Geller & Wegner 1996; here-
after MGW96); the virial masses are not large enough
to produce the apparent velocity difference. Thus, FP
distances used to determine whether or not the sub-
clusters in A2626 are bound and infalling serve as an
independent test of the virial estimators, or alterna-
tively, as a means of measuring the mass beyond the
virialized region of the cluster.
Possible pitfalls to this approach include (1) dis-
tance biasing FP zeropoint (or shape) differences be-
tween subclusters, (2) deviations from radial infall,
2
and (3) significant interactions between the virial-
ized regions of the two subclusters. FP variation
with cluster environment is an area of active research
(e.g. Guzman et al. 1992, Worthey, Trager, & Faber
1996, Jørgensen, Franx & Kjaergaard 1996; hereafter
JFK96); current indications are that distance biases
are smaller than 5% for ellipticals within clusters and
are related to observable variations in the stellar pop-
ulations within cluster galaxies. As discussed below,
subcluster pairs suffering from pitfall numbers 2 and 3
can be identified through corresponding observational
signatures. We argue that the radial infall model and
FP distances provide a promising method for deter-
mining cluster masses on scales larger than the viri-
alized region in low redshift clusters; if coupled with
estimates of cluster light at larger radii, these mass
measures should suffice to determine whether mass
to light ratios vary significantly between the virial-
ized core and the surrounding infall region. For higher
redshift cluster pairs, our approach could be used to
quantify the level of environmentally induced funda-
mental plane zeropoint variations.
We describe the double cluster A2626, the FP ob-
In Section 3,
servations, and the reductions in §2.
we derive FP coefficients, discuss constraints on the
stellar populations in each subcluster, and derive con-
straints on their relative distances. Section 4 con-
tains a discussion of the radial infall model applied
to A2626. Conclusions are summarized in Section 5.
Throughout the paper we use H0 = 100h km/s/Mpc.
2. Galaxy Sample and Data
Below we describe Abell 2626 (Abell 1958), sum-
marize previous observational results detailed else-
where (MGW96), present the new observations, and
describe the reductions and analysis in detail.
2.1. Abell 2626
A redshift survey with the Decaspec (Fabricant &
Hertz 1990) mounted on the MDM Hiltner 2.4 m tele-
scope in Fall 1993 and 1994 revealed that Abell 2626
(hczi ∼ 17, 500 km/s, richness class 0) is composed of
at least two systems with mean line of sight velocities
which differ by ∼ 2, 500 km/s (MGW96). As is clear
from Fig. 1 and Table 1, there are two main compo-
nents with cluster–like dispersions: group A centered
at a velocity of hczi = 16, 533 ± 141 km/s and group
B centered at a velocity of hczi = 19, 164 ± 138 km/s
(90% statistical confidence limits). As discussed in
MGW96, the 11 galaxies centered at hczi = 21, 173 ±
119 km/s are most likely part of a background, low
density structure because (1) the dispersion is low
(σ = 200+119
−52 km/s) and (2) a large fraction of these
galaxies are gas rich- 73% have emission lines com-
pared to 39% (33%) for group A (B). If the velocity
of the low density structure corresponds to Hubble
flow then its distance from groups A and B is from
20h−1 Mpc to 50h−1Mpc.
The group A and B masses, velocity dispersions,
central densities and X–ray properties all differ. As
noted in Table 1, the virial mass of subgroup A is
roughly twice that of B, and the line of sight veloc-
ity dispersions are σA = 658+111
−81 km/s and σB =
415+117
−72 km/s (90% statistical confidence limits). The
absence of observed X–ray emission from group B can
be used to place a limit on the central gas density
in group B (MGW96).
In particular, the Einstein
imaging proportional counter (IPC) image of the re-
gion reaches roughly 50 times fainter than the peak
in the X–ray emission from group A. The group A
central gas density is ∼ 3× 10−3 cm−3, so the central
gas density in group B must be ≤ 5 × 10−4 cm−3.
Although the contrast in the projected galaxy densi-
ties of groups A and B is difficult to quantify, the
smoothed distribution (Figure 3 in MGW96) indi-
cates that the central density in group B is ∼ 7 times
less than the central density in group A. Thus, the ra-
tio of gas to galaxy density may be the same in both
groups.
As discussed below, if groups A and B are on a ra-
dial infall trajectory, then the minimum implied grav-
itational mass is more than the sum of their virial
masses. However, it is also possible that the velocity
difference is just due to Hubble flow. The magni-
tude distributions of the confirmed members of both
groups indicate that the merger hypothesis is favored
over Hubble flow at 93% confidence (MGW96). To
further investigate this issue we measure fundamen-
tal plane distances to each cluster and determine the
line of sight distance difference directly. To do so
we first use R band CCD photometry to identify a
sample of early type galaxies with redshift confirmed
membership.
2.2. Photometry
On November 21, 1995, we used a thinned, Tek-
tronix 10242 CCD mounted on the MDM Hiltner
2.4 m telescope to image galaxies in A2626 and J8
(Jackson 1982). The 24µ pixels are 0.275 ′′ on a side.
The seeing varied during the night between 0.85 ′′ and
1.05 ′′. Because the night was non–photometric, we
determine the photometric zeropoint of each R band
image externally. For the A2626 galaxies we use pho-
tometric images taken previously (MGW96) on the
MDM 1.3 m to zeropoint our images. Specifically, we
use aperture photometry of isolated stars in the 2.4 m
and 1.3 m frames to determine the 2.4 m zeropoint.
The 1.3 m images are reduced to the Kron–Cousins
system using Landolt (1992) standards. For J8 we
rely primarily on published photoelectric galaxy aper-
ture photometry (Colless et al. 1993) to determine
the R band zeropoint but also use stellar aperture
photometry from a photometric 1.3 m image for one
galaxy (Saglia et al. 1997). In addition to our primary
imaging run, we acquired additional Hiltner 2.4m im-
ages through the generosity of Paul Schechter; the
reductions for these data were similar.
The R band galactic extinction is 0.062 mag toward
A2626 (NASA/IPAC Extragalactic Database, Savage
& Mathis 1979) and 0.185 mag toward J8 (Saglia
et al. 1997). We approximate the k correction as
kR = zg where zg is the galaxy redshift (Frei & Gunn
1994). We account for the peculiar velocity compo-
nent of the line of sight velocity when applying the
cosmological dimming correction; specifically, two of
the factors of (1 + z) in the cosmological dimming
are due to relativistic effects and two are due to the
change in the geometry of the universe between emis-
sion and detection. Writing zg = zd + zp, where zd
and zp are the Hubble flow and peculiar velocity com-
ponents of the redshift, the cosmological dimming is
written C = +5 log{1 + zd} + 5 log{1 + zg}; in terms
of the observed and peculiar redshift, the correction
is C = +10log{(1 + zg)p1 − zp/(1 + zg)}. Thus, C
is model dependent. As an example, if groups A and
B are at the same distance and the peculiar velocity
correction were incorrectly taken to be zp = 0, the
fundamental plane distances would be biased so that
group B would appear closer than group A by ∼ 1%.
We bias subtract, flat field (using twilight flats),
and clean the images using standard IRAF tasks. For
each galaxy, we calculate sky subtracted radial pro-
files and uncertainties. Uncertainties have a Pois-
son component added in quadrature with an assumed
1% flat fielding uncertainty. We fit these profiles to
summed R1/4laws and exponential disks,
Ib(R) = Ib0e−7.67[(R/Rb)1/4−1]
(2-1)
3
Id(R) = Id0e−R/Rd
where Ib0 (Rb) and Id0 (Rd) are the characteristic
surface brightnesses (scale lengths). Our fitting pro-
cedure accounts for the effects of PSF smoothing and
pixel extent as described in Saglia et al. 1993b.
More specifically, we determine the parameters of
the R1/4and disk components by minimizing the χ2
difference between the seeing convolved model and
the observed profile. The expected surface brightness
I(R, dR) in the bin of width 2dR and radius R is
I(R, dR) =
[F (R + dR) − F (R − dR)]
4πRdR
(2-2)
where F (R) is the seeing convolved integral flux
within radius R. The integral flux follows from the
bulge, disk and PSF parameters. Specifically,
0
F (R) = RZ ∞
dkJ1(kR)p(k)(cid:16) Ib(k) + Id(k)(cid:17) (2-3)
where J1 is a Bessel function, and p(k), Ib(k), and
Id(k) are the Fourier transforms of the PSF, the bulge,
and the exponential disk. Following Saglia et al.
1993b, we use a PSF of the form
p(k) = e−(kb)γ
(2-4)
and find it to be a good fit to stars on the 2.4 m images
with 1.4 ≤ γ ≤ 1.6. We use the analytic approxima-
tion to the Fourier transform of the R1/4law (Saglia
et al. 1993b), and the exact form of the transform of
the disk component (e.g. Bracewell 1986).
The fitting procedure has three steps:
• Fit the PSF parameters b and γ using radial
profiles of ∼ 4 stars within each 2.4 m image;
best fit parameters minimize the χ2 difference
between the observed and model profile.
• Feed initial guesses for the bulge and disk pa-
rameters into a simplex minimization routine
(Press et al. 1988) which minimizes the χ2 dif-
ference between the seeing convolved theoretical
and observed profiles. At each iteration the full
Fourier integral is solved for each point in the
radial profile; with an HP–735 the minimization
takes several minutes.
In addition to the un-
certainties in the observed profile, a theoretical
uncertainty is introduced to account for imper-
fections in the smoothing corrections (e.g. the
exact form of the PSF). Specifically, the uncer-
tainty for each profile point has a component
which is proportional to the fractional effect of
the smoothing on the model profile. This has
the positive effect of making the fit results less
sensitive to the inclusion/exclusion of the heav-
ily smoothed inner points in the profile and to
the details of the smoothing operation. We in-
clude a theoretical uncertainty which is 10% of
the fractional effect of the smoothing on the
model profile. This is consistent with a phi-
losophy that we can correct for seeing effects at
the 90% level.
• We fit a bulge only model to the profile and
compare the minimum χ2 for this model to the
bulge plus disk χ2. For galaxies where χ2 is only
minimally improved by the inclusion of the disk
component we use the bulge only fit. Once the
fitting is complete, we use the best fit parame-
ters to calculate Re– the half luminosity radius,
µe– the surface brightness at Re, and hµei– the
mean surface brightness within Re.
We test our fitting routines by (1) fitting artificial
galaxy images, (2) comparing results from multiple
images of the same galaxy, and (3) comparing our re-
sults to those of Saglia et al. (1997) in the cluster
J8. We create artificial images by adding a galaxy
described by an R1/4 profile to a flat background of
103 cts/pixel. We then smooth the artificial image
and introduce Poisson noise. By using delta func-
tions instead of R1/4 profiles we produce stars which
are then used to measure the PSF parameters b and
γ. The process of fitting artificial galaxy images tests
the accuracy of our fitting in that case where the
true galaxy profile is well described by an R1/4 pro-
file. We find that the parameters Re and hµei are
constrained to an accuracy of ∆ log(Re) ∼ 0.02 and
∆hµei ∼ 0.07 for Re as small as 50% the FWHM
of the PSF. As expected, the accuracy of the com-
bination log(Re) − 0.33 ∗ hµei (which appears in the
equation describing the fundamental plane) is con-
strained to ∼ 0.01 because the errors in Re and hµei
are correlated.
We also compare results from subarcsecond seeing
images of seven A2626 galaxies to results from poorer
quality images. The mean offset between the param-
eters in the good and poor seeing images of the same
galaxies are consistent with zero, and the scatter is
small. Specifically, ∆hµei = −0.0262 ± 0.0311 with
4
an RMS of 0.0823, ∆ log(Re) = −0.0073±0.0074 with
an RMS of 0.0196, and ∆ (log(Re) − 0.33 ∗ hµei) =
0.0013 ± 0.0037 with an RMS of 0.0097.
We observed five galaxies in the EFAR cluster J8
(Wegner et al. 1996) for which we have photometric
zeropoints. These observations provide the strongest
test of our methods because the data, reductions and
analyses differ significantly. There are no significant
offsets between our values and EFAR values (Saglia et
al. 1997); specifically, ∆hµei = −0.0766±0.1223 with
an RMS of 0.2735, ∆ log(Re) = −0.0257±0.0326 with
an RMS of 0.0729, and ∆ (log(Re) − 0.33 ∗ hµei) =
−0.0006±0.0112 with an RMS of 0.0249. These three
tests of our photometric methods indicate that our
procedure is accurate and makes only a small con-
tribution to the observed scatter in the fundamental
plane (see below).
2.3. Spectroscopy
On November 22 and 23, 1995, we used the De-
caspec (Fabricant & Hertz 1990) plus MkIII spectro-
graph mounted on the MDM Hiltner 2.4 m telescope
to obtain spectra of four galaxy fields in Abell 2626,
one field in J8 (Wegner et al. 1996), and multiple
spectra of five HD stellar velocity standards. The De-
caspec fibers subtend 2.3′′ in the focal plane. We used
a grism blazed at 5,700 A with 600 l/mm, yielding a
typical PSF of 5.3 A at 2.18 A per pixel and coverage
from 4,300 A to 6,500A. Four or five thirty minute
exposures were obtained for each of the A2626 fields;
shorter exposures were obtained for the brighter J8
field. We extracted the spectra, using arc lamp ex-
posures taken on either side of each object exposure
to dispersion correct. We then combined the spec-
tra from the four to five exposures, excluding cosmic
rays through sigma clipping of large, positive devia-
tions. With the Decaspec there are four sky spectra
for each object spectrum. We combined the sky spec-
tra for each object, scaled the sky spectrum using the
flux in the 5577 A sky line, and then removed the
sky contribution from the object spectrum. The final
spectra range in signal to noise (S/N) from S/N=20
to 75 per pixel at 5,300 A.
We obtained spectra of more galaxies, stellar ve-
locity and Lick linestrength standards (Worthey et
al. 1994) using the Modspec mounted on the MDM
Hiltner 2.4m telescope in Nov and Dec '96. We used
a 1,200 l/mm grating, the 200 mm camera, a thinned
CCD with 24 µm pixels, and a 1.7′′ wide slit; stan-
dard longslit reductions led to spectra with 2 A reso-
lution, 1 A/pixel sampling, and coverage from 4,800A
to 5,800A. Galaxy exposure times ranged from 20 min
to 2.5 hrs, and the spectra have S/N≥ 30 per pixel.
The higher resolution of this setup allowed us to accu-
rately measure galaxy dispersions as low as 100 km/s.
For each run, galaxy dispersions and velocities are
extracted using the stellar templates from that run
and the cross correlation peak fitting available in fx-
cor (IRAF.RV); we use the approximate wavelength
region 4,800–5,800A which contains the strong Mg
features. The correlation peak width is transformed
to an intrinsic galaxy dispersion using translation ta-
bles created for each stellar template; the transla-
tion tables are produced by cross correlating the stel-
lar templates against Gaussian broadened versions of
themselves. The final galaxy dispersion is the mean
of the dispersions from each stellar template in the
run. Finally, an aperture correction
log(cid:18) σcor
σobs(cid:19) = 0.038 log(cid:18) rap
rn
ap
θn
e
θe(cid:19)
(2-5)
is applied. This aperture correction accounts for the
falling dispersion profile in early type galaxies; we
normalize to an aperture of radius rn
ap = 1.7′′ and a
galaxy with θn
e = 20′′. For the longslit spectra we use
rap = 1.025pxy/π, where x and y are the width and
length of the rectangular aperture (Jørgensen, Franx
& Kjaergaard 1995, Baggley 1996).
We estimate the uncertainties in our dispersions
using Monte Carlo techniques. For each galaxy spec-
trum we broaden a stellar template to mimic the mea-
sured dispersion, scale the template to the observed
cts/pix in the region around 5,200 A, and add the sky
at the observed level. We sample this artificial galaxy
spectrum 100 times,
introducing the Poisson noise
and sky subtracting, and then measure the dispersion.
We use the RMS of the measured dispersions around
the input value as a measure of the velocity dispersion
uncertainty (see Table 2). Not surprisingly, the sim-
ulations indicate that averaging the dispersions mea-
sured by cross correlating against multiple templates
does not improve the accuracy. However, these sim-
ulations consider only Poisson noise effects; we aver-
age the multiple dispersion measurements from each
galaxy spectrum to reduce template mismatch sys-
tematics.
We have multiple spectra for 21 galaxies. The dis-
tribution of dispersion differences scaled by the un-
certainties for the 28 comparison pairs has an RMS
of 0.97, indicating that the Monte Carlo uncertain-
5
ties are a reasonably good estimate of our true dis-
persion uncertainties. The mean scaled difference
(h(σMod − σDeca)/ǫσi where ǫσ is the uncertainty in
the difference) between Modspec and Decaspec mea-
surements for 23 comparison pairs has a mean of
−0.11 ± 0.19 with an RMS scatter about this mean
of 0.89; the variance weighted, mean logarithmic dif-
ference between Modspec and Decaspec dispersions
is −0.001 ± 0.013. Because there is no evidence for
a systematic difference between dispersions measured
with the Decaspec and the Modspec, we apply no cor-
rection. We note that the Modspec was used in mea-
suring dispersions for 10 of the 16 galaxies in group A
and 5 of the 8 galaxies in group B; so any systematic
offset between Decaspec and Modspec measurements
would add to the FP scatter within each group, but
would not bias the estimate of the distance difference.
Table 2 contains a list of the galaxies and their
properties; for the galaxies with multiple measure-
ments, we use the variance weighted average disper-
sion. The columns of Table 2 contain the galaxy tag
(first letter corresponds to group membership), the
coordinates, the redshift cz, the velocity dispersion σ
and uncertainty ǫσ, the R band apparent magnitude
MR, the mean surface brightness hµei within the half
light radius, the half light radius θe, the M gb equiva-
lent width, and the bulge luminosity fraction FB.
For 6 galaxies in J8 we compare our measured dis-
persions with those of Saglia et al. (1997). There is
evidence of an offset; the distribution of (σEF AR −
σobs)/ǫσ has a mean of −0.80 ± 0.31 (where ǫσ is the
uncertainty in the dispersion difference). The aver-
age EFAR velocity dispersion is 15 ± 7 km/s lower
than ours. This offset does not affect our estimates
of the relative distances to the two subclusters, but
would complicate efforts to bring our distances onto
the EFAR system.
We place constraints on possible differences in the
stellar populations of the two subclusters by using
IRAF scripts to measure M gb linestrengths in each
galaxy spectrum. The rest wavelengths given in
Burstein et al. (1984) for the feature and continuum
bands are corrected using the redshifts in Table 2. We
broaden our spectra to the nominal 8.6 A resolution
of the Lick system (Worthey & Ottaviani 1997) be-
fore making measurements. We tranform to the Lick
system using an expression of the form
M gLICK
b
= af (σ)M gobs
b + cap,
(2-6)
where a is a scale factor, f (σ) is a velocity disper-
6
sion correction we determine by broadening our Lick
standards, and cap is an aperture correction of the
same form as in Eq. 2-5 but with a coefficient of
0.050 instead of 0.038. We determine a separately
for each run; observations of Lick standards indicate
a = 1.060±0.016 for the Nov '96 and a = 1.086±0.028
for the Dec '96 data. We use overlapping galaxy
observations to measure a = 1.066 ± 0.034 for the
Nov '95 data. The aperture correction is similar to
that used by Jørgensen (1997), but generalized to the
form used for the velocity dispersions above (Baggley
1996).
We test the accuracy of our M gb measurements us-
ing multiple galaxy observations. Twenty three com-
parison sets from 18 different galaxies yield an RMS
of 0.051 in ∆ log M gb. Thus, we estimate the uncer-
tainty of a single observation is approximately 9%.
The linestrengths listed in Table 2 are averages of
multiple observations where appropriate. Finally, we
compare our linestrengths to preliminary EFAR M gb
linestrengths in 6 J8 cluster galaxies; the EFAR values
are lower: ∆ log M gb = −0.0733± 0.014 with an RMS
of 0.039. The linestrengths in Table 2 are corrected
to the EFAR system.
3. Fundamental Plane Analysis
In this section we, determine the coefficients and
zeropoint of the FP within the two clusters (§3.1),
discuss the evidence for stellar population variations
(§3.2), and then interpret the observed FP zeropoint
offset (§3.3).
3.1. Determining FP Coefficients
Even if the assumption that elliptical and S0 galax-
ies within both clusters are similarly distributed within
the FP is valid, the FP coefficients and cluster dis-
tances can be biased by differences in selection (e.g.
Lynden–Bell et al. 1988, Baggley 1996). Our galaxy
sample is drawn from a list of known elliptical and
S0 cluster members sorted by central aperture mag-
nitude (aperture is 2′′× 2′′ square); the parameters of
our final sample indicate that central aperture mag-
nitude is correlated with log θe − 0.3 hµei, a quantity
similar to the combination which appears in the FP.
Thus, our selection is similar to selection by isopho-
tal diameter Dn (Dressler et al. 1987). We seek to
minimally bias the coefficients by (1) combining the
samples from both clusters into a single FP fit and
(2) weighting galaxies within each cluster by the sam-
ple completeness. Finally, we examine the variation
of the zeropoint offset between the two clusters as a
function of variations in the FP coefficients (§3.3).
Our galaxy sample extends over 1.5 mag in cen-
tral aperture magnitude. We divide this range into
four bins and calculate the completeness within each
separately for the two clusters. The completeness in
each bin is simply the number of galaxies within our
final FP sample divided by the estimated total popu-
lation of elliptical and S0 cluster members. We esti-
mate the total cluster population within each bin to
be Ncl = Ntot ∗ fcl, where Ntot is the total number
of elliptical and S0 galaxies identified in the R band
survey of the cluster, and fcl is the fraction of all el-
liptical and S0 galaxies with redshifts that lie within
the velocity range of cluster cl.
Weighting by the inverse of the completeness ap-
pears to be ill advised for the faintest galaxies within
each cluster. For these two galaxies the completeness
is small enough (weight is 13.2 (17) for the faintest in
group A (B)), that weighting by the incompleteness
would lead to them dominating the fit. We exclude
the faintest galaxy from each sample when determin-
ing the FP coefficients, but include them when cal-
culating the best estimate of the cluster zeropoint.
Finally, we normalize the weights within each cluster
so that the ratio of weighting for cluster A:B is 15:7,
the ratio of the number of galaxies in each.
We fit the galactic properties to a FP of the form
log θe = α log σ + β hµei + γ, where α and β are
the (cluster invariant) coefficients of the plane and
γ is a distance dependent zeropoint (Faber et al.
1987, Djorgovski & Davis 1987, Jørgensen, Franx
& Kjaergaard 1993, JFK96). We combine galaxies
from both groups in a single fit by using deviations
around median values of distance dependent quanti-
ties (Baggley 1996). The corrected quantities hµei
and log σ are distance independent, so we transform
log θe alone; specifically, we remove an estimate of the
zeropoint within each subcluster by expressing log θe
as the variation around the γf it which minimizes the
cluster scatter in γ
log θe → log θe − γf it
(3-1)
Note that the transformation depends on the FP co-
efficients, so it must be reapplied during each fit iter-
ation.
We determine the best fit FP coefficients by mini-
mizing the sum of the absolute value of the weighted,
orthogonal deviations from the plane (JFK96, Bag-
gley 1996). The best fit values are α = 1.30 and
β = 0.31; the scatter in log Re is 0.090 (23% distance
uncertainty). Our FP has the form
log θe = (1.30 ± 0.36) log σ + (0.31 ± 0.06)hµei + γ.
(3-2)
The uncertainties are determined using a bootstrap
resampling procedure (JFK96, Baggley 1996) to re-
sample the galaxy list and fit the FP 104 times; the
uncertainty intervals are half the width of the 68%
confidence region. For comparison, the FP coeffi-
cients in the case where all galaxies are given equal
weight are α = 1.27 and β = 0.27. Our best fit FP
coefficients are statistically consistent with the val-
ues α = 1.24 ± 0.07 and β = 0.33 ± 0.01 determined
in a sample of 207 galaxies with Gunn r photome-
try (JFK96); as discussed by JFK96, their coefficients
are consistent with previous studies (e.g. Faber et al.
1987, Djorgovski & Davis 1987, Bender, Burstein &
Faber 1992, Guzman, Lucey & Bower 1993, Saglia,
Bender & Dressler 1993, Jørgensen, Franx & Kjaer-
gaard 1993). Fig. 2 contains plots of the group A
(filled) and B (hollow) galaxies along two projections
of the FP; the lines in each figure represent the FP
for groups A (solid) and B (dashed).
3.2. Constraints on the Stellar Populations
We compare the distribution of M gb equivalent
widths within the two subclusters to place constraints
on stellar population variations which could introduce
offsets in the cluster FP zeropoints, biasing the rela-
tive distances to the two clusters (e.g. JFK96). Fig. 3
contains a plot of log (M gb) versus log σ for the galax-
ies in groups A (solid points) and B (hollow points).
The lines represent best fit models for groups A (solid)
and B (dashed).
Rather than allowing both the slope and zeropoint
to vary, we constrain the slope to be that found by
Jørgensen (1997) for a sample of ∼ 300 galaxies.
Specifically, the combined correlations of M g2 with
M gb and M g2 with σ imply
log M gb = 0.307 log σ − b,
(3-3)
where the zeropoint for the large galaxy sample is
b = 0.034 (Jørgensen 1997). By minimizing orthogo-
nal deviations from the line, we find b = 0.054± 0.016
for sample A and b = 0.042 ± 0.022 for sample B.
The zeropoint of each cluster is statistically consistent
with the zeropoint from the Jørgensen sample. We
7
can measure the offset between the two clusters some-
what more accurately; the zeropoint difference be-
tween the two clusters is ∆b = bA−bB = 0.012±0.019,
statistically consistent with no offset.
Trends in M g2 have been noted as a function of
radius from the center of the Coma cluster (Guzman
et al. 1992) and as a function of cluster velocity dis-
persion for a sample of 11 clusters (JFK96; Jørgensen
1997). Another study indicates that elliptical galax-
ies in low velocity dispersion environments have both
higher metallicities and younger stellar populations
than galaxies in higher velocity dispersion environ-
ments (Rose et al.
1994). The velocity disper-
sion difference between clusters A (σ = 658 km/s)
and B (σ = 415 km/s) should introduce an offset
of ∆b ∼ −0.016 into the M gb–σ correlation. This
expectation is inconsistent with our observations at
the 1.5σ level; we would require additional galaxy
linestrengths to resolve an offset of that magnitude.
3.3. FP Zeropoint Differences
We determine the zeropoint of each group by using
the median γ; the 16 galaxies in group A yield γA =
−8.4083 with an RMS around this value of 0.0879,
and the 8 galaxies in group B yield γB = −8.3632
with an RMS of 0.1149. The homogeneous Malmquist
bias has a marginal effect on the relative distances to
these two clusters (Lynden–Bell et al. 1988); we cal-
culate δγA = −0.0033 and δγB = −0.0114, resulting
in an increase in ∆γ of ∼ 0.0081. Thus, the zeropoint
difference between the two clusters is
∆γ = γA − γB = −0.037 ± 0.046,
(3-4)
where we have assumed the uncertainty in the zero-
point is the scatter about that value divided by √N ,
where N is the number of galaxies (note that this is
a conservative estimate of the true uncertainties be-
cause the scatter is sensitive to outliers). As discussed
in §2.2, if we assume groups A and B are at the same
distance rather than at distances proportional to their
mean redshifts, the cosmological dimming corrections
change so that ∆γ increases by ∼ 0.004.
In general, zeropoint offsets can be caused by stel-
lar population or metallicity induced M/LR differ-
ences and/or distance differences. Because there is no
compelling evidence for stellar population differences,
we use the observed zeropoint offset and uncertainty
as relative distance constraints. The quantity γ (in
Eqn. 3-2) can be written γx = log (Γ/Dx) where Dx
is the distance of galaxy x and Γ follows from the
standard form implied by the FP
Re = Γσα hIei−2.5β .
(3-5)
Assuming the intrinsic FP zeropoint log Γ is the same
for all galaxies, the relative distances to two galaxies
x and y is
log(cid:18) Dx
Dy(cid:19) = γy − γx.
(3-6)
Fig. 4 contains a plot of the distances to the galaxies
in groups A (solid) and B (hollow) relative to the me-
dian distance to group A versus the galaxy redshift.
Also marked with vertical lines are the (homogeneous
Malmquist bias corrected) median distances to groups
A (DA; solid line) and B (log (DB/DA) = −0.037;
dashed line); the horizontal lines mark the mean ve-
locities for groups A (solid) and B (dashed). The error
bars on the dashed, vertical line represents the rela-
tive distance uncertainty. The large star marks the
position of cluster B if its distance relative to cluster A
were reflective of pure Hubble flow (∆D ∝ ∆hvi); the
Hubble flow hypothesis implies log (DB/DA) = 0.065.
The zeropoint difference is consistent with zero,
and it is sufficient to rule out Hubble flow at 2.2σ;
therefore, the FP measurements suggest that cluster
B is closer to us than cluster A (in §5.1 we return to
this issue in more detail). Naturally, the zeropoint
offset depends on the coefficients of the FP, which are
only constrained to within ∼20% of the best fit values.
Using the distribution of γA − γB from the 104 boot-
strap resampling fits, we find that in 99.05% of the
simulations the zeropoint offset between the two clus-
ters is less than the Hubble flow offset, rejecting the
Hubble flow hypothesis. This probability corresponds
to a +2.35σ deviation in a Gaussian, similar to the
2.2σ estimate above which we derived from the scat-
ter about the median distance in each cluster. The
bootstrap probability is superior because it includes
variations in the FP coefficients (and is less sensitive
to outliers than the scatter estimator).
4. Radial Infall Model as a Mass Constraint
The radial infall model has been applied to many
close galaxy cluster pairs to determine whether or not
the pairs will merge (e.g. Beers, Geller & Huchra
1982, Beers et al. 1991, Beers et al. 1992, Colless &
Dunn 1995, Scodeggio et al. 1995). In the following
we use the measured line of sight separation and this
model to estimate the total binding mass required to
8
explain the observed infall velocity. We then compare
this mass to virial estimates of the cluster binding
mass.
4.1. Radial Infall Model
For two isolated, bound objects within an expand-
ing universe, the dynamics can be parametrized in the
standard way
(4-1)
l = (cid:0)GMGB2(cid:1)1/3
B (cid:19)1/3
v = (cid:18) GMG
t = B(η − sin η)
(1 − cos η)
sin η
(1 − cos η)
where η is a development angle, l is the separation,
v is the relative velocity, t is the age of the universe,
B is an undetermined coefficient, and MG is the sum
of the individual masses (e.g. Peebles 1993). When
we observe two clusters on a merger trajectory at the
present epoch, we measure the projected separation
of the clusters ∆l⊥ and a line of sight velocity dif-
ference ∆vLOS. Modelling the collision as a radial
infall, we parametrize the merger in terms of the an-
gle φ between the merger trajectory and the line of
sight (∆vLOS = ∆v cos φ and ∆l⊥ = ∆l sin φ).
With four free parameters in this model (η, B, φ
and MG) and three observables (the age of the uni-
verse tH , the projected separation ∆l⊥ and the line
of sight velocity difference ∆vLOS), the total cluster
mass and age of the universe are a function of the an-
gle φ and the present epoch value of the development
angle ηH .
MG = 2.3 × 1014M⊙(cid:0)(1 + cos ηH ) cos2 φ sin φ(cid:1)−1
1000 km/s(cid:19)2(cid:18) ∆l⊥
1 Mpc(cid:19)
(cid:18) ∆vLOS
(4-2)
tH = 9.8 × 108 yrs (cid:18) (ηH − sin ηH ) sin ηH
cot φ(cid:19)
(1 − cos ηH )2
1 Mpc(cid:19)(cid:18) 1000 km/s
∆vLOS (cid:19)
(cid:18) ∆l⊥
Using an approximate age of the universe to deter-
mine ηH , we arrive at the total cluster mass function
MG(φ). Fortunately the dependence of MG on tH is
rather weak (see Fig. 5).
If the line of sight sepa-
ration between the subclusters ∆lLOS is measured, φ
is determined, yielding the total system mass (which
9
we term the cluster merger mass MG). MG can then
be compared to other mass estimates, which typically
probe the cores of clusters.
Departures from radial infall and the physical ex-
tent of the clusters can bias the merger mass. Gen-
erally speaking, the large mass and rarity of galaxy
clusters make them good candidates for this analysis.
We plan to study the accuracy of merger masses us-
ing numerical simulations of cluster evolution within
"realistic" environments.
Note that there are scenarios in which the merger
mass would be substantially biased; it is possible to
recognize these scenarios observationally. In the first
scenario, the two merging clusters are at small sep-
interactions among the cluster
aration (≤ 1 Mpc);
components, significant overlap of their mass profiles,
and small departures from pure radial infall combine
to bias the inferred merger velocity and the implied
merger mass. Fortunately, the signatures of cluster
mergers have been extensively studied (e.g. Geller &
Beers 1982, Dressler & Shectman 1988, Jones & For-
man 1992, Mohr, Fabricant & Geller 1993, Pearce,
Thomas & Couchman 1994, Mohr et al. 1995, Buote
& Tsai 1995), and typically there are merger clues in
the galaxy and gas distributions.
In the second scenario, a third object of compara-
ble mass to the two merging clusters is close enough
to invalidate the two body analysis. In this case, the
observed distribution of nearby galaxy clusters should
be enough to determine whether or not the radial in-
fall model is appropriate.
4.2. Merger Mass in A2626
To calculate MG we require (see Eqn. 4-2) the
observed rest frame velocity difference and the pro-
jected separation. The rest frame velocity difference
is ∆v = 2, 486 ± 112 km/s. The projected separation
between the centers of mass is assumed to be the dis-
tance between the bright, central elliptical in group A
(its position is consistent with the peak in the X–ray
emission) and the centroid of the 30 group B members
with measured redshifts (using the centroid of group
A members slightly increases the projected separa-
tion and the merger mass). At a cluster distance of
l = 175h−1 Mpc, the projected separation between
the two groups is ∆l⊥ = (0.707 ± 0.052)h−1 Mpc.
The uncertainty reflects the statistical uncertainties
associated with centroiding group B. The fractional
uncertainty in the merger mass MG due to ∆vLOS
and ∆l⊥ is 12%.
Fig. 5 displays the ratio of the cluster merger mass
MG to the sum of the virial masses of groups A and
B (MA + MB = 9.1 × 1014h−1M⊙; see Table 1 and
MGW96) versus the line of sight separation between
the subclusters, ∆lLOS = ∆l⊥/ tan φ. The dotted
line corresponds to tH = 18 Gyr and the solid line
corresponds to tH = 13 Gyr. The FP distances to
galaxies within groups A and B yield a ratio of the
group B distance to the group A distance (lB/lA)
which can be used to calculate ∆lLOS. Specifically,
we estimate the line of sight separation between the
subclusters as ∆lLOS ∼ 175h−1(1− lB/lA) Mpc. The
sample of 24 FP distances to groups A and B yields
∆lLOS = (14 ± 18)h−1 Mpc.
Because distance uncertainties in this case are
large, our method does not yield tight mass con-
straints in Abell 2626; however, if groups A and B are
merging (99% confidence from FP distances and 93%
confidence from galaxy magnitude distributions), it is
clear that the virial masses of groups A and B under-
estimate the total system binding mass. Specifically,
the sum of the masses of groups A and B must be
at least 1.5 × 1015M⊙, compared to the virial sum of
9.1+3.1
−2.7 × 1014M⊙ (MGW96), a factor of 1.65 higher.
The virial mass range corresponds to 90% statistical
confidence limits, where it is assumed the mass uncer-
tainties are dominated by the velocity dispersion un-
certainties (Heisler, Tremaine & Bahcall 1985). The
virial sum is inconsistent with the minimum merger
mass at the ∼ 3σ level.
5. Discussion
The study of double clusters with the fundamental
plane holds promise for significant progress in two ar-
eas. First, the study of appropriately isolated nearby
double clusters should lead to new constraints on the
total cluster binding mass. These constraints will not
only provide an independent test of the virial, hy-
drostatic, and weak lensing mass estimators, but will
also provide information about the mass (and typical
M/L) outside the virialized cluster region. Second,
the study of more distant double clusters (at z ≥ 0.1)
provides a means of directly probing for environmen-
tally driven biases in FP distance estimates.
5.1. Fundamental Plane Analysis
Here we apply this approach to the double cluster
A2626 at cz ∼ 17, 500 km/s. Because of the cluster
10
distance, it is critical to make seeing corrections while
extracting θe and hµei. Our method is similar to the
one described by Saglia et al. (1993b), but includes
a seeing–correction uncertainty which makes the pho-
tometric parameters less sensitive to the in/exclusion
of central points in the profile. We also note that
the cosmological dimming correction to the surface
brightness has a peculiar velocity dependence which
may be important in cases where many FP distance
estimates are combined to produce a single, more ac-
curate cluster distance (§2.2).
Through numerous cross checks we demonstrate
that our estimates of the combination log(θe)− β hµei
which appears in the FP are sufficiently accurate that
their errors make no significant contribution to the FP
scatter; using a Monte Carlo approach we estimate
the velocity dispersion uncertainties (§2.3). These pa-
rameters for the 8 galaxies in group B and the 16 in
group A are listed in Table 2.
We combine galaxies from both clusters in a single
fit, weighting to account for incompleteness (§3.1).
Our best fit FP coefficients (α = 1.30 ± 0.36 and
β = 0.31± 0.06) are statistically consistent with mea-
surements in other clusters (see Fig. 2). The RMS
scatter about this plane in ∆ log Re is 0.09, corre-
sponding to 23% distance uncertainties per galaxy.
We examine the distribution of M gb equivalent
widths with the galaxy spectra from both subclus-
ters (§3.2). We find that there is no evidence for sig-
nificant differences in the M gb–σ relation (see Fig.
3). Our result does not contradict the recently
noted correlation between cluster environment and
linestrength (JFK96, Jørgensen 1997); the expected
offset in M gb (given the velocity dispersion differences
between groups A and B) is too small to detect with
our data.
The difference in the fundamental plane zeropoints
for the two clusters is γA − γB = −0.037± 0.046, con-
sistent with no offset (§3.3). Under the assumption
that this zeropoint difference is indicative of distance
differences, log (DB/DA) = −0.037±0.046, where the
uncertainty follows from scatter around the median
distance in each cluster with the best fit FP coeffi-
cients. We use the 104 bootstrap resampling simula-
tions to measure the variation of the distance offset
as FP coefficients vary. The relative distance con-
straint is robust enough to reject the Hubble flow hy-
pothesis (log (DB/DA) = 0.065) in 99% of the simu-
lations. Under the Hubble flow hypothesis, the two
clusters are not interacting gravitationally, and the
velocity difference reflects pure Hubble flow. An anal-
ysis of the R band magnitude distributions of con-
firmed members of both clusters rules out Hubble flow
with 93% confidence (MGW96).
If the subclusters were on an outgoing trajectory
(bound or unbound), then the line of sight distance
difference would have to be greater than their cur-
rent recession velocity times the age of the universe
(∆lLOS > vLOStH ∼ 34 Mpc) which is greater than
the Hubble flow value: log (DB/DA) ≥ 0.0645; thus,
models where groups A and B are on outgoing tra-
jectories are ruled out with higher confidence than
pure Hubble flow. Therefore, our relative distances
indicate that the subclusters are bound and infalling
with 99% confidence (see Fig. 4).
5.2. Radial Infall Model
Assuming bound and infalling subclusters, we model
these kinematics with a radial infall model (Beers,
Geller & Huchra 1982); this two body model is appro-
priate for separations large compared to the scale of
the virialized region of the cluster and in cases where
there are no other massive clusters in the neighbor-
hood. Given the projected separation, line of sight
velocity difference, and line of sight separation, the
radial infall model provides an estimate of the total
cluster binding mass which depends weakly on the
adopted age of the universe. From Fig. 5, the total
binding mass is at least 1.65 times higher than the
virial sum, a difference significant at ∼ 3σ. The total
binding mass could be much larger than this mini-
mum.
The differences in these two mass estimates could
indicate (1) deviations from radial infall, (2) an error
in the virial estimate, or (3) significant mass beyond
the virialized region. The X–ray image of A2626 pro-
vides no indication of interactions between the two
subclusters, and there is no third subcluster, so the
radial infall model should be appropriate in A2626.
If the virial estimators are biased low, then the radial
infall model indicates that M/LR ∼ 1000h (see Ta-
ble 1), which would be a surprisingly high value (e.g.
Ramella et al. 1989). These arguments suggest that
the minimum merger mass is larger than the virial
mass because it is sensitive to mass in the cluster in-
fall region.
Taking the merger mass as a lower limit on the
mass within the cluster infall region, we can address
whether there is evidence for variation of the mass to
light ratio between the cluster core and infall region.
The galaxy light is calculated within a region which
only extends to a projected radius of 1.5h−1 Mpc;
presumably both the galaxy and mass distributions
extend beyond this region. We have no CCD pho-
tometry over the larger region to directly measure
the total cluster light. Assuming that the projected
galaxy distribution falls off as ∼ R−1.5 (e.g. Mohr
et al. 1996) with a similar luminosity function to
the galaxies in the central region, the total cluster
light grows as ∼ √R for R larger than the core ra-
dius. Under these assumptions, within a diameter of
∼ 4h−1 Mpc there is enough light to yield a mass–to–
light ratio of M/LR ∼ 600h throughout the cluster
even with the higher mass estimate from the radial
infall model. Of course, the binding mass could be
many times larger than the minimum value, so the
mass–to–light ratio could increase significantly out-
side the virialized cluster core. A study of the cluster
light extending to larger radius and a larger sample of
FP distances would yield more concrete information
regarding any variation in the cluster mass–to–light
ratio in A2626.
FP studies of other double clusters, coupled with
(1) a more detailed accounting of the radial distribu-
tion of cluster light and (2) larger redshift samples to
enable a more accurate estimate of the virial mass,
should provide interesting constraints on the distri-
bution of cluster mass and possible differences in the
efficiency of galaxy formation beyond the virialized
region.
We thank Roberto Saglia and Glenn Baggley for
their comments on an earlier version of this manuscript
and for their generosity in allowing us to compare
our measurements to their unpublished results. We
thank Guy Worthey for stimulating discussions, Paul
Schechter for obtaining improved 2.4 m images of
several galaxies in our sample, and Emilio Falco for
attempting additional imaging. This research was
funded in part by NAGW–2367 and NAG5–3401.
REFERENCES
Abell, G. 1958, ApJS, 3, 211
Baggley, G. 1996, D.Phil. Dissertation, Oxford Uni-
versity
Beers, T.C., Geller, M.J. & Huchra, J.P. 1982, ApJ,
257, 23
Beers, T.C., Forman, W., Huchra, J.P., Jones, C. &
Gebhardt, K. 1991, AJ, 102, 1581
11
Beers, T.C., Gebhardt, K., Huchra, J.P., Forman, W.,
Jones, C. & Bothun, G.D. 1992, ApJ, 400, 410
Bender, R., Burstein, D., & Faber, S.M. 1992, ApJ,
399, 462
Bracewell, R.N., 1986, The Fourier Transform and Its
Applications, (McGraw–Hill: New York)
Buote, D. & Tsai, J. 1995, ApJ, 452, 522
Burstein, D., Faber, S., Gaskell, C.M., Krumm, N.
1984, ApJ, 287, 586
Colless, M., Burstein, D., Wegner, G., Saglia, R.P.,
McMahan, R., Davies, R.L., Bertschinger, E. &
Baggley, G. 1993, MNRAS, 262, 475
Colless, M. & Dunn, A. 1995, ApJ, 458, 435
David, L.P., Jones, C. & Forman, W. 1995, ApJ, 445,
578
Djorgovski, S. & Davis, M. 1987, ApJ, 313, 59
Dressler, A., Lynden–Bell, D., Burstein, D., Davies,
R.L., Faber, S.M, Terlevich, R.J. & Wegner, G.
1987, ApJ, 313, 42
Dressler, A. & Shectman, S. 1988, AJ, 95, 985
Faber, S.M., Dressler, A., Davies, R.L., Burstein, D.,
Lynden–Bell, D., Terlevich, R.J. & Wegner, G.
1987, in Nearly Normal Galaxies, ed. S.M. Faber,
(New York: Springer–Verlag), 175
Fabricant, D.G. & Hertz, E. 1990, SPIE Proc. 1235,
747
Frei, Z. & Gunn, J.E. 1994, AJ, 108, 1476
Geller, M.J. & Beers, T.C. 1982, PASP, 94, 421
Guzman, R., Lucey, J.R., Carter, D. & Terlevich, R.J.
1992, MNRAS, 257, 187
Guzman, R., Lucey, J.R., & Bower, R.G. 1993, MN-
RAS, 265, 731
Heisler, J., Tremaine, S. & Bahcall, J.N. 1985, ApJ,
298, 8
Jackson, R.E. 1982, PhD Dissertation, University of
California, Santa Cruz
Jones, C. & Forman, W. 1992, in Clusters and Su-
perclusters of Galaxies (NATO ASI Vol. 366), ed.
A.C. Fabian, London: Kluwer, p. 49
Jørgensen, I. Franx, M. & Kjaergaard, P. 1993, ApJ,
411, 34 (JFK93)
Jørgensen, I. Franx, M. & Kjaergaard, P. 1995, MN-
RAS, 276, 1341
Jørgensen, I. Franx, M. & Kjaergaard, P. 1996, MN-
RAS, 280, 167 (JFK96)
Jørgensen, I. 1997, astro–ph 9702076
Lacey, C. & Cole, S. 1993, MNRAS, 262, 627
Lacey, C. & Cole, S. 1994, MNRAS, 271, 676
Landolt, A. 1992, AJ, 104, 340
Lucey, J.R. & Carter, D. 1988, MNRAS, 235, 1177
12
Lynden–Bell, D., Faber, S.M., Burstein, D., Davies,
R.L., Dressler, A., Terlevich, R.J. & Wegner, G.
1988, ApJ, 326, 19
Mohr, J.J., Fabricant, D.G. & Geller, M.J. 1993, ApJ,
413, 492
Mohr, J.J., Evrard, A.E., Fabricant, D.G. & Geller,
M.J. 1995, ApJ, 447, 8
Mohr, J.J., Geller, M.J., Fabricant, D.G., Wegner,
G., Thorstensen, J. & Richstone, D.O. 1996, ApJ,
470, 724
Mohr, J.J., Geller, M.J. & Wegner, G. 1996, AJ, 112,
1816; (MGW96)
Pearce, F.R., Thomas, P.A. & Couchman, H.M.P
1994, MNRAS, 268, 953
Peebles, P.J.E. 1993, Principles of Physical Cosmol-
ogy, (Pinceton University Press: Princeton), p 535
Press, W.H., Flannery, B.P., Teukolsky, S.A. & Vet-
terling, W.T. 1988, Numerical Recipes in C, (Cam-
bridge University Press: Cambridge)
Ramella, M., Geller, M.J., & Huchra, J.P. 1989, ApJ,
344,57
Rose, J.A., Bower, R.G., Caldwell, N., Ellis, R.S.,
Sharples, R.M. & Teague, P. 1994, AJ, 108, 2054
Saglia, R.P., Bender, R. & Dressler, A. 1993a, A&A,
279, 75
Saglia, R.P., Bertschinger, E., Baggley, G., Burstein,
D., Colless, M., Davies, R.L., McMahan, R.K. &
Wegner, G. 1993b, MNRAS, 264, 961
Saglia, R.P., Bertschinger, E., Baggley, G., Burstein,
D., Colless, M., Davies, R.L., McMahan, R.K. &
Wegner, G. 1997, MNRAS, submitted
Sandage, A. & Bedke, J. 1994, The Carnegie Atlas of
Galaxies: Volumes I & II, (Carnegie Institution of
Washington: Washington, D.C.)
Savage, B.D. & Mathis, J.S. 1979, Ann Rev Astron
Astro, 17, 73
Scodeggio, M., Solanes, J.M., Giovanelli, R. &
Haynes, M.P. 1995, ApJ, 444, 41
Wegner, G., Colless, M., Baggley, G., Davies, R.L.,
Bertschinger, E., Burstein, D., McMahan, R.K. &
Saglia, R.P. 1996, ApJS, 106, 1
White, S.D.M., Navarro, J.F., Evrard, A.E. & Frenk,
C.S. 1993, Nature, 366, 429
Worthey, G., Faber, S.M., Gonzalez, J.J. & Burstein,
D. 1994, ApJS, 94, 687
Worthey, G., Trager, S.C. & Faber, S.M. 1996, in
Fresh Views of Elliptical Galaxies, eds. Buzzoni,
A., Renzini, A. & Serrano, A. ASP Conference Se-
ries, Vol 86, 203
Worthey, G., & Ottaviani, D. L. 1997, ApJS, in press
This 2-column preprint was prepared with the AAS LATEX
macros v4.0.
A. Effects of Culling the Galaxy Sample
Four of our 24 galaxies show peculiar morphologies; here we detail the peculiarities and demonstrate that
eliminating them from the analysis does not significantly alter our conclusions. The following are morphological
types which we derived by comparing deep 2.4 m images and the Carnegie Atlas of Galaxies (Sandage & Bedke
1994):
• AH: bright nucleus with possible bar; probable SB0
• AL: possible spiral structure; SB0 or Sa
• AM: bright nucleus with possible bar; probable SB0
• AP: appears to have a dust ring; probable RSB0
Without these four galaxies, Group A (B) contains 12 (8) galaxies; we now apply the analysis detailed in §3 to
this smaller sample. The best fit form of the FP is
log θe = (1.39 ± 0.65) log σ + (0.32 ± 0.06)hµei + γ,
(A1)
and the scatter around this plane in log Re is 0.080 (20% distance uncertainty). These coefficients are consistent
with the values from the full sample, but the scatter is somewhat smaller and the α uncertainty is larger. The FP
coefficient uncertainties are determined (as before) through bootstrap resampling and refitting the sample.
The RMS scatter around the median distance estimator in group A (B) is 0.066 (0.115); the zeropoint difference
corrected for homogeneous Malmquist bias is
∆γ = −0.026 ± 0.045.
(A2)
The zeropoint difference and uncertainty (from the scatter) are somewhat smaller than for the whole sample (see
Eqn. 3-4). This measurement is inconsistent with the Hubble flow hypothesis at ∼ 2.0σ or 97.7% confidence. The
bootstrap refitting produces a distribution of ∆γ which includes the variation in the FP coefficients; 5000 refits of
this smaller sample rule out the Hubble flow hypothesis with 98.2% confidence. In summary, removing 4 galaxies
which have morphological peculiarities slightly reduces the significance with which the Hubble flow model can be
rejected in Abell 2626.
Table 1
Subcluster Properties
ID N
¯v
σ
Mvir[1014M⊙] Mvir/LR
A
B
67
30
16,533±141
19,164±138
658+111
−81
415+117
−72
6.6+2.4
−1.5
2.3+1.4
−0.7
630
570
Intervals are statistical 90% confidence limits
13
Table 2
Galaxy Properties
Tag
RA(1950)
Decl
cz
σ
ǫσ MR
23 34 29.30
AA
23 35 18.21
AB
23 34 22.71
AC
23 33 40.11
AD
23 33 23.09
AE
23 35 33.91
AF
23 32 40.07
AG
23 33 40.68
AH
23 34 32.46
AI
23 34 35.93
AJ
23 33 58.71
AK
AL
23 34 08.75
AM 23 33 49.52
23 33 57.86
AO
23 33 47.81
AP
23 35 19.60
AZ
BA
23 34 07.96
23 33 19.80
BB
23 33 57.29
BC
23 33 08.31
BD
23 33 17.67
BE
BF
23 33 33.61
23 33 17.50
BG
BH
23 33 12.20
20 22 22.3
20 31 56.7
20 56 55.6
20 45 31.4
20 42 02.1
21 07 54.7
21 00 01.6
20 42 35.6
20 18 14.5
20 59 41.0
21 01 12.4
20 49 29.9
20 49 12.1
20 51 11.8
20 49 21.8
20 30 15.3
20 31 29.6
20 33 12.0
20 35 11.2
20 48 02.8
20 22 54.2
20 49 36.5
20 52 03.1
20 35 20.1
17,538
17,159
16,819
17,911
15,941
16,728
15,979
16,644
17,804
16,665
16,856
17,645
16,129
16,113
16,859
16,063
19,729
19,369
19,240
19,150
19,215
19,825
18,818
19,136
287
223
218
190
205
191
163
161
185
178
202
152
252
190
172
118
148
219
187
214
133
119
133
151
12
22
22
22
28
12
19
6
7
35
20
3
21
27
8
11
3
21
17
10
3
3
6
5
14.06
13.93
14.37
14.75
15.03
14.97
14.71
15.02
14.93
15.35
15.35
14.61
14.40
15.91
15.07
16.02
14.84
14.31
14.36
14.21
15.18
15.89
16.02
17.09
hµei
19.79
19.47
19.56
19.93
19.22
19.23
19.33
20.15
19.24
18.94
19.23
19.67
21.03
17.66
19.70
19.74
19.19
19.61
20.06
20.48
19.17
19.77
19.71
18.34
θa
e
M gb
6.232
5.477
4.846
4.817
3.063
3.161
3.727
4.732
3.231
2.319
2.564
4.554
9.403
0.997
3.743
2.136
3.351
4.900
6.243
8.148
2.836
2.705
2.431
0.761
4.83
4.50
4.41
4.64
4.42
4.17
3.83
4.54
4.14
4.67
4.86
3.67
4.87
4.68
4.86
3.89
3.80
4.27
4.42
4.45
4.27
3.79
4.61
4.98
F b
B
0.53
0.34
0.79
0.63
0.55
0.74
0.85
0.50
0.84
0.90
0.70
0.39
0.64
1.00
0.30
0.70
0.69
0.74
0.69
1.00
0.62
0.38
0.68
0.33
ain arc seconds
bbulge luminosity fraction
14
Fig. 1.- The galaxy velocity distribution around Abell 2626. The histogram contains 108 galaxy velocities from the
redshift survey of MGW96. Group A is centered at v = 16, 500 km/s, and group B is centered at v = 19, 150 km/s.
The high fraction of galaxies with emission lines in the group of 11 centered at v ∼ 21, 500 km/s (73% compared
to 39%/33% for groups A/B) indicates they are part of a low density background structure.
15
Fig. 2.- Two projections of the FP. Both plots contain group A (solid) and B (hollow) galaxies. The lines define
the best fit group A (solid) and B (dashed) FP; offsets in the fits reflect subcluster distance differences. The best fit
FP (with weighting to account for incompleteness– see text) is of the form log θe = 1.30 log σ + 0.31 hµei + γcl where
γcl is the distance dependent cluster zeropoint. The RMS scatter around the best fit in log θe is 0.09, corresponding
to a 23% distance uncertainty per galaxy.
16
Fig. 3.- Plot of log (M gb) vs log (σ) for the galaxies in groups A (solid) and B (hollow). The lines mark the best fit
relations for groups A (solid) and B (dashed) constrained to have the slope found using a much larger galaxy sample
(Jørgensen 1997). The apparent zeropoint offset between the two samples is statistically insignificant (0.012±0.019;
see text), providing no evidence for significant stellar population differences between the subclusters.
17
Fig. 4.- Plot of fundamental plane relative distances versus redshift for galaxies in groups A (solid) and B
(hollow). Galaxies in groups A and B are clearly offset in velocity (see Fig. 1), and also appear to be offset in
distance. Vertical lines mark the (homogeneous Malmquist bias corrected) median distances to groups A (solid) and
B (dashed). Horizontal lines mark the mean velocities of groups A (solid) and B (dashed). The large star marked
with DHubble lies at the expected position of group B if its trajectory relative to group A were pure Hubble flow.
The median distances to groups A and B indicate that log (DB/DA) = −0.037± 0.046; the 1σ distance uncertainty
is indicated by the error bar on the vertical, dashed line. The sample of 24 galaxies rules out the Hubble flow
hypothesis, DB = DHubble and log (DB/DA) = 0.065, with 99% confidence (see §3.3).
18
Fig. 5.- The ratio of the cluster merger mass MG to the sum of the group A and B virial masses MA + MB =
9.1×1014h−1M⊙ as a function of the line of sight separation ∆lLOS of the two groups. The two solutions correspond
to different assumptions regarding the age of the universe tH : tH = 13 Gyr (solid) and tH = 18 Gyr (dotted). The
statistical uncertainty in the merger mass MG is 12%. If groups A and B are merging, as suggested by the FP
distances and the galaxy magnitude distributions, the merger mass is at least 1.65 times greater than the sum of
the virial masses. The FP distances indicate that ∆lLOS = 14 ± 18 h−1 Mpc.
19
|
astro-ph/0210358 | 3 | 0210 | 2004-04-14T17:36:14 | Testable anthropic predictions for dark energy | [
"astro-ph",
"gr-qc",
"hep-ph"
] | In the context of models where the dark energy density $\rD$ is a random variable, anthropic selection effects may explain both the "old" cosmological constant problem and the "time coincidence". We argue that this type of solution to both cosmological constant problems entails a number of definite predictions, which can be checked against upcoming observations. In particular, in models where the dark energy density is a discrete variable, or where it is a continuous variable due to the potential energy of a single scalar field, the anthropic approach predicts that the dark energy equation of state is $p_D=-\rho_D$ with a very high accuracy. It is also predicted that the dark energy density is greater than the currently favored value $\Omega_D\approx 0.7$. Another prediction, which may be testable with an improved understanding of galactic properties, is that the conditions for civilizations to emerge arise mostly in galaxies completing their formation at low redshift, $z\approx 1$. Finally, there is a prediction which may not be easy to test observationally: our part of the universe is going to recollapse eventually. However, the simplest models predict that it will take more than a trillion years of accelerated expansion before this happens. | astro-ph | astro-ph |
Testable anthropic predictions for dark energy
J. Garriga1,2 and A. Vilenkin 2
1 Departament de F´ısica Fonamental, Universitat de Barcelona,
Mart´ı i Franqu`es 1, 08193 Barcelona, Spain and
2 Institute of Cosmology, Department of Physics and Astronomy,
Tufts University, Medford, MA 02155, USA
(Dated: September 27, 2018)
In the context of models where the dark energy density ρD is a random variable,
anthropic selection effects may explain both the "old" cosmological constant problem
and the "time coincidence". We argue that this type of solution to both cosmologi-
cal constant problems entails a number of definite predictions, which can be checked
against upcoming observations. In particular, in models where the dark energy den-
sity is a discrete variable, or where it is a continuous variable due to the potential
energy of a single scalar field, the anthropic approach predicts that the dark energy
equation of state is pD = −ρD with a very high accuracy. It is also predicted that
the dark energy density is greater than the currently favored value ΩD ≈ 0.7. An-
other prediction, which may be testable with an improved understanding of galactic
properties, is that the conditions for civilizations to emerge arise mostly in galaxies
completing their formation at low redshift, z ≈ 1. Finally, there is a prediction
which may not be easy to test observationally: our part of the universe is going to
recollapse eventually. However, the simplest models predict that it will take more
than a trillion years of accelerated expansion before this happens.
I.
INTRODUCTION
The "old" cosmological constant problem - why don't we see the large vacuum energy
density ρΛ which is expected from particle physics? - and the "time coincidence" problem
- why do we live at the epoch when the dark energy component ρD starts dominating? -
may find a natural explanation in models where ρD is a random variable. The idea is to
introduce a dynamical dark energy component X whose contribution ρX varies from place
to place, due to processes which occurred in the early universe. Then
ρD = ρΛ + ρX
will also vary from place to place, and the old cosmological constant problem takes a different
form. The question is not why ρΛ is much smaller than η4, where η is some high energy
physics mass scale, such as the supersymmetry breaking scale η ∼ T eV , but why do we
happen to live in a place where ρΛ is almost exactly cancelled by ρX. This line of enquiry is
rather quantitative, since we can ask what is the probability for us to observe certain values
of ρD ∼ 10−11(eV )4, or what is the probability for the time coincidence.
Explicit particle physics models for a variable ρX have been reviewed in [1]. Two examples
which have been thoroughly discussed in the literature are a four-form field strength, which
can vary through nucleation of membranes [2, 3], and a scalar field with a very low mass [3, 4].
Assuming one such mechanism, and using a theory of initial conditions such as inflation,
2
one can calculate the "a priori" probability distribution P∗(ρD)dρD. This is defined as the
fraction of co-moving volume which at some fiducial initial time (which we conventionally
take to be the time of recombination) had the value of the dark energy density in the interval
dρD. Inflation is also responsible for smoothing out the value of ρD over comoving distances
much larger than the size of our presently observable universe.
By itself, P∗ is not sufficient to calculate probabilities for our observations. Selection
effects which bias the measurement of ρD must be included, and the most important one
1 While ρD may be very large in most places, there is
in this case is anthropic [5 -- 9].
nobody there to observe such extreme values. If ρD > 0, galaxy formation stops once the
dark energy becomes dominant over the matter density. Some galaxies are seen at redshifts
of order z ∼ 5, but not much higher, indicating [9] that galaxies will not form in regions
where ρD & (1 + zEG)3ρ0. Here, ρ0 is the matter density at the present time t0, and zEG ≈ 5
is the redshift at the time tEG ∼ (1 + zEG)−3/2t0 when the earliest galaxies formed. Also,
for a negative ρD the universe recollapses on a time scale tD ∼ GρD−1/2, where G is
Newton's constant. This time should be larger than the earliest time tEI which is required
for intelligence to develop [7, 13]. Thus, observers will only exist within a tiny "anthropic
range"
(1)
−(Gt2
EI)−1 . ρD . (Gt2
EG)−1.
It should be noted that, aside from the above minimal requirements, anthropic selection
includes all other ways in which ρD disfavours the existence of observers. For instance, in
regions where ρD < 0, the matter density is larger than ρD throughout the cosmic evolution.
If ρD is too large, all galaxies formed in that region will be very dense, and as a result,
very inhospitable. This occurs also for a large ρD > 0, since galaxies must form before ρD
starts dominating. We shall come back to this issue in Sections III and V.
The selection effect can be implemented quantitatively by assuming the mediocrity prin-
ciple, according to which our civilization is typical in the ensemble of all civilizations in the
universe. The probability to find ourselves in a region with given values of ρD is thus given
by [14]
dP(ρD) ∝ P∗(ρD)nciv(ρD)dρD.
(2)
Here, nciv(ρD) refers to the number of civilizations which will ever form per unit co-moving
volume in regions where the dark energy density was equal to ρD at the time of recombina-
tion. 2
Needless to say, the determination of both factors in the r.h.s. of Eq. (2) leaves room for
some uncertainties. However, we shall argue that there are reasons to be optimistic. If the
distribution (2) is to explain both cosmological constant problems, then a number of rather
generic predictions can be made, rendering these ideas very testable.
In the next section, we review the calculation of the prior probability distribution P∗(ρD).
The anthropic factor nciv(ρD) is discussed in section III. In the same section, we argue that
the anthropic approach can succeed only if the conditions for civilizations to evolve arise
mostly in galaxies formed at low redshifts, z ∼ 1. Anthropic predictions for the dark energy
1 Anthropic selection effects associated with the possible variation of the amplitude of density fluctuations
[10, 11] and of the baryon to photon ratio [10, 12] have also been discussed in the literature.
2 As we shall argue, in models where both cosmological constant problems can be solved anthropically, ρD
has not varied appreciably since the time of recombination, and therefore it can be treated as constant in
time.
3
equation of state, for the energy density ρD, and for the Hubble parameter h are discussed
in sections IV, V. The prediction for the future of the universe in unveiled in section VI.
Finally, our conclusions are briefly summarized and discussed in section VII.
II. THE PRIOR DISTRIBUTION
The first task in determining (2) is to estimate P∗. The vacuum energy density is of
order ρΛ ∼ η4 & (T eV )4, and therefore ρD must have a natural range of variation of order
η4 or larger. Weinberg noted [9] that a function P∗(ρD) that varies smoothly on scales
ρD ∼ η4, should behave as a constant in the utterly narrower interval (1) - unless of course,
the function would happen to have a zero or a pole in that interval (which would be an utter
coincidence). This led him to conjecture that for values of ρD in the anthropic range the
prior probability would be constant,
P∗(ρD) ≈ const.
(3)
Outside of this range the form of P∗ is irrelevant, because the factor nciv vanishes. Weinberg's
conjecture is subject to verification. As mentioned in the Introduction, P∗ is calculable,
provided that the dynamics of ρX is known, and assuming an inflationary model which
would determine its spatial distribution at the time of recombination. Analysis of explicit
models shows that (3) is not automatically guaranteed [4], but it does seem to be satisfied
in generic models.
There are basically two reasons [1, 4] why a non-flat P∗ may result from the process of
randomization of ρD which occurs during inflation (this randomization is due to quantum
diffusion in the case where X is a scalar field, or to nucleation of membranes in the case when
X is a four-form). The first reason is the differential expansion induced by the dark energy
component. During inflation, the expansion rate is determined by H 2 = (8πG/3)(Vinf +ρD).
Although ρD is very small compared with the inflationary potential Vinf , its effect may
build up over time, in such a way that more thermalized volume is generated with high
values of ρD. In this way, P∗(ρD) could be biased towards large values. Let us denote by
τ (X, H) the characteristic time needed for the dynamics of X to sample (at a fixed point in
space) all values of ρD within the anthropic range (∆ρD)anth. The differential expansion is
characterized by the parameter
q = (∆H)τ = (4πG/3)H −1(∆ρD)anthτ (X, H).
(4)
If q ≫ 1, then P∗ is exponentially steep in the range of interest. This case is ruled out
by observations, because it predicts a very large ρD, even after selection effects have been
factored in. If q ∼ 1, the distribution P∗ may have a moderate dependence on ρD within
the anthropic range. This dependence affects the position of the peak of the distribution for
the observed values of ρD, Eq. (2), and hence it affects our predictions. While models of
this sort are not ruled out, they require a very unnatural adjustment of parameters, since
q is determined by a combination of rather different pieces of dynamics. Hence, we shall
disregard this marginal possibility as non-generic. Finally, there is a wide class of models
where q ≪ 1 is satisfied without any fine tuning [1, 4], and hence we shall take this to
be the generic case. Numerical simulations confirm that in this case the bias effect due to
differential expansion is insignificant [18].
4
The second reason why P∗ may be non-flat is the following. Even if the differential
expansion is negligible, and the prior distribution for X is flat, this does not automatically
guarantee that the prior for ρD will be flat, unless the relation between X and ρD is linear
in the range of interest. Through this effect, it is possible to have a moderate variation of
P∗(ρD) within the anthropic range. But again, this would require a contrived adjustment
of parameters and we shall dismiss this case as non-generic (see also [19] for a discussion of
this issue).
As an example, let us consider the case where ρX = V (φ) is the potential energy density
of a scalar field φ,
ρD = ρΛ + V (φ).
(5)
The field must change very slowly on a cosmological time-scale, so that its potential energy
behaves as an effective cosmological constant. This requires the slow-roll conditions [4]
V ′ ≪ 10ρD/mp,
V ′′ ≪ 102ρD/m2
p
(6)
to be satisfied up to the present time (when ρD ∼ ρ0, with ρ0 the present matter density).
The constraint q ≪ 1 on the differential expansion yields [4]
V ′2H 4/GV 3 ≫ 1.
(7)
During inflation, the scalar field is randomized by quantum fluctuations, and at recombina-
tion it is distributed according to the "length" in field space,
P∗(φ)dφ ∝ dφ.
.
(8)
(9)
Therefore 3,
P∗(ρD)dρD ∝
dρD
V ′(φ)
Thus, the flatness of the prior depends on how much V ′ changes in the anthropic range.
As we shall see, variations in this range may occur, but they do not bias the probability
distribution for ρD in any significant way, unless we adjust some parameters specifically for
this purpose.
Consider a potential of the form
V (φ) =
1
2
µ2φ2,
(10)
where µ2ρΛ < 0, so that it is possible to have ρD very small even if ρΛ is large. Eqs. (6)
lead to the condition [4]
(11)
µ ≪ 10−120m3
pρΛ−1/2.
Such a small mass parameter may seem unrealistic, but it can naturally arise, for instance,
in a low energy effective theory with a suitable discrete symmetry [3] (for other proposals,
see [1, 19, 20] and references therein). Note that (11) does not correspond to a fine tuning,
but just to a strong suppression. The condition (7) translates into
µ ≫ H 3
0 H −2 ∼ 10−169mp,
(12)
3 Note that near the points where V ′(φ) = 0, we have ρD ≈ A + Bφ2 and V ′(φ) ∼ φ ∼ (ρD − A)1/2, which
is integrable. Hence, the zeroes of V ′(φ) are not a concern.
P∗(ρD)dρD =
κ + (M 4/η) cos(φ/η)−1 .
dρD
(17)
where H0 is the present Hubble rate, and in the last step we have used H ∼ 10−7mp,
corresponding to a GUT scale of inflation. The conditions (11) and (12) leave very many
orders of magnitude available for the parameter µ, and so fine tuning is not necessary. From
(5),
5
(13)
0 = −2ρΛ/µ2 and κ = µ2φ0. We are interested in the vicinity of ρD = 0, where it is
ρD = κ(φ − φ0) +
where φ2
easy to show from (9) that [4]
µ2
2
(φ − φ0)2,
P∗(ρD)dρD ∝ [1 + O(ρD/ρΛ)]dρD ≈ dρD.
(14)
Since ρD ≪ ρΛ in the anthropic range, the distribution is indeed flat to a very good accuracy.
For contrast, we may consider the "washboard" potential
where κ was given above and M and η are different mass scales. Let us assume that 4
ρD = ρΛ + κφ + M 4 sin(φ/η),
M 4 ≪ H 2
0 mpη ∼ (η/mp)ρ0.
(15)
(16)
Then the field will typically be found away from the local minima, with a probability dis-
tribution
In the case κ ≫ M 4/η, the distribution (17) is still flat, as in (3).
Both κ and M 4/η should be much smaller than H 2
0 mp in order to satisfy the slow roll
condition.
In the
opposite case, where M 4/η ≫ κ, the a priori distribution can have a sizable variation within
If η ≪ mp, this range is very wide in the field space,
the anthropically allowed range.
δφ & ρ0/κ ≫ mp. This means that the oscillations in P∗ will average out on scales much
smaller than the anthropic range, and effectively we recover (3). Clearly, the only way to
avoid this averaging effect is if η & mp, and
M 4 ∼ (∆ρD)anth.
(18)
The last equation is to ensure that a significant range of values of φ/η is sampled in the
anthropic range (∆ρD)anth . 103ρ0, so that changes in the slope of the potential are appre-
ciable. Otherwise the distribution for P∗ will be almost flat. Thus, aside from the fact that
the washboard potential is already a somewhat contrived example [19], Eq. (18) implies an
otherwise unnecessary adjustment of the parameter M.
In what follows, we shall only consider models where there is no such ad-hoc adjustment.
In this sense, our predictions may not be completely inescapable, but they can be considered
generic. The situation can be compared with the predictions of inflation that the density
parameter is Ω = 1 and the spectrum of density perturbations is nearly flat. It is certainly
possible, in the context of inflation, to have an open universe with Ω < 1, or to have
a markedly non-flat spectrum of density perturbations. But to achieve this, additional
parameters must be introduced and adjusted to the desired outcome.
4 If M 4 ≫ H 2
0 mpη, the slow roll condition is not satisfied today and the field φ will be in any one of the local
minima of the washboard. With some generic requirements on the inflationary parameters, the minima
will have equal a priori probability within the anthropic range [1].
6
III. THE ANTHROPIC FACTOR
We now consider the effect of the anthropic factor nciv in Eq. (2). The physical situation
is rather different for positive and negative ρD, so we consider these two cases separately.
For positive ρD, the main change introduced by nciv is that the time of earliest galaxy
formation tEG in the anthropic range (1) is effectively replaced by the time at which the
bulk of galaxy formation occurs. This is because a few early birds will not make a difference
once we apply the principle of mediocrity. More precisely, we should take into consideration
that the morphology of some galaxies could make them less suitable for the development of
civilizations, and therefore
nciv(ρD) =Z dα n(α, ρD) Nciv(α).
(19)
Here, α denotes the set of parameters characterizing the type of galaxy (e.g. its size, density,
etc.), n(α, ρD) is the number density of such galaxies that form per co-moving volume in
regions characterized by ρD, and Nciv(α) is the number of civilizations per galaxy of type α.
Suppose that the above integral receives a dominant contribution from galaxies of type αG.
Then
nciv(ρD) ∝ n(αG, ρD),
(20)
and the relevant time for anthropic considerations is the time at which this type of galaxies
form, which we shall denote by tG. With the assumption of a flat prior P∗, it was shown in
[11, 39] that the most probable value for a positive ρD is the one characterized by
This fact was used in order to explain the observed time coincidence
tD ∼ tG.
tD ∼ t0.
(21)
(22)
The last relation follows from (21), assuming that stars and civilizations develop on a
timescale not much greater than tG, and therefore tG is comparable to t0, defined as the
time when most civilizations make their first determination of ρD.
Connected with the above discussion, there is a prediction of the anthropic approach,
which can be checked by a combination of observations and theoretical analysis. In a not so
distant future, our understanding of galactic evolution and morphology may improve to the
point where we can tell with some confidence which galaxies are suitable for sustaining plan-
etary systems similar to our own, where civilizations can develop. The anthropic approach
to the cosmological constant problems (CCPs) predicts that the conditions for civilizations
to emerge will be found mostly in galaxies that formed (or completed their formation) at a
low redshift, z ∼ 1.
In the standard cold dark matter cosmology, galaxy formation is a hierarchical process,
with smaller objects merging to form more and more massive ones. We know from observa-
tions that some galaxies existed already at z = 5, and the theory predicts that some dwarf
galaxies and dense central parts of giant galaxies could form as early as z = 10 or even 20.
The fraction of matter bound in giant galaxies (M ∼ 1012M⊙) at z = 1 (∼ 20%) is some-
what less than that in objects of mass ∼ 109M⊙ at z = 3, or in objects of mass ∼ 107M⊙
at z = 5 [30]. If civilizations were as likely to form in early galaxies as in late ones, then
7
Eq. (21) would indicate that, for a typical observer, the cosmological constant should start
dominating at a redshift zG & 5. The corresponding dark energy density,
ρD ∼ (1 + zG)3ρ0,
(23)
would be far greater than observed. Clearly, the agreement becomes much better if we
assume that the conditions for civilizations to emerge arise mainly in the types of galaxies
which form at lower redshifts, zG ∼ 1.
We now point to some directions along which the choice of zG ∼ 1 may be justified. One
problem with dwarf galaxies is that if the mass of a galaxy is too small, then it cannot retain
the heavy elements dispersed in supernova explosions. Numerical simulations suggest that
the fraction of heavy elements retained is ∼ 30% for a 109M⊙ galaxy and is negligible for
much smaller galaxies [37]. The heavy elements are necessary for the formation of planets
and of observers, and thus one has to require that the structure formation hierarchy should
evolve up to mass scales ∼ 109M⊙ or higher prior to the dark energy domination. This gives
the condition zG . 3, but falls short of explaining zG ∼ 1.
Another point to note is that smaller galaxies, formed at earlier times, have a higher
density of matter. This may increase the danger of nearby supernova explosions and the rate
of near encounters with stars, large molecular clouds, or dark matter clumps. Gravitational
perturbations of planetary systems in such encounters could send a rain of comets from the
Oort-type cloud towards the inner planets, causing mass extinctions5.
Our own Galaxy has definitely passed the test for the evolution of intelligence, and the
principle of mediocrity suggests that most observers may live in galaxies of this type. Our
Milky Way is a giant spiral galaxy. The dense central parts of such galaxies were formed
at a high redshift z & 5, but their discs were assembled at z ∼ 1 or later [32]. Our Sun is
located in the disc, at a distance ∼ 8.5 kpc from the Galactic center6. If this situation is
typical, then the relevant epoch to use in Eq. (23) is the epoch zG ∼ 1 associated with the
formation of discs of giant galaxies.
The above remarks may or may not be on the right track, but we emphasize once again
that if CCPs have an anthropic resolution, then, for one reason or another, the evolution
of intelligent life should require conditions which are found mainly in giant galaxies, which
completed their formation at zG ∼ 1.
In order to estimate n(αG, ρD) in Eq. (20), we shall need a simple quantitative criterion
to specify the relevant type of galaxies. The most important parameter characterizing a
5 The cross-section for disruption of planetary orbits is much smaller, and it would take a rather substantial
increase of the density for this process to become statistically important. A.V. is grateful to David Spergel
for a discussion of this issue.
6 It has been noted [31] that this distance is close to the corotation radius, where the orbital velocity of the
stars coincides with the rotational velocity of the spiral pattern. In other words, the motion of the Sun
relative to the spiral arms is rather slow, and as a result, the periods between spiral arm crossings are
rather long (∼ 108 yrs). Spiral arms are the primary sites of supernova explosions. They are also rich in
giant molecular clouds, and are therefore very hazardous to life. It has been argued in [31] that spiral arm
crossings are responsible for the major mass extinctions observed in the fossil record. Then one expects
that habitable planetary systems are to be found mainly in the vicinity of the corotation radius, since
mass extinctions at a rate much greater than once in 108 yrs may be too frequent for intelligent life to
evolve. (Note that it took us 6.5 × 107 yrs to evolve since the last great extinction.)
8
galaxy is its mass M. For the Milky Way it is MM W ∼ 1012 M⊙ [33], and the above
discussion suggests that we identify the relevant galaxies with gravitationally bound halos
of this mass. (Note that this is also the typical mass of L∗ galaxies, which contain most of
the luminous stars in the Universe.) It should be recognized, however, that the choice of
this characteristic mass scale is somewhat uncertain, so we shall illustrate how our results
are affected by choosing a larger or a smaller mass.
Our Galaxy is a member of the Local Group cluster, whose mass has been estimated as
[34] MLG ∼ 4 × 1012 M⊙. It is conceivable that the gas captured in this cluster is later
accreted onto the member galaxies and thus affects the properties of their discs. There
seems to be no justification to consider larger mass objects, and we shall regard MLG as an
upper bound on the potentially relevant mass scales. On the lower mass end, we shall use
M ∼ 1011 M⊙, which is roughly the mass of the bright part of our Galaxy, up to ∼ 10 kpc
from the center. (We note that MM W is probably a more reasonable choice, because the
properties of the disc depend on the total mass of the halo [35].)
We now consider negative ρD. The scale factor of a universe filled with nonrelativistic
matter and dark energy with ρD < 0 is given by
a(t) = sin2/3(cid:18) t
tD(cid:19) ,
(24)
where tD ≡ (1/6πGρD)1/2. The matter density ρM initially decreases while the universe
expands, but at t = πtD/2, when it reaches the value ρM = −ρD, the universe stops its
expansion and starts recontraction. The matter density grows in the contracting phase, and
thus ρM ≥ ρD throughout the evolution. The structure formation in a universe with a
negative ρD proceeds as usual until t ∼ tD, but then the growth of density perturbations
accelerates during the contraction, so that all overdensities collapse to form bound objects
prior to the big crunch. For tD & t0, giant galaxies will form at about the same time as
they did in our part of the universe and will have similar properties (with a possible caveat
indicated below). However, for tD ≪ t0 halos of the galactic size will be forced to collapse
at a much earlier time t ∼ tD, and their density will therefore be much higher than that of
our Galaxy. This would probably make such halos unsuitable for life.
These considerations suggest that the anthropic factor effectively constrains tD to be in
the range
tD & t0
(25)
for both positive and negative ρD. There is, however, an additional factor that could make
negative ρD less probable. For ρD > 0, structure formation effectively stops at t > tD, and
the existing structures evolve more or less in isolation. This may account for the fact that
discs of giant galaxies take their grand-design spiral form only relatively late, at z ∼ 0.3.
The discs are already in place at z ∼ 1, but they have a very unsettled, irregular appearance
[32]. On the other hand, for ρD < 0 the clustering hierarchy only speeds up at t > tD, and
quiescent discs which may be necessary for the evolution of fragile creatures like ourselves
may never be formed.
Another factor to consider is the characteristic time tI needed for intelligence to develop.
For positive ρD, this factor is unimportant, since the time after the dark energy domination
is practically unlimited, but for negative ρD the available time is bounded by t < πtD, and
the effect of tI requires a closer examination.
We first note that tI ≪ t0 is unlikely, since then it is not clear why it took so long for
intelligence to develop on Earth. (The total time of biological evolution, from the origin of
9
life on Earth till present, is estimated at ∼ 3.5×109 yrs.) For tI ≫ t0, we note that the main
sequence lifetime of stars believed to be suitable to harbor life is t⋆ ∼ (5 − 20) × 109yrs ∼ t0
(see [11] for a discussion of this point). If tI ≫ t0 ∼ t⋆, most of these stars will explode as red
giants before intelligence has a chance to develop. Carter [6] has argued that this is the most
likely scenario7. In this case, the number Nciv is suppressed by a factor ∼ min{t⋆, tD}/tI ∼
t0/tI, where we have used (25) in the last step. For positive ρD, the suppression is by a
factor ∼ t⋆/tI, which is of the same order of magnitude.
We conclude that the precise value of tI has little effect on the relative probability of
positive and negative ρD. If the accelerated clustering hierarchy is detrimental for life, then
the probability for negative ρD is suppressed; otherwise the two signs of ρD are equally likely.
In either case, we should not be surprised that ρD is positive in our part of the universe. In
the following sections we shall focus on the positive values of ρD.
IV. PREDICTION FOR THE EQUATION OF STATE
A rather generic prediction of models where both CCP's are solved anthropically is that
the equation of state of dark energy is given by pD = wρD, with
w = −1 ± 10−5.
(26)
The error bars correspond to the precision to which the observable universe can be aproxi-
mated by a homogeneous and isotropic model. In models where ρX is the energy density of
a four-form field, this equation of state is guaranteed by the fact that the four-form energy
density is a constant and can only change by the nucleation of branes (other than that,
it behaves exactly like an additional cosmological constant). If ρX is a generic scalar field
potential, the slow roll conditions (6) are likely to be satisfied by excess, by many orders
of magnitude, rather than marginally. For instance, for the quadratic potential (10), these
conditions imply the constraint (11). It would be contrived to arrange for the condition to
be satisfied marginally, since the whole point of the present approach is to have ρΛ cancelled
regardless of its precise value (which is not known to us even by order of magnitude). If
the slow roll conditions are satisfied by excess by just more than three orders of magnitude,
then the kinetic energy of the scalar field will be less than its potential energy by more than
six orders of magnitude, and Eq. (26) follows.
There are certainly models for dark energy, some of them with anthropic input, were
(26) is not satisfied. For instance, Kallosh and Linde [13] recently considered a supergravity
model where the time coincidence problem is solved anthropically, and where (26) does
not hold. However, their model does not solve the old CCP, since it is assumed that the
cosmological constant vanishes in the observable matter sector due to some unspecified
mechanism. Likewise, (26) does not hold in the usual quintessence models [21], which have
no anthropic input at all, but which do not address the CCP's [1, 19], or in models of
k-essence [22], where only the time coincidence is partially addressed.
A possibility worth discussing is the case of models where the slow roll parameters are
themselves random fields. Consider, for instance, the following model:
ρD = ρΛ + µ2(ψ)φ2.
(27)
7 The coincidence tI ∼ t⋆ is unlikely, since the evolution of life and evolution of stars are governed by
completely different processes.
10
If the probability distribution for the new scalar field ψ were such that all values of µ2 are
equiprobable, then one might imagine that the order of magnitude of µ2 would be such that
the slow roll conditions would be marginally satisfied. 8
V. PREDICTIONS FOR ΩD AND h.
Currently favoured values for the dark energy density and for the Hubble parameter are
ΩD ≈ .7 and h ≈ .7 [24, 25], both with error bars of the order of 10%. While observations
are not very accurate, we would like to challenge the status quo and boldly use the anthropic
approach to the CCP's to make predictions for these two parameters. As we shall see, this
approach predicts that ΩD is likely to be somewhat higher, and that h is likely to be smaller
than those currently favoured values.
The basic reason why we expect ΩD to be larger is the following [14, 15]. The growth
of density fluctuations in a universe with a positive cosmological constant effectively stops
at the redshift zD when the cosmological constant starts dominating. This is given by
(1 + zD) ∼ (ΩD/ΩM )1/3, where ΩM = 1 − ΩD is the matter density parameter. According
to (21), we expect zD ∼ zG, where zG is the epoch when the relevant galaxies were formed.
(For
With zG ∼ 1, this corresponds to (ΩD/ΩM ) ∼ 8, which in turn implies ΩD ∼ .9.
zG > 1, we would obtain an even higher value for ρD.) This prediction can be made more
8 The new field ψ must also be a light field and hence its prior distribution is in principle calculable. We
can actually consider a more general form of the potential for ψ and φ,
ρD = ρΛ + V (ψ, φ).
(28)
Around any point (ψ0, φ0) on the curve γ defined by V (ψ, φ) = −ρΛ, the potential can be approximated
by a linear function of the fields. Moreover, we can always rotate coordinates in field space so that ψ is
directed along γ, and φ is orthogonal to it,
ρD ≈ Vφ(ψ0, φ0)(φ − φ0).
Here Vφ is the gradient of the potential at that point. During inflation, both fields ψ and φ are randomized
by quantum fluctuations. Hence, the prior probability distribution is given by the area in field space
P∗dψdφ ∝ dψdφ, which leads to
(29)
P∗(ρD)dρD =(cid:20)Zγ
dψ
Vφ(cid:21) dρD.
Along the curve γ, the values of Vφ that will carry more weight per unit distance along the curve are
those for which the slope is smaller. In the published version of this paper, it was (incorrectly) concluded
from this observation that "given a model where the slope of the potential is variable, smaller values of
the slope are preferred a priori, and there is no reason to expect that the slow roll conditions should be
satisfied only marginally." However, this is not necessarily correct in general, since the slope could be large
for a very large range of ψ, and large slopes may end up dominating the integral in Eq. (29). A detailed
analysis of this point has been given in [42], where it is shown that there is a wide class of two-field models
for which a small slope is favoured a priori (and for which the conclusions of Section IV do indeed apply).
However, there is an equally broad class of models for which a large slope is favoured a priori. In that
case, the posterior probability distribution for the slope is such that the slow roll condition is satisfied
only marginally, and some departure from the vacuum equation of state may be expected (see [43, 44]).
quantitative [17] by using the distribution (2). As we shall see, the precise predictions
depend not only on ΩD but also on h.
11
Throughout this section, we shall assume that ρD > 0 as part of our prior. In a universe
filled with pressureless matter and with a dark energy component ρD > 0, the scale factor
behaves as
a(t) = sinh2/3(cid:18) t
tD(cid:19) ,
where tD ≡ (1/6πGρD)1/2. A primordial overdensity will eventually collapse, provided that
its value at the time of recombination is larger than a certain value δrec
. In the spherical
collapse model, this is estimated as δrec
rec , where xrec = x(trec) [27]. Here, we
have introduced the variable
c (ρD) = 1.13x1/3
c
x(t) ≡
ΩD(t)
ΩM (t)
= sinh2(cid:18) t
tD(cid:19) .
(30)
The number of galaxies n(M, ρD) of mass M that will form per unit comoving volume in
a region characterized by the value ρD of the dark energy density, is proportional to the
fraction of matter that eventually clusters into this type of galaxies. In the Press-Schechter
approximation [26, 27], this is given by
nciv(ρD) ∝ n(M, ρD) ∝ erfc(cid:18) δrec
√2σrec(M)(cid:19) .
c (ρD)
(31)
Here, erfc is the complementary error function, and σrec(M) is the dispersion in the density
contrast at the time of recombination trec. As argued in the preceeding section, we shall
assume that most civilizations are formed in galaxies characterized by a mass M ∼ MM W ≈
1012M⊙ (although we shall also consider slightly larger and smaller masses).
The factor nciv depends on the parameter σrec, which in turn depends on the amplitude
of density perturbations generated during inflation. The value of σrec can be inferred from
the normalization of CMB anisotropies, but for this task, both the present value of ΩD and
the value of the Hubble parameter h would be needed. Since these are the parameters we
wish to make predictions about, it would be somewhat contrived to use them at this point
to make an inference about σrec.
Another factor to consider is that σrec may be different in distant regions of the universe
(where, as a consequence, galaxies would form earlier or later). In models where the inflaton
field has only one component, the value of σrec is the same in all regions of the universe.
However, if the inflaton field has more than one component, the amplitude of density per-
turbations depends on the path followed by the inflaton on its way to the minimum of the
potential. In such models, it is possible for σrec to vary over distances much larger than the
presently observable universe.
To make our discussion sufficiently general, we shall consider that σrec is itself a random
variable with unspecified prior. This prior may be determined by processes occurring during
inflation, or it may just reflect our ignorance of the actual value of the fixed parameter σrec.
Then, Eq. (2) is generalized to
dP(ρD, σrec) ≈ ncivP∗(ρD, σrec)dρDdσrec.
(32)
In this context, the generic expectation that the prior does not depend on ρD in the anthropic
range [see Eq. (3)], translates into
P∗(ρD, σrec) ≈ P∗(σrec).
dP
dln y
0.08
0.06
0.04
0.02
0.01
0.1
1
10
100
y
FIG. 1: The distribution (34).
Substituting (31) into (32), we have
dP(ρD, σrec) ∝ erfc .80x1/3
σrec !P∗(σrec) dxrecdσrec,
rec
12
(33)
where we have used that ΩM (trec) ≈ 1 in all regions of interest, so that dρD ∝ dxrec.
Introducing y = xrecσ−3
rec, the change of variables (xrec, σrec) → (y, σrec) produces a Jacobian
proportional to σ3
recP∗(σrec)dydσrec, where f (y) does
not depend on σrec. Integrating over σrec leads to the normalized distribution
rec, and we have 9 dP(y, σrec) ≈ f (y)σ3
dP(y) = (.80)3π−1/2erfc(.80y1/3) y d ln y,
(34)
which is uncorrelated with σrec.
The variable y can be expressed in terms of observable quantities, as we shall see below,
and from (34) we should expect y ∼ 1 by order of magnitude (See Fig. 1). More precisely,
9 The appearance of the factor σ3
rec in the posterior distribution was noted in Ref. [11]. This factor implies
that the prior P∗(σrec) should decay faster than σ−3
rec at large values of the density contrast. Otherwise,
the posterior distribution is not normalizable (and we should expect both a large value of the effective
vacuum energy and of the linear density contrast, in contradiction with observations). In the published
version of the present paper this factor was omitted, due to an incorrect normalization of Eq. (33). We
thank Takahiro Tanaka for drawing our attention to this point.
we expect y > .79 with probability
and y > .07 with probability
P (y > .79) = .68
(1σ c.l.),
P (y > .07) = .95
(2σ c.l.).
13
(35)
(36)
We shall denote these two equations as the 1σ and 2σ confidence level predictions for y. Let
us now show how these translate into confidence level curves for the expected values of the
parameters ΩD and h. Here, and in what follows, ΩD will denote the present value of the
dark energy density parameter in our observable universe.
Let us first express the "observed" value of y, which we shall denote as y0, in terms of
ΩD and h. The density contrast at present is given by σ0 = G(x0, xrec)σrec, where, assuming
zrec ≫ 1, the growth factor is given by [17] G(x0, xrec) = x−1/3
dw
rec F (ΩD), with
F (ΩD) =
5
6
Ω−1/2
D
w1/6(1 + w)3/2 .
(37)
Therefore,
0
Z ΩD/(1−ΩD)
(cid:21)3
y0 =(cid:20) F (ΩD)
σ0
.
(38)
The linearized density contrast at present σ0 can be inferred from measurements of CMB
temperature anisotropies, as described e.g.
in [17, 28]. Since the spectrum is expressed as
a function of wavelength, the mass scale has to be converted into a length-scale. A halo
of mass M corresponds to a co-moving radius R(M) = (3M/4πρ0)1/3. The mean matter
density of the universe is given by ρ0 = 1.88 × 10−29ΩM h2 g/cm3, which leads to
R(M) = .95h−2/3Ω−1/3
M (cid:18) M
1012M⊙(cid:19)1/3
Mpc.
Assuming an adiabatic primordial spectrum of scalar density perturbations, characterized
by a spectral index n, we have
σ0(R) = (c100Γ)(n+3)/2δHK 1/2(R).
(39)
Here, c100 = 2.9979 is the speed of light in units of 100km s−1 and
Γ = ΩM h exp[−Ωb(1 + √2hΩ−1
M )]
is the so-called shape parameter, with Ωb the density parameter in baryons. For numerical
estimates, we shall take Ωbh2 ≈ .02. The dimensionless amplitude of cosmological pertur-
bations inferred from the COBE DMR experiment is given by [28, 29]
δH = 1.91 × 10−5
Ω−.80−.05 ln ΩM
M
[1 − .18(1 − n)ΩD − .03rΩD].
(40)
The parameter r denotes the ratio of tensor to scalar amplitudes. Note that the effect of
tensors is to make δH a bit smaller (although not very significantly). Finally,
exp[1.01(1 − n)]
p1 + r(.75 − .13Ω2
D)
K(R) =Z ∞
0
q(n+2)T 2(q)W 2(q Γ h R Mpc−1)dq,
14
h
0.85
0.8
0.75
0.7
0.65
0.6
0.55
0.6 0.65 0.7 0.75 0.8 0.85
ΩD
2
σ
1PSfrag replacements
σ
FIG. 2: Contours of the function y0(ΩD, h) given in (38), corresponding to the 1σ (lower curves)
and 2σ (upper curves) predictions represented by Eqs. (35-36). The excluded region lies to the left
of the curves. The thick solid lines assume that the dominant contribution to nciv is in galaxies of
mass M = MM W = 1012M⊙. For comparison, we show the predictions for different choices of the
mass. The short dashed curves correspond to the mass of the local group MLG = 4× 1012M⊙, and
the long dashed curves correspond to the mass of the bright inner part of our galaxy M = 1011M⊙.
A scale invariant spectrum of density perturbations is assumed.
where the transfer function is given by T (q) = (2.34q)−1 ln(1 + 2.34q)[1 + 3.89q + (16.1q)2 +
(5.46q)3 + (6.71q)4]−1/4 and the window function is given by W (u) = 3u−3(sin u − u cos u).
Substituting (39) in (38), and using ΩD + ΩM = 1, we obtain the function y0 = y0(ΩD, h).
Contour lines of this function, corresponding to the 1σ and 2σ predictions represented by
Eqs. (35-36), are plotted in Fig. 2, assuming that the dominant contribution to nciv is in
galaxies of mass M = MM W = 1012M⊙ (thick solid lines). We also consider the predictions
for different choices of the mass, as discussed in Section III. The short dashed curves cor-
respond to the mass of the local group MLG = 4 × 1012M⊙, and the long dashed curves
correspond to the mass of the bright inner part of our galaxy M = 1011M⊙. The effect of
a tilt in the spectral index is plotted in Fig. 3. Both of these figures ignore the effect of
tensor modes in the normalization (40). Tensor modes tend to lower the value of δH, and
hence they tend to make the bounds somewhat less stringent. The effect, however, is not
dramatic. Even for r as large as .5, the effect on the curves is comparable to the effect of
lowering the spectral index by .05.
Expressions similar to Eqs. (34)-(38) were already contained in the exhaustive analysis of
the problem given by Martel, Shapiro and Weinberg (MSW) in [17], where σrec was treated
as a fixed parameter. However, our use of these expressions is somewhat different. MSW
15
0.6 0.65 0.7 0.75 0.8 0.85
ΩD
2
σ
1
σ
PSfrag replacements
h
0.85
0.8
0.75
0.7
0.65
0.6
0.55
FIG. 3: Effect of a tilt in the spectral index of density perturbations. As in Fig. 2, the thick solid
lines correspond to a scale invariant spectrum n = 1, and a mass M = MM W = 1012M⊙. The
long dashed line and the short dashed lines correspond to tilted spectra, with n = .95 and n = .9
respectively.
noted that the existing observations indicate a value of ΩD ∼ 0.6 − 0.7 and used Eq. (33),
with h = 0.7 to show that this range corresponds to probabilities from 2% to 12%, depending
on the values chosen for the galactic scale M and the spectral index of perturbations n. They
concluded that "anthropic considerations do fairly well as an explanation of a cosmological
constant with [ΩD] in the range 0.6 − 0.7". However, one cannot help but feel disappointed
by the somewhat low values of the probabilities.
Our approach here is that anthropic models should be used as any other models -- to
make testable predictions. Thus, the goal is not so much to explain the value of ΩD after it
is determined by observations, but to predict that value at a specified confidence level. The
contour lines in Figs. 2,3 indicate the 1σ and 2σ predictions of the model. If M ∼ MM W
proves to be the relevant mass giving the dominant contribution to nciv, then the currently
favoured model with ΩD ≈ .7 and h ≈ .7 is virtually excluded by the anthropic approach at
the 2σ level. Instead, this approach favours lower values of h and higher values of ΩD.
These predictions can be turned around. If the values ΩD ≤ 0.7, h ≥ 0.7 are confirmed
by future measurements, then our model will be ruled out at a 95% confidence level, again
assuming M ∼ MM W and a scale invariant spectrum. For a tilted spectrum, slightly lower
values of ΩD are allowed at the same confidence level. The observational situation at the
time of this writing is far from being clear. CMB and supernovae measurements yield [23, 24]
ΩD ≈ 0.7 , while the observations of galaxy clustering give [38] ΩM = 0.18 ± 0.8, and thus
ΩD ≈ 0.8.
16
VI. THE FUTURE OF THE UNIVERSE
We finally discuss the anthropic prediction which is not likely to be tested any time soon.
In all anthropic models, ρD can take both positive and negative values, so the observed
positive dark energy will eventually start decreasing and will turn negative, and our part of
the universe will recollapse to a big crunch.
To be specific, we shall consider a scalar field model with a very flat potential. In the
anthropic range (1), the potential can be approximated as a linear function,
where V ′
the evolution is described by the usual slow roll equations
0 is a constant and we have set φ = 0 at V = 0. Once the dark energy dominates,
V (φ) ≈ −V ′
0φ,
(41)
3H φ = V ′
0,
H 2 =
8π
3m2
p
V ′
0φ,
where H = a/a and a(t) is the scale factor. The solution of (42), (43) is
φ(t) = −φ∗[1 − (t/t∗)]2/3,
where −φ∗ is the present value of φ and
a(t) = exp[4πm−2
p (φ2
∗ − φ2(t))],
t∗ = 8πtD(φ∗/mp)2
(42)
(43)
(44)
(45)
(46)
is the time from the present to the beginning of recollapse.
The slow roll condition (6) implies that φ∗ & mp. As we discussed in Sec. IV, we do not
expect this condition to be only marginally satisfied, and thus φ∗ ≫ mp. Then it follows from
Eqs. (46) and (45) that t∗ ≫ 8πtD and therefore we should expect our region of the universe
to undergo accelerated expansion for at least another trillion years before recollapse.10
The slow roll approximation breaks down at φ ∼ −mp, so the above equations cannot be
used to describe the evolution at φ > 0, where the potential becomes negative. A general
analysis of models with negative potentials has been given in [36], where it is shown that at
φ2 ≫ V (φ).
φ ≫ mp the dynamics becomes dominated by the kinetic energy of the field,
The corresponding evolution is described by
φ(t) =
ln(tc − t) + const,
mp√6π
a(t) ∝ (tc − t)1/3,
(47)
(48)
where tc is the time of the big crunch. The linear approximation (41) for the potential breaks
down at sufficiently large φ, but in this regime the form of the potential is unimportant and
Eqs. (47), (48) still apply.
10 This is in contrast with the model of Kallosh and Linde [13] discussed in Section IV, where the universe
is expected to recollapse within 10-20 billion years.
17
During the dark energy dominated expansion, the ordinary nonrelativistic matter is di-
luted by the exponential factor (45). When the contraction starts, the density of matter
begins to grow as ρM ∝ (tc − t)−1. However, the kinetic energy of the field φ grows much
φ2 ∝ (tc − t)−2, and thus ordinary matter forever remains a subdominant component
faster,
of the universe.
VII. CONCLUSIONS
We now summarize the predictions that follow from the anthropic approach to the CCP's.
(1) In the simplest models where the dark energy density takes discrete values, or where
the dark energy density is due to the potential energy of a single scalar field, the dark energy
equation of state is predicted to be that of the vacuum,
pD = wρD,
(49)
where w = −1 with a very high accuracy. This distinguishes the anthropic models we
discussed here from other approaches, such as quintessence [21] or k-essence [22].11
(2) The anthropic predictions for the dark energy density ΩD and for the Hubble parame-
ter h are given in Figs. 2 and 3 of Section V12. We show the areas in the Ωd−h plane that are
excluded at 1σ and 2σ confidence levels. The excluded areas depend on the assumed galactic
mass M and on the spectral index n of the density fluctuations. For M = MM W = 1012M⊙
the currently popular values ΩD = 0.7, h = 0.7 are marginally excluded at 2σ confidence
level for a scale invariant spectrum n = 1. Lowering the spectral index relaxes the bounds
somewhat. For h > 0.65 and n > .95, the 1σ prediction is ΩD > 0.79. These anthropic
constraints get weaker when the relevant mass scale M is increased. For example, with
M = 4 × 1012M⊙ a value as low as ΩD = 0.63 is still allowed at the 2σ level for a scale
invariant spectrum. The 1σ prediction in this case is ΩD > 0.78 (for h = 0.65).
(3) Conditions for intelligent life to evolve are expected to arise mainly in giant galaxies
that form (or complete their formation) at low redshifts, zG . 1.
(4) The accelerated expansion will eventually stop and our part of the universe will
recollapse, but, at least in the framework of the simplest models, it will take more than a
trillion years for this to happen. Of course, this prediction is not likely to be tested anytime
soon. 13.
The above predictions apply to models where both CCP's are solved anthropically. For
comparison, we may consider other models. For instance, it is conceivable that a small value
of the cosmological constant will eventually be explained within the fundamental theory.
(We note the interesting recent proposal by Dvali, Gabadadze and Shifman [40] in this
regard.) Even then, the coincidence problem will still have to be addressed. One possibility
11 For a discussion of multifield models, see [42]. There is a class of two field models where the above
prediction does not hold. These are the models where the prior distribution favours a large slope of the
potential. In general, the equation of state pD = w(z)ρD depends on the redshift z, and increases with
time as the dark energy fields pick up kinetic energy, on their way to negative values of the potential. The
prediction for w(z) in the general case has been discussed in [43, 44].
12 These predictions are implicit in the earlier analysis by Martel, Shapiro and Weinberg [17]
13 Again, this prediction is somewhat different in two field models where the prior distribution favours a
large slope of the potential [42]
V (φ) = Λ(2 − cosh√2φ)
(50)
is that ρD is truly a constant, while the amplitude of the density fluctuations σrec is a
stochastic variable. With some mild assumptions about the prior probability distribution
P∗(σrec), it can be shown [1] that most galaxies are then formed at about the time of vacuum
domination. In this class of models, predictions (1) and (3) still hold, while the other two
predictions no longer apply.
Another possibility has been recently discussed by Kallosh and Linde [13]. They assumed
18
an M-theory inspired potential
with a stochastic variable Λ. An interesting property of this potential is that its curvature
is correlated with its height (at φ = 0). As a result, the universe tends to recollapse
within a few Hubble times after the dark energy comes to dominate. Assuming that other
contributions to the vacuum energy are somehow cancelled (that is, that the old CCP
is solved by some unspecified mechanism), Kallosh and Linde argue that the coincidence
tD ∼ tI is to be expected, where tI is the time it takes intelligent life to evolve (they assume
it to be ∼ 1010 yrs). Predictions (1)-(3) are not applicable to this model. The model does
predict recollapse of the universe, but the corresponding timescale (∼ 1010 yrs) is much
shorter than the anthropic prediction (4).
Anthropic arguments are sometimes perceived as handwaving, unpredictive and unfal-
sifiable lore, of questionable scientific validity.
In our view, the results presented in this
paper should dispel this notion. Here, we have used the anthropic approach to make sev-
eral quantitative predictions, some of which may soon be checked against observations. It
should also be emphasized that, for the particular case of dark energy, there are at present
no alternative theories explaining both CCP's, or making generic predictions of comparable
accuracy.
The present bound on the equation of state parameter w from the CMB and supernovae
measurements is [41] w < −0.7, which is consistent with the anthropic prediction of w = −1.
The value of w = −1 is usually associated with a plain cosmological constant. However,
if in addition to this equation of state, observations confirm some of the other predictions
presented above, this may be taken as an indication that the dark energy is dynamical. Thus,
a better understanding of structure formation and galactic evolution may in fact reveal a
crucial property of dark energy, with important implications for particle physics.
Acknowledgements
We are grateful to Steven Barr, David Spergel and Rosanne Di Stefano for useful discus-
sions. This work was supported by the Templeton Foundation under grant COS 253. J.G. is
partially supported by MCYT and FEDER, under grants FPA2001-3598, FPA2002-00748.
A.V. is partially supported by the National Science Foundation.
[1] J. Garriga and A. Vilenkin, Phys. Rev. D64 023517 (2001).
[2] J.D. Brown and C. Teitelboim, Nucl. Phys. 279, 787 (1988). For recent discussions, see also
J. L. Feng, J. March-Russell, S. Sethi and F. Wilczek, Nucl. Phys. B602, 307 (2001); R.
Bousso and J. Polchinski, JHEP 0006:006 (2000); T. Banks, M. Dine and L. Motl, JHEP
0101:031 (2001).
19
[3] G. Dvali and, A. Vilenkin, Phys. Rev. D64, 063509 (2001).
[4] J. Garriga and A. Vilenkin, Phys. Rev. D61, 083502 (2000).
[5] B. Carter, In I.A.U. Symposium 63: confrontation of cosmological theories with observational
data (ed. M.S. Longair), p. 291. Dordecht:Reidel (1974).
[6] B. Carter, Phil. Trans. R. Soc. Lond. A 310, 347 (1983).
[7] J.D. Barrow and F.J. Tipler, the Anthropic Cosmological Principle (Oxford University Press,
New York, 1986).
[8] For an early attempt to apply anthropic arguments to the cosmological constant, see also
P.C.W. Davies and S. Unwin, Proc. Roy. Soc. 377, 147 (1981).
[9] S. Weinberg, Phys. Rev. Lett. 59, 2607 (1987).
[10] M. Tegmark and M. Rees, Astrophys. J. 499, 526 (1998).
[11] J. Garriga, M. Livio and A. Vilenkin, Phys. Rev. D61, 023503 (2000).
[12] A. Aguirre, Phys. Rev. D64, 083508 (2001).
[13] R. Kallosh and A. Linde, "M-theory, Cosmological Constant and the anthropic principle",
hep-th/0208157 (2002).
[14] A.Vilenkin, Phys. Rev. Lett. 74, 846 (1995).
[15] G. Efstathiou, M.N.R.A.S. 274, L73 (1995).
[16] S. Weinberg, in Critical Dialogues in Cosmology, ed. by N. G. Turok (World Scientific, Singa-
pore, 1997).
[17] H. Martel, P. R. Shapiro and S. Weinberg, Ap.J. 492, 29 (1998).
[18] V. Vanchurin, A. Vilenkin and S. Winitzki, Phys. Rev. D61, 083507 (2000).
[19] S. Weinberg, Phys. Rev. D61 103505 (2000); "The cosmological constant problems," Talk
given at 4th International Symposium on Sources and Detection of Dark Matter in the Universe
(DM 2000), Marina del Rey, California.
[20] J. Donoghue, JHEP 0008:022 (2000).
[21] P.J.E. Peebles and B. Ratra, Ap. J. Lett. 325, L17 (1988); R.R. Caldwell, R. Dave and P.J.
Steinhardt, Phys. Rev. Lett. 80, 1582 (1998).
[22] C. Armendariz-Picon, V. Mukhanov and P. J. Steinhardt, Phys. Rev. D 63, 103510 (2001).
[23] J. L. Sievers et. al., astro-ph/0205387; W.J. Percival et. al., astro-ph/0206256.
[24] S. Perlmutter et al., Ap.J. 483, 565 (1997); S. Perlmutter et al., astro-ph/9812473 (1998);
B. Schmidt et al., Ap.J. 507, 46 (1998); A. J. Riess et al., A.J. 116, 1009 (1998).
[25] W. L. Freedman et al., arXiv:astro-ph/0012376.
[26] W.H. Press and P. Schechter, Astrophys. J. 187,425 (1974).
[27] H. Martel and P.R. Shapiro, astro-ph/9903425.
[28] A. R. Liddle and D. H. Lyth, Cambridge, UK: Univ. Pr. (2000) 400 p.
[29] E. F. Bunn and M. J. White, Astrophys. J. 480, 6 (1997).
[30] H.J. Mo and S.D.M. White, astro-ph/0202393.
[31] E.M. Leitch and G. Vasisht, astro-ph/9802174.
[32] R.G. Abraham and S. van der Bergh, Science 293, 1273 (2001).
[33] M.I. Wilkinson and N.W. Evans, Mon. Not. Roy. Astron. Soc. 310, 645 (1999).
[34] P.J.E. Peebles, Principles of Physical Cosmology, Princeton University Press, Princeton, 1993.
[35] H.J. Mo, S. Mao and S.d.m. White, MNRAS 304, 175 (1999).
[36] G. Felder, A. Frolov, L. Kofman and A. Linde, hep-th/0202017.
[37] M. Mac Low and A. Ferrara, astro-ph/9801237.
[38] N. Bahcall et. al. astro-ph/0205490.
[39] S. Bludman, Nucl. Phys. A663-664,865 (2000).
[40] G. Dvali, G. Gabadadze and M. Shifman, hep-th/0202174.
[41] J.R. Bond et. al., astro-ph/0210007.
[42] J. Garriga, A. Linde and A. Vilenkin, arXiv:hep-th/0310034.
[43] R. Kallosh, J. Kratochvil, A. Linde, E. V. Linder and M. Shmakova, JCAP 0310 (2003) 015
[arXiv:astro-ph/0307185].
[44] S. Dimopoulos and S. Thomas, Phys. Lett. B 573 (2003) 13 [arXiv:hep-th/0307004].
20
|
astro-ph/9709283 | 1 | 9709 | 1997-09-28T18:24:12 | Image Separation vs. Redshift of Lensed QSOs: Implications for Galaxy Mass Profiles | [
"astro-ph"
] | Recently, Park and Gott reported an interesting observation: image separation of lensed QSOs declines with QSO redshift more precipitously than expected in any realistic world model, if the lenses are taken to be either singular isothermal spheres or point masses. In this Letter I propose that the observed trend arises naturally if the lensing galaxies have logarithmic surface mass density profiles that gradually change with radius. If the observed lack of central (odd) images is also taken into account, the data favor a universal dark matter density profile over an isothermal sphere with a core. Since the trend of image separation vs. source redshift is mostly a reflection of galaxy properties, it cannot be straightforwardly used as a test of cosmological models. Furthermore, the current upper limits on the cosmological constant may have to be revised. | astro-ph | astro-ph |
Mon. Not. R. Astron. Soc. 000, 000 -- 000 (1997)
Printed 28 December 2018
(MN LATEX style file v1.4)
Image Separation vs. Redshift of Lensed QSOs:
Implications for Galaxy Mass Profiles
Liliya L. R. Williams⋆
Institute of Astronomy, Madingley Road, Cambridge, CB3 0HA
Accepted for publication in MNRAS
ABSTRACT
Recently, Park and Gott reported an interesting observation: image separation of
lensed QSOs declines with QSO redshift more precipitously than expected in any
realistic world model, if the lenses are taken to be either singular isothermal spheres
or point masses. In this Letter I propose that the observed trend arises naturally if the
lensing galaxies have logarithmic surface mass density profiles that gradually change
with radius. If the observed lack of central (odd) images is also taken into account,
the data favor a universal dark matter density profile over an isothermal sphere with
a core. Since the trend of image separation vs. source redshift is mostly a reflection of
galaxy properties, it cannot be straightforwardly used as a test of cosmological models.
Furthermore, the current upper limits on the cosmological constant may have to be
revised.
Key words: gravitational lensing -- cosmology -- galaxies: structure
1
INTRODUCTION
Since gravitational lensing takes place over cosmological dis-
tances, it can in principle be used to measure cosmologi-
cal parameters, such as mass density, Ω, and cosmological
constant, Λ. For example, the frequency of multiply imaged
QSOs is a sensitive function of Λ, and therefore the observed
abundance of lensed QSOs has been used to place upper lim-
its on Λ (Turner 1990; Fukugita et al. 1992; Kochanek 1996).
Another example is the relation between image separation,
θim and source redshift, zs of multiply split QSOs, which
mostly depends on the sum of Ω and Λ, and thus should
serve as a good indicator of the curvature of the Universe.
(Gott, Park & Lee 1989; Fukugita et al. 1992, Park & Gott
1997, hereafter PG97). If the Universe is flat and galaxies
are singular isothermal spheres, the θim -- zs relation should
be flat, i.e. independent of source redshift. If the Universe
is open, there should be a mild decline of image separations
with redshift; a ∼ 20% decline is expected between zs ∼ 1.5
and 5, in an extreme Ω = 0 case.
The mildness of the expected θim -- zs trend can be ex-
plained as follows. For any given lens, and fixed lens and
source redshifts the image separation of images with compa-
rable brightness is quite independent of source impact pa-
rameter. Since optimal lens redshift for any high redshift
source is roughly the same, zl ∼ 0.5, and angular diame-
ter distances vary little past z ∼ 0.5, the observed image
separation should change little with zs. If a population of
lenses is considered, having a range of properties, like scale
lengths and velocity dispersions, there will be a correspond-
ing spread in image separations, but still little or no trend
with zs.
Yet the observed θim of lensed QSOs declines strongly
with zs (See Figure 1 of PG97, or Figure 2 of this paper.)
PG97 present a list of source redshifts and image separations
of 20 multiply imaged QSOs. The source redshifts span a
range between 1.4 and 4.5, and image separation between
0.5′′ and 7′′. There is a drop of at least a factor of 4 in image
separation from the low-zs to high-zs cases. According to
PG97, in a flat Universe such distribution can be ruled out
at 99% confidence level, while in the most extreme empty
Universe, it is ruled out at 97%. To make the confidence
level decrease below 95%, one would need to 'remove' 2 -- 3
largest separation or highest redshift cases from the sample.
(See PG97 for a discussion of statistical significance, possible
errors, etc.)
In this Letter, I suggest that the trend, which is due
to the lack of high redshift wide separation lensed QSOs is
naturally reproduced if a set of three conditions is met, (i)
galaxies' central mass profiles have logarithmic slopes that
change with radius, (ii) there is a dispersion in galaxy prop-
erties, like central surface mass densities or velocity disper-
sion, and (iii) the characteristic length scale of galaxy dark
matter halos varies with galaxy luminosity as r ∝ La, where
0 < a <∼ 0.5.
⋆ Email: [email protected]
c(cid:13) 1997 RAS
2
L. L. R. Williams
2 MODEL
In this section I describe the lensing model, and introduce
some simplifying assumptions about the lenses, galaxies, and
sources, QSOs. I do not aim to produce a detailed model be-
cause the small number of lensed QSOs in the current sample
does not necessitate it. I need not assume any particular cos-
mological model, because the effect implied by the present
model is considerably stronger than that of cosmology.
2.1 Model Assumptions
2.1.1 Lenses
I consider two types of galaxy mass profile, an isothermal
sphere with a core, ISC, and a universal dark matter pro-
file, NFW, derived from numerical simulations of Navarro,
Frenk & White (1995, 1996). These profiles were chosen be-
cause their logarithmic surface mass density gradually flat-
tens with decreasing radius. The reason for using these,
as opposed to constant power law profiles, like a singular
isothermal sphere, is discussed in Section 3. Both ISC and
NFW provide a plausible description of the dark matter ha-
los of galaxies. ISC is commonly used because of the ob-
served shapes of the rotation curves of spiral galaxies. Addi-
tionally, based on the results of the HST Snapshot Survey,
Maoz & Rix (1993, hereafter MR93) conclude that ellipticals
must possess ISC-type dark matter halos in order to repro-
duce the observed image separation distribution of multiply
imaged QSOs. The NFW profile has been recently claimed
as a universal dark matter halo profile describing halos of
gravitationally bound structures over 4 orders of magnitude
in mass. It is also supported on theoretical grounds (Evans &
Collett 1997). I assume that all galaxy-lenses are circularly
symmetric.
A description of the lensing properties of these profiles
can be found in Schneider et al. (1992), Bartelmann (1996),
and Williams & Lewis (1997). It suffices to say here that
ISC is roughly flat within core radius rc, and has an isother-
mal density profile outside. The vertical normalization of
the profile is fixed by its central surface mass density, κ0,
in terms of critical density for lensing, Σcrit. NFW is sin-
gular at the centre, its logarithmic slope gradually steepens
from ρ ∝ r−1 at the centre, to ρ ∝ r−3 at large radii, and
is isothermal around the characteristic scale length, rs. The
vertical normalization of its projected surface mass density
is proportional to κs, which is given in terms of Σcrit.
As was pointed out by PG97, the scatter in the observed
θim -- zs points is large. One of the main sources of scatter is
probably the spread in galaxy-lens properties. For example,
a singular isothermal sphere lens of a given σv produces a
constant bending angle at the lens, and hence image sepa-
ration varies as Dls/Dos, where the D's are the lens -- source
and observer -- source angular diameter distances. If all lens-
ing galaxies had the same σv, one would expect the observed
angular image separations to be proportional to the corre-
sponding Dls/Dos values, whereas a plot of Dls/Dos vs. θim,
in 8 systems where both zs and zl are known reveals no cor-
relation at all, and a scatter in θim of a factor of a few larger
than the scatter in Dls/Dos. This strongly suggests that the
spread in galaxy-lens properties is important. Therefore I as-
sume a family of galaxy-lenses, each having the same scale
length, rc and rs for ISC and NFW respectively, but a range
of κ0 and κs, which is equivalent to a range in velocity dis-
persions.
I assume that rc and rs and galaxy luminosity function
do not evolve with redshift, at least in the interval most
relevant here, i.e. optimal lens redshift for high-z sources,
zl ∼ 0.3 − 1. This assumption is supported by the recent
spectroscopic observations of Lilly et al. (1995), which indi-
cate that the luminosity function of elliptical galaxies, i.e.
the galaxy population that is believed to provide the bulk
of the lensing optical depth (see MR93), evolves very little
from z ∼1 to the present.
When the source impact parameter is sufficiently small,
all the NFW lenses, and ISC lenses with κ0 > 1 produce
3 images, whereas the observed multiply imaged systems
mostly have an even number of images, either 2 or 4. This
is thought to be because the central (odd) image is demag-
nified below visibility. I will return to the importance of
the central odd image later, in Section 2.2. The circularly
symmetric lenses considered here cannot produce 5 image
systems; observed 4(+1) cases are a property of elliptical
lenses. However the present treatment will not suffer if only
symmetric lenses are considered.
For sources at large redshifts, the optimal lens redshift
increases very slowly with zs in all cosmologies except for
the most extreme Λ dominated cases. So I will assume that
zl is constant, independent of zs. The critical surface mass
density for lensing, Σcrit changes little with zs, if zl is fixed,
and sources are at high zs. For example, in an Ω=0.3 open
Universe, moving the source from zs=1 to 5 reduces Σcrit by
17%, if zl = 0.2. Therefore, I will assume that galaxy-lens
projected surface mass density is given by κ0 and κs, with
no dependence on redshifts.
2.1.2 Sources
Analytic fits to QSO luminosity function (QLF) usually take
the form of a double power law, with steep bright-end slope
and shallow faint-end slope; the transition occurs at some
characteristic luminosity, L0, which evolves with redshift,
and is commonly parameterized as L0(zs) ∝ (1+zs)α. Boyle
et al. (1990) derive a pure luminosity evolution for z <∼ 2
QSOs, with α=3.2, while Hewett et al. (1993) conclude that
QLF changes shape with redshift, and that evolution must
slow down beyond z ∼ 1.5 compared to the predictions of
Boyle et al. For 2 <∼ z <∼ 3 they derive α ∼1.5. QSO evolu-
tion at higher redshifts, up to 5 is less constrained, though
it is sometimes assumed that the shape of the QLF does
not evolve beyond z ∼ 2, while L0 evolves such that it
'slides' brightward along the high luminosity part of the
QLF (Wallington & Narayan 1993). I assume that at any
given redshift there are no QSOs brighter than L0, and that
QLF at L < L0 is NQSO(L) ∝ L−s, with s = 1.2 (MR93).
I adopt the α parameterization of QSO evolution for high
redshifts, and derive results for α=0, and 2.
2.2 Results
image separation and total
We are interested in two lensing properties of the galaxy-
lenses:
image magnification.
Given a galaxy-lens, a source can have a range of impact
parameters, each resulting in a different image magnifica-
tion, and image separation, even though the latter tends to
c(cid:13) 1997 RAS, MNRAS 000, 000 -- 000
stay rather constant, roughly equal to twice the Einstein
ring radius of the lens. To account for a range of source im-
pact parameters, I calculate the impact parameter weighted
averages of image magnification, hµi and image separation,
hθi as a function of κ0 or κs, for ISC and NFW respectively.
Figure 1 shows the results. Vertical normalization of im-
age separation is irrelevant for now. The important feature
is that for both profiles the image separation increases with
κ0 or κs, while total image magnification decreases. In other
words, given a population of galaxy-lenses with a range of κ0
or κs, image magnification goes inversely with image sepa-
ration. Now, image magnification couples to source redshift,
as follows. Flux of an L0 QSO is given by,
f ∝ µ(zs)L0(zs)/DL(zs)2
∝ µ(zs)(1 + zs)α/DA(zs)2(1 + zs)4
∝ µ(zs)(1 + zs)α−4,
(1)
where µ(zs) is total image magnification, and DL and DA
are the luminosity and angular diameter distances respec-
tively. In the second and third lines I used the fact that
DL = (1 + z)2DA ∝ (1 + z)2, since angular diameter dis-
tance stays roughly constant for high-zs sources. So to make
it into a magnitude limited sample, with a certain flim, an
L0 QSO at zs has to be magnified by at least
µ(zs) ∝ flim(1 + zs)4−α,
(2)
with fainter QSOs magnified more. Because larger redshift
QSOs undergo larger magnifications, a fixed flim of any
given survey translates into a lower limit on hµi as a func-
tion of zs, which in turn implies an upper limit on hθi, as
read off from Figure 1. The resulting relation is plotted as
solid (ISC) and dashed (NFW) curves in Figure 2, for two
values of QSO luminosity function evolution parameters, α;
0, and 2. Both sets of curves reproduce the upper envelope
of the observed points quite well. The paucity of observed
points does not allow to differentiate between α of 0 and 2,
or between ISC and NFW profiles.
The curves are scaled vertically and horizontally to
roughly match the observed distribution. Adjusting the scal-
ing in the vertical direction yields the angular size of rc
and rs respectively. For ISC angular core radius is 0.37′′,
while for NFW angular scale radius is 1.25′′. For typical lens
redshifts of 0.5 these translate into rc = 1.4h−1 kpc, and
rs = 4.6h−1 kpc. These values are roughly what one would
derive from modelling individual multiple-image systems us-
ing ISC/NFW profiles. Scaling the curves in the horizontal
direction gives the average minimum magnification of lensed
QSOs as a function of zs. For α = 0 models, hµi/flim =
0.039(1 + zs)4, and for α = 2, hµi/flim = 0.4(1 + zs)2; i.e. if
QSO characteristic luminosity does not evolve, a lower limit
on magnification of QSOs at z = 3 is hµi/flim=10, whereas
it is 6.4 if L0(zs) ∝ (1+zs)2. Surveys with brighter flim have
higher hµi, but the dependence on (1+zs) is weak, and most
existing surveys have flim of 1 -- 2 magnitudes within each
other.
Let us summarize the conclusions thus far. Roughly,
Figure 1 implies that average image magnification is in-
versely proportional to the galaxy central surface mass den-
sity, hµi ∝ κ−1
0,s; and that average image separation is di-
rectly proportional to the galaxy central surface mass den-
sity, hθi ∝ κ0,s. In general, the observed image separation
is also proportional to the characteristic scale length of the
c(cid:13) 1997 RAS, MNRAS 000, 000 -- 000
Image Separation of Lensed QSOs
3
Figure 1. Magnification and image separation vs. the central
surface density of two different types of lenses: ISC and NFW.
Both magnification and image separation are impact parame-
ter weighted averages of these quantities. The dotted line is the
impact parameter weighted average magnification of a singular
isothermal sphere. Note that for ISC and NFW profiles the mag-
nification and image separation go in the opposite sense with
increasing κ0 and κs, while for SIS the average magnification is
always a constant (see Section 3).
galaxy-lens, so hθi ∝ κ0,src,s, but since I have so far assumed
that all galaxies have the same characteristic scale length, I
have not used the latter dependence. The scale length can
be a function of central concentration, rc ∝ κβ
0,s. Combin-
ing these relations gives, hµi ∝ hθi− 1
1+β , and together with
eq.(2) it implies,
hθi ∝ (1 + zs)−(1+β)(4−α)
(3).
With β = 0 this equation basically reproduces the curves
in Figure 2. Note that β does not have to be 0, any value
between roughly -0.5 and 2 will do, i.e. the total exponent in
eq. (3) should be around -- few. I will return to the discussion
of β and its implications for the dark matter halos of galaxies
in Section 4.
Thus both ISC and NFW galaxy-lens models can re-
produce the observed θim -- zstrend, and neither is preferred
based on these observations alone. However, another widely
documented observation can break the tie; all multiply im-
aged QSOs lack the central odd-numbered image, i.e. ob-
served cases are either doubles or quadruples. This has long
been interpreted as evidence in favor of centrally condensed
galaxy profiles, which would demagnify the central image
below visibility. The other possible explanation, dust obscu-
ration in the centres of galaxies, is largely ruled out because
radio observations of multiply split QSOs also do not reveal
central images (Myers et al. 1995). I apply this argument
here: ISC model predicts that κ0 ∼ 1 − 2 models should
have central images of brightnesses comparable to the pri-
mary image, while with the NFW models the central images
of low κs lenses should be 5-10 times demagnified compared
4
L. L. R. Williams
Figure 2. Image separation vs. source redshift. The points are the
observed multiply imaged QSOs (taken from Table 1 of PG97);
stars represent the lenses from the HST Snapshot Survey (MR93),
circles are from other, ground based surveys. Predictions of the
isothermal sphere with a core, ISC and universal dark matter
halo, NFW models (labeled) were normalized vertically and hori-
zontally, as described in Section 2.2. Both ISC and NFW profiles
can account for the observed trend.
to the primary image. This rules strongly in favor of the
NFW model.
3
IMPLICATIONS FOR GALAXY MASS
PROFILES
The key feature of the galaxy lens population that allows
to reproduce the envelope is that image separation goes in-
versely with image magnification. This allows magnification
bias, which acts through image magnification, to couple im-
age separation to source redshift. This property of a lens
population is a direct consequence of the individual galax-
ies having a changing logarithmic slope, as in the ISC and
NFW models. Lenses that do not have this property, for ex-
ample singular isothermal spheres and point masses, cannot
reproduce the observed envelope.
For a singular isothermal sphere (SIS) the average im-
pact parameter weighted magnification can be calculated to
be 4, independent of the velocity dispersion, σv, while image
separation is proportional to σ2
v. So hµi, being a 'universal'
constant, does not couple to hθi, and thus SIS cannot repro-
duce the observed trend. The property that hµi is constant
is not unique to SIS, but can be generalized to any family of
mass profiles with no in-built scale length, like single power-
laws and point masses. In such models the only parameter is
the absolute mass normalization of the profile, which deter-
mines the size of Einstein ring radius, and hence the image
separation. However, hµi in such self-similar profiles is a con-
stant, independent of absolute mass normalization.
Therefore galaxies must have changing logarithmic sur-
face mass density profiles in order to be able to reproduce
the θim -- zs relation.
4 DISCUSSION
There are two interesting aspects in the observed distribu-
tion of image separations vs. source redshift. In this Letter I
set out to explain one of these, the lack of high redshift wide
separation lenses, i.e. the θim -- zs anticorrelation. It turns out
that the model presented in Section 2 also naturally explains
the other interesting feature of the θim -- zs plot, namely the
existence of wide separation cases; there are 8 lensed QSOs
with θim > 3′′. Most models currently found in the literature
have trouble predicting a large population of wide separation
lenses; for example MR93 predict the peak in θim distribu-
tion at <∼ 1′′, with vanishingly small number of cases above
3′′. So it has been argued that large θim cases are not due to
isolated galaxies, but are the result of cluster-aided galaxy
lensing. QSO 0957+561, with 6.1′′ between its two images,
is adduced as supporting evidence. However, if cluster-aided
galaxy lensing is the correct explanation, then wide sepa-
ration lenses should be found at all redshifts, and not just
at low zs, as is currently the case (Figure 2). Furthermore,
PG97 show that with the help of cluster lensing, image sep-
aration should increase with zs. Since this is clearly not ob-
served, cluster aided lensing is probably not important in
general (0957 must be a special case), and the large number
of wide separation lenses needs an explanation.
Let us derive the distribution of image separations as
predicted by the model of Section 2. (I will use the ISC
model because all the derivations required in this section
can be carried out analytically with this model. However
the general arguments presented here will also apply to the
NFW profile.) So far only the dark matter halo properties
have been discussed. These need to be related to the observ-
able galaxy properties, like luminosity. I will adopt relations
similar to those used in MR93 (their Section 2.2.1);
rc ∝ La, M/L ∝ Lb, hence κ0 ∝ L1−2a+b
(4).
Optical effective radii of ellipticals are observed to scale as
L1.2 (Lauer 1985), and since it is commonly assumed that
they are linearly proportional to the galaxies' dark mat-
ter scale length, a is usually taken to be 1.2. From the
observations of the fundamental plane of ellipticals (Kor-
mendy & Djorgovski 1989), b ≈ 0.25. Similar assumptions
were made by Kochanek (1996). Parameter β of eq. (3) is
β = a/(1 − 2a + b)†.
The frequency of multiply imaged QSOs as a function
of image separation is given by,
dF
dhθi
=
dN (L)
dL
[rcyr(L)]2
dL
flim (cid:17)s
dhθ(L)i (cid:16) hµ(L)i
,
(5)
where dN (L)/dL ∝ L−1.2e−L is the Schechter luminosity
function (Schechter 1976); L is in units of L∗, and I assume
a constant slope of -1.2. The lensing cross section for a galaxy
of luminosity L is given by the radial caustic in the source
† Note that MR93 and Kochanek (1996) assumed a=1.2, and
b = 0.25, and hence effectively their β = −1.04, which would not
reproduce the θim -- zs anticorrelation, see eq. (3).
c(cid:13) 1997 RAS, MNRAS 000, 000 -- 000
plane, whose radius is yr(L) = (κ2/3
0 − 1)3/2. The image
separation is approximately equal to the diameter of the
tangential critical curve in the lens plane, hθi ≈ 2rcpκ2
0 − 1.
Using eq. (4),
dhθ(L)i
dL
=
2rc,∗La−1(κ2
(κ2
0[1 − a + b] − a)
0 − 1)
.
(6)
Here, rc,∗ and κ0,∗ refer to the core radius and κ0 of an L∗
galaxy. I take rc,∗ to be 1.4h−1kpc, equal to the constant
core radius derived in Section 2.2. The value of κ0,∗ is then
calculated from rc,∗ and an assumed asymptotic line of sight
velocity dispersion, σ∗ = 300km s−1; κ0,∗ = 5. The last term
in eq. (5) is the magnification bias.
The function dF/dhθi of eq. (5) is plotted in Figure 3,
for a range of a corresponding to the allowable range of β, in
eq. (3), a=0.0, 0.2, 0.4, and 0.6; b = 0.25 was used through-
out. The angular splitting by an L∗ galaxy is denoted by an
arrow. These predicted distributions apply to the source red-
shift range where the upper envelope, described by eq. (3)
and plotted in Figure 2 does not cut in, i.e. for zs <∼ 2.5. The
solid histogram is the observed distribution for the same
source redshift cutoff. Even though the small number of cur-
rently observed lenses does not allow to make any precise
conclusions, it is apparent that a ∼ 0.4 reproduces the his-
togram quite well. The corresponding value of β is 0.9, which
is perfectly consistent with the allowed range of β, -0.5 to
2 (see eq. [3]). Thus, a ∼ 0.4 (β ∼ 0.9) reproduces both the
θim -- zs anticorrelation, and the observed frequency of wide
separation lenses. In this scenario, the θim ∼ 7′′ lenses are
due to ∼ 2L∗ galaxies, with a mass within 10h−1kpc of
3.7 1011h−1M⊙.
The major deviation of the present model from those
found in the literature is in the value of a, the power law
index relating the dark matter characteristic scale length of
a galaxy to its luminosity. As mentioned earlier, a is usually
assumed to be 1.2, while consistency with lensing observa-
tions in the framework of the present model implies a ∼ 0.4.
5 CONCLUSIONS
If the observed distribution of image separation vs. source
redshift of lensed QSOs is real and not a result of false
lenses, like physical QSO pairs, then it has the following im-
plications for galaxy dark matter halos: (i) galaxies must
have changing logarithmic surface mass density profiles,
with a 'universal dark matter' model being preferred over an
isothermal sphere with a core, (ii) there must be a spread in
galaxy properties, like the central surface mass density, (iii)
the characteristic length scale of dark matter halos should
scale with luminosity as La, where a ∼ 0.4.
The corollary of the present model is that higher zs mul-
tiply imaged QSOs are statistically more magnified, and are
preferentially lensed by galaxies with lower central surface
mass densities and intrinsic luminosities. These predictions
can be tested if each lens case is modelled individually, which
will become possible when the lensing galaxies of most mul-
tiply split QSOs are detected.
The strong dependence of the observed θim -- zs relation
on galaxy properties means that it cannot be easily used as
a test for the curvature of the Universe unless we know a lot
more about the mass distribution of galaxies. Furthermore,
c(cid:13) 1997 RAS, MNRAS 000, 000 -- 000
Image Separation of Lensed QSOs
5
Figure 3. Predicted image separation distribution for lensed
QSOs. The curves are labeled by parameter a, defined by rc ∝ La.
Vertical normalization is arbitrary. The predictions are valid for
zs where the magnification bias effect, discussed in Section 2.2
and shown in Figure 2, does not impose an upper limit on θim of
high-zs lensed QSOs, i.e. for zs <∼ 2.5. Accordingly, the observed
distribution to which the curves apply, is restricted to zs < 2.5
and is shown as the solid histogram. Values of a ∼ 0.4 seem to
describe the observed distribution adequately. In these models
image separation is a monotonically increasing function of galaxy
luminosity; image separation produced by an L∗ galaxy is indi-
cated by an arrow.
because of the implications the present model has on the
lensing properties of galaxy population, like the dependence
of the predicted frequency of lensed QSOs on parameter a
(see Figure 3), the current upper limits on Λ may have to
be revised.
ACKNOWLEDGMENTS
I am grateful to Joachim Wambsganss for useful suggestions.
I would like to acknowledge the support of PPARC fellow-
ship at the Institute of Astronomy, Cambridge, UK.
REFERENCES
Bartelmann, M. 1996, A&A, 313, 697
Boyle, B. J., Shanks, T., & Peterson, B. A. 1990, MNRAS, 243, 1
Evans, N. W., & Collett, J. L. 1997, Preprint, also available as
astro-ph/9702085
Fukugita, M., Futamase, T., Kasai, M., & Turner, E. L. 1992,
ApJ, 393, 3
Gott, R. J. III, Park, M.-G., & Lee, H. M. 1989, ApJ, 338, 1
Hewett, P. C., Foltz, C. B., & Chaffee, F. H. 1993, ApJ 406, L43
Kochanek, C. S., 1996. ApJ, 466, 638
Kormendy, J. & Djorgovski, G. 1989, ARA&A, 27, 235
Lauer, T. R. 1985, ApJ, 292, 104
Lilly, S. J., Tresse, L., Hammer, F., Crampton, D., & Le Fevre,
O. 1995, ApJ, 455, 108
6
L. L. R. Williams
Park, M.-G., & Gott, R. J. III. 1997, Preprint, also available as
astro-ph/9702173 (PG97)
Maoz, D., & Rix, H.-W. 1993, ApJ, 416, 425 (MR93)
Myers, S. T., et al. 1995, ApJ, 447, L5
Navarro, J. F., Frenk, C. S., & White, S. D. M. 1995, MNRAS,
275, 720
Navarro, J. F., Frenk, C. S., & White, S. D. M. 1996, ApJ, 462,
563
Schechter, P. L., 1976. ApJ, 203, 297
Schneider, P., Ehlers J. and Falco E. E. , 1992, Gravitational
Lenses, Springer-Verlag Press
Turner, E. L., 1990. ApJ, 365, L43
Wallington, S., & Narayan, R. 1993, ApJ, 403, 517
Williams, L. L. R., & Lewis, G. F. 1997, MNRAS, submitted
c(cid:13) 1997 RAS, MNRAS 000, 000 -- 000
|
astro-ph/9910336 | 1 | 9910 | 1999-10-19T05:32:43 | Conference highlights of "Toward a New Millennium in Galaxy Morphology" | [
"astro-ph"
] | A summary of the highlights of the conference, "Toward a New Millennium in Galaxy Morphology" is presented. In this review, I cover the major topics addressed at the conference, including both the observational and theoretical frameworks suggested for understanding the morphology of galaxies. | astro-ph | astro-ph |
Conference Highlights
Toward a New Millennium in Galaxy Morphology1
The most basic process in any observational science is taxonomy - the classification of objects by their
natural relationships. For almost a century, the Hubble classification has been the dominate morphological
paradigm, but it appears new ideas and systems are needed. Many galaxies at high redshift as well as a
fraction of galaxies in the nearby universe are unclassifiable in the Hubble sequence. Some nearby examples
are low surface brightness galaxies, including dwarf spheroidals, and galaxies in clusters, especially large cD
galaxies. With so much complexity in the galaxy population, new systems and ideas are needed.
We can ask at the turn of the century if optical images contain enough information to adequately
classify galaxies. Perhaps other wavelengths, such as the infrared, UV, radio and sub-mm are also needed
to fully cover the basic features of extragalactic systems. Spectroscopy of a galaxy, including kinematic
information, or HI line profiles might also be better morphological indicators than a galaxy's image.
To address these issues, over 90 astronomers clustered in Johannesburg, South Africa between 13-18
September to discuss the future of galaxy morphology, and how classification relates to galaxy evolution
and formation. Over 60 talks and a number of posters were presented during the meeting.
Several new methods of classifying galaxy images were proposed at the conference. These new methods
are in many ways different from the qualitative Hubble system. Not only are they quantitative in nature,
with parameters such as asymmetry, concentration of light, Fourier components, measures of bar strengths,
and automatic neural networks, but most do not reproduce the Hubble sequence, but attempt to modify or
replace it (Abraham, Bershady, Conselice, Frei, Takamiya, Windhorst). Physical morphological systems
based on comparing the form of a galaxy with its spectral-type look promising; these are similar to the
ideas proposed by W.W. Morgan in the 1950s. There is still however, no all encompassing system that can
relate all galaxies in different environments.
High redshift galaxies, such as those seen in the Hubble Deep Field (HDF) have brought about this
new interest in morphology. Although, most of the HDF, and medium deep survey (MDS) images of distant
galaxies are sampling rest frame ultraviolet morphologies, the most disturbed galaxies remain disturbed
when viewed with NICMOS on HST. Therefore, these galaxies are intrinsically different from nearby ones;
their morphologies are not the result of morphological k-corrections. We are however, only sampling the
young to middle aged stars when viewing galaxies in optical/near infrared wavelengths. Limiting ourselves
to these wavelengths, we are missing a large fraction of the light output from galaxies. The effects of dust
on the morphology of galaxies, and much of the gas is being missed.
Our view of the dusty universe is provided by ISO, IRAS, SCUBA and other instruments that probe
long-ward of 5µm (Cesarsky, Laurent, Mirabel, Sanders, Sauvage, Scoville). Not only does a significant
fraction of the light emitted from galaxies occur at these wavelengths, but because of dust effects several
features can only be seen by examining galaxies at these wavelengths. Hidden star formation in mid
ISOCAM images of nearby galaxies, prove that dust is a significant factor affecting the morphology of
galaxies. Luminous and ultra-luminous infrared galaxies are probably major galaxy mergers, with their
1Conference was held in Johannesburg, South Africa from 13-18 September 1999. Proceedings will be edited by D.L Block,
I. Puerari, A. Stockton and D. Ferreira and published in a special edition of Ap&SS
-- 2 --
optical morphology dominated by dust absorption. SCUBA sources seen at redshifts z < 6, are probably
analogs to nearby ULIGs and possibly are progenitors of ellipticals with L > 1012 L⊙. Many of these
questions will be answered with the next generation of sub-mm instrumentation (Combes).
Using near infrared images, particularly in the K band is useful for determining the underlying
structure of the old stellar populations in galaxies. Bars and other morphological features are more easily
detected in the NIR. The spiral structure of galaxies, specifically the pitch angle, changes between optical
and NIR wavelengths (Block). Using Fourier analysis, three principle NIR spiral galaxy morphologies are
found. The pitch angle of the spiral arms in these families, viewed in the NIR correlates with the shape of
velocity rotation profiles (Block), an effect not seen in optical wavelengths.
Using dust corrected rest frame optical fluxes, infrared, sub-mm fluxes from high redshift galaxies and
the sub-mm background, it is possible to obtain an estimate of the star formation history of the universe
(Lagache, Thompson). Determining the star formation rate in a galaxy is tricky, and estimates of star
formation rates are based on dust correction assumptions. Although, we do not know if the dust in high
redshift galaxies behaves the same as dust in nearby galaxies. To paraphrase Mayo Greenberg, "Galaxies
evolve, stars evolve, so why not dust?". Additionally, we do not know the full effects of stellar winds, SNe,
magnetic fields, and possibly other feedback mechanisms (Chu) to say nothing of the effects of black holes
and AGN phenomenon on the morphology of galaxies.
The masses of black holes correlate with the masses of their host galaxy's spheroidal components
(Macchetteo) suggesting a possible feedback mechanism and related formation. AGNs might also play an
important role in star formation in galaxies. The space density of AGNs is related to the star formation
history of the universe (Miley). Jet outflows are an important component for the formation of stars in
galaxies, and hence could have a significant effect on their morphologies.
These physical mechanisms can change how a galaxy looks in the general, but perhaps more detailed
features deserve a closer look. Rings, bars, and globular clusters all hold clues to galaxy formation. Rings
(Buta) can be located at resonances, are sites of star formation and are potentially triggered by bars, or
interactions with other galaxies. Bars also produce spiral arms, star formation, and drive secular evolution
in spiral galaxies. The bar frequency among spiral galaxies in the local population seen in the near infrared
is at about 60% (Eskridge, Knapen). The bar fraction decreases at higher redshifts (Abraham), suggesting
that bars might form later in the evolution of galaxies. The specific frequency and properties of globular
clusters, especially metallicity distributions, can also be used to determine the evolutionary history of
galaxies (B. Elmegreen).
An overlooked feature for morphological studies of galaxies is their gas content. While some forms of
ionized gas can be seen at optical wavelengths, most gas is detectable in the radio. HI 21cm lines images of
galaxies are generally more extended than in the optical, and more commonly show signs of interaction and
mergers than galaxies in optical light (Sancisi). Interactions between galaxies is a critical aspect for their
evolution, and any complete morphological system must include dynamical indices. The effects of mergers
and interactions on galaxies, including the intensity of the induced star formation, can vary depending on
the initial orbital conditions (Horellou).
HI gas dynamics can be used to determine the underlying physical nature of galaxy disks, including
dark matter halos. Recent observations suggest that dark matter halos are triaxial and possibly rotating in
some galaxies (Freeman). These rotating dark matter halos, predicted by CDM could be responsible for
the structure of HI disks. High resolution rotation curves also show that the luminous mass of galaxies
does not trace dark matter potentials in halos (Sofue). Vertical structures of edge-on disk galaxies are also
-- 3 --
inconsistent with slowly rising rotation curves (van der Kruit).
Knowing the structure of galaxies naturally leads towards learning how galaxies formed and evolved.
Observationally we know that the galaxy population looked different in the past (Abraham, Conselice,
Illingworth, Windhorst) but how did this evolution occur? Evidence for dissipative collapse, where galaxies
are created from the inside out, includes metallicity gradients that correlate with bulge luminosity, and
old bulges (Goudfrooij). On the other hand, properties of spirals from Sm to S0s suggest galaxies formed
through a secular process (Pfenniger). Additionally dynamical evolution of galaxies based on semi-analytical
and N-body simulates using likely initial conditions of the universe suggest galaxies formed hierarchically.
Although, dark matter halos in these models are too concentrated and fail to produce correct sizes for disk
galaxies (Steinmetz). There is however, observational evidence for hierarchical formation of galaxies at
high redshifts (Illingworth). Morphologies of certain spirals can also change in special environments such as
galaxy clusters through high speed galaxy interactions (Lake).
The observations and theory described above is beginning to reveal how galaxies were formed, evolved
and how fundamental properties relate to morphology. A good taxonomy system for extragalactic objects
should classify galaxies according to characteristics that best describe galaxy evolution; perhaps interactions,
star formation history, gas content, and dark matter/rotation curve profiles, are among these. Although
such a morphological system does not presently exist, the accumulation of observational data and the
further development of theory will suggest what parameters are the best for understanding the structure
and evolution of galaxies.
Christopher J. Conselice
University of Wisconsin, Madison
|
0804.2074 | 1 | 0804 | 2008-04-13T16:44:06 | Bohdan's Impact on Our Understanding of Gamma-ray Bursts | [
"astro-ph"
] | Bohdan Paczy'nski was one of the pioneers of the cosmological GRB model. His ideas on how GRBs operate and what are their progenitors have dominated the field of GRBs in the hectic nineties during which the distances and the origin of GRBs were revealed. I discuss here Bohdan's contributions in some historical perspective. | astro-ph | astro-ph |
**FULL TITLE**
ASP Conference Series, Vol. **VOLUME**, **YEAR OF PUBLICATION**
**NAMES OF EDITORS**
Bohdan's Impact on Our Understanding of Gamma-ray
Bursts
Tsvi Piran
Racah Institute for Physics, The Hebrew University, Jerusalem, Israel
91904
Abstract.
Bohdan Paczy´nski was one of the pioneers of the cosmological
GRB model. His ideas on how GRBs operate and what are their progenitors have
dominated the field of GRBs in the hectic nineties during which the distances
and the origin of GRBs were revealed. I discuss here Bohdan's contributions in
some historical perspective.
1. Prologue
I first met Bohdan in the summer of 1977 at the IOA. I don't remember the
exact date, but I remember the time. It was between 3am and 4am. Several
of us were trying to open the last remaining bottle of wine at the traditional
IOA summer (Argentinean) barbecue. Something did not work out. Bohdan
approached us and succinctly pointed out that the process will be much easier
if we remove first the plastic warp that covered the cork. We did. It worked.
This was a simple demonstration of Bohdan's ability to see the obvious simple
fact that everyone else around stared at - but no one sees.
Our next meeting took place several years later at Princeton. Bohdan and
his family moved to the Institute housing, where I lived as well. I have spent
many pleasant evenings visiting them. We enjoyed many discussions. When we
did not discuss life in America versus life in Europe those conversations turned
naturally to Gamma-Ray Bursts (GRBs) - a subject that fascinated both of us.
2.
Introduction - GRBs as we understand them today
I begin with a brief description of the present picture of GRBs and their mod-
els. This section is not intended to be a comprehensive review (see Piran 1999;
M´esz´aros 2002; Piran 2005; M´esz´aros 2006; Woosley & Bloom 2006; Zhang,
2007 for recent reviews and Fishman & Meegan, 1995; Piran 1995 for a histor-
ical perspective). My objective is to outline our current understanding so that
Bohdan's contributions and impact can be put into the right perspective and be
appreciated.
GRBs are short and intense (∼ 10−7 −10−5ergs/cm2 ) bursts of soft (10 keV
- 2 MeV) gamma rays coming from random directions in the sky. GRBs were
discovered accidentally in the late sixties by the Vela satellites - defense satellites
that were sent to monitor the outer space treaty. The discovery was reported
only in 1973 (Klebesadel et al.). In the early nineties the BATSE detector on-
board of the COMPTON-GRO satellite indicated that GRBs are cosmological
1
2
Tsvi Piran
(Meegan 1992, Mao & Paczy´nski 1992, Piran 1992). Towards the late nineties
the BeppoSAX satellite discovered X-ray afterglows (Costa et al., 1997). Op-
tical (van Paradijs et al., 1997) and Radio (Frail and et al., 1997) afterglows
followed shortly. Direct redshift measurement of the optical afterglow of GRB
970508 (Matzger et al., 1997) established the cosmological origin of GRBs. Ra-
dio scintillation (Goodman, 1997) and later VLBI observations (Taylor et al.,
2004) confirmed the relativistic bulk motion, predicted earlier by the fireball
model.
In the following years afterglow observations revealed a wealth of in-
formation on host galaxies (Djorgovski et al. 2003) and the positions of GRBs
within host galaxies, which was always within star forming regions (Paczy´nski
1998, Fruchter, 2006).
GRB belong to two groups (Kouveliotou et al. 1993) according to their
durations T90: Long GRBs (with T90 > 2sec) and short (with T90 < 2). Short
bursts are also typically harder and hence they are called at times short hard
bursts (SHBs). The association of several long GRBs with supernovae (Galama
et al. 1998; Bloom et al. 1999; Stanek et al. 2003; Hjorth, et al. 2003) confirmed
the Collapsar model (Woosley 1993; Paczy´nski 1998; McFadyen & Woosley,
1998) according to which (long duration) GRBs arise during the collapse of
massive stars. The origin of short GRBs is less certain. Unlike long GRBs, short
ones do not arise in star forming regions. They arise in both elliptical and spiral
galaxies (see e.g. Nakar 2007 for a review). The redshift distributions shows that
the observed short bursts are much nearer than the observed long ones (Guetta
& Piran, 2006; Nakar et al., 2006). These observations are consistent with
earlier suggestions that short GRBs arise from neutron star mergers (Paczy´nski,
1986; Eichler et al., 1999). Within this model the delay corresponds to the
gravitational radiation spiral in time. The delay is crucial also in positioning
the progenitors away from their birth places, possible even outside the original
galaxy.
The internal-external shocks model (also called the fireball model) (Good-
man 1986; Paczy´nski, 1986; Shemi & Piran, 1990; M´esz´aros & Rees 1992, Rees &
M´esz´aros, 1994, Narayan, Paczy´nski & Piran, 1992; Sari & Piran 1997, Wijers,
Rees & M´esz´aros, 1997) is depicted in Fig. 1. It suggests that GRBs involve
ultrarelativisitic (baryonic or Poynting flux) jets that emerge from compact ob-
jects, most likely black holes. Internal collisionless shocks or instabilities within
the relativistic outflow, that take place at 1014 − 1016cm from the central engine,
accelerate particles to relativistic velocities and generate strong magnetic field.
Synchrotron or SSC emission of these particles produce the prompt emission.
The outflow interacts later with the surrounding medium producing a relativis-
tic blast wave - a relativistic analogue of a Supernova remnant. This long lasting
blast wave accelerates particles and generates magnetic fields that produce the
multiwavelength afterglow. The predictions (Sari, Narayan & Piran, 1998; Sari,
Halpern & Piran, 1999) of this "standard model" agreed nicely the observed
light curves and spectra of GRB afterglows (Wijers & Galama, 1999; Kulkarni
et al., 1999; Panaitescu & Kumar 2002; Yost S.A., et al., 2003). Recent Swift
observations indicate that the picture is somewhat more complicated than what
is described here. Still for the purpose of this review this description is sufficient.
Bohdan's GRBs
3
Figure 1.
The standard internal-external shocks model of GRBs (see e.g.
Piran 1999; M´esz´aros 2002; Zhang and M´esz´aros 2004; Piran 2005; M´esz´aros
2006 for reviews). 1. Central engine and acceleration of the flow. 2. Coasting
phase; terminal Lorentz factor Γ > 100. 3. Internal shocks; prompt gamma-
ray emission . 4. The interaction of the outflow with the surrounding matter
produces the external shocks system with a blast wave propagating outwards
producing the afterglow.
3. Prehistory - GRBs before BATSE
The history of GRBs should be divided to two period. Pre and post BATSE. At
the pre-BATSE era (up to the late eighties and very early nineties) a consensus
formed that GRBs originate in magnetic instabilities on galactic neutron stars.
What was considered as a strong evidence for this model was the detection of
20 and 40 keV lines which were interpreted as cyclotron lines of a ∼ 1012 Gauss
field - a magnetic field expected in neutron stars.
Bohdan was skeptical. He never believed the observed 20 and 40 keV lines
e.g.(see Narayan, Paczy´nski & Piran, 1992) - insisting that there was never the
case that the same line was observed simultaneously from the same burst by
two different detectors. With no lines there was no compelling case for galactic
neutron stars and the road to cosmological sources was open.
If the bursts
are cosmological they release much more energy ∼ 1051ergs or more. A basic
problem with any cosmological model was that GRBs have non-thermal spectra
indicating that they are optically thin. However, the temporal variability (less
4
Tsvi Piran
than a fraction of a second) indicated that the emission region is small (less than
109cm). If 1051ergs of soft gamma-rays (with numerous photons above 500 keV)
are released from such a source there would be copious pair production that will
lead to an optically thick source. This is the so called "Compactness problem".
The "Compactness problem". Left:
Figure 2.
light curves for different
bursts. Variability on very short time scales is clearly seen. Right: The spec-
trum of GRB 910601. The nonthermal spectrum is clearly seen. If 1051ergs
of radiation with many photons above mec2 (as seen in this spectrum) are
released within a small region (as indicated by the temporal variability) nu-
merous pairs will form and the radiation-pairs plasma will be optically thin.
The breakthrough concerning the resolution of the "Compactness problem"
came in 1986 when two back to back papers where publisehd in the Astrophysical
Journal Letters by Jeremy Goodman (1986) and Bohdan Paczy´nski (1986). Both
papers describe an idealized problem in which a hot radiation, a fireball, with
T0 ≫ mec2 is released within a small region. They argues that the radiation will
form a radiation-pair plasma fluid that will expand rapidly. As the fireball
accelerates it cools (in its local frame). When the local temperature drops
below mec2 the pair begin to annihilate. There are enough pairs to keep the
fireball optically thick until the local temperature drops to ∼ 20keV when the
fireball become optically thin. By this time the fireball has reached a relativistic
motion with Γ ∼ T0/(20keV). The photons escape freely now. Because of the
relativistic motion the escaping photons will be seen by an observer at rest with
Bohdan's GRBs
5
approximately the original temperature T0. Jeremy solved the problem in the
impulse approximation, considering the case of an "explosion". Bohdan, on
the other hand, considered a steady state approximation in which the evolution
was approximated as a steady wind. The results were more or less similar.
The papers were the first to suggest that GRBs involve relativistic motion. A
concept that was verified more than ten years later first using radio scintillations
(Goodman, 1997) and later with direct VLBI measurements (Taylor et al., 2004).
While the two papers are closely related, they are not similar. Goodman focuses
on the physics and is very concerned with the fact that at the end the resulting
photons have a thermal like spectrum. Bohdan, on the other hand, is not so
worried about this issue and he uses this model to put forwards his idea that
GRBs are in fact cosmological. He even goes on to suggest, in the discussion
section of his work, that some reoccurring sources arise because of gravitational
lensing of cosmological events.
At that time we spent many hours discussing what could drive such events.
Our discussions focused on neutron star mergers. Objects that at the time
became popular among relativists as it was realized that they are prime candi-
dates for detection as sources of gravitational radiation. Our discussions evolved
around the final stages in which tidal interaction between the two stars will tear
them apart, and Bohdan borrowed here a lot from his work on binaries. Still
Bohdan was reluctant to publish a paper with a model describing mergers as
GRB sources. My guess was that what he really cared most at the time was
to "win the cosmological war" on the distances to the bursts. He was proba-
bly worried that linking the cosmological idea to the very speculative merger
model (Neutron star mergers were considered as somewhat esoteric events in
those days), might weaken his cosmological case. Bohdan speculated on this
merger possibility very briefly in two or three sentences in his 1986 paper and
returned to it (actually to a variant - black hole neutron star merger (Paczy´nski,
1991)) only in the early nineties. I went on and published these ideas, with a
detailed model on how a merger can produce a GRB, in 1989 (Eichler, Livio,
Piran & Schramm 1989). We wrote our joint paper on this subject only in 1992
(Narayan, Paczy´nski & Piran, 1992).
I will return to this paper later in this
talk.
4. BATSE, the Cosmological-Galactic War and the Great Debate
The Burst and Transient Source Experiment, BATSE, on board of the Compton-
GRO satellite, was expected to demonstrate that GRBs are galactic. BATSE's
spectrograph was supposed to detect the lines and its ability to localize the
bursts within a few degrees was sufficient to demonstrate that the bursts are
within the galactic disk. It is difficult to imagine the shock when BATSE's first
results were announced in the fall of 1991: No lines, Isotropic distribution
with no dipole moment (corresponding to our position relative to the galactic
center), and last but not least a paucity of weak bursts (Meegan et al., 1992).
The results were accepted with disbelief and at first were sort of ignored. Most
of the talks, at least most of the theoretical talks, given at the first Huntsville
symposium, that took place in October 1991, dealt with various aspects of the
galactic neutron star model. A whole session was devoted to magnetospheric
6
Tsvi Piran
instabilities on neutron stars. One speaker, J. P. Lasota, withdrew his talk on
a galactic model and stated that it became irrelevant in view of the new data.
Bohdan (Paczy´nski, 1992) and me (Piran, Narayan & Shemi 1992) were the only
speakers that dealt with cosmological models in that meeting.
Both Bohdan (Mau & Paczy´nski, 1992) and me (Piran 1992) rushed to
publish letters on the essential cosmological interpretation of BATSE's result.
We stressed that the combination of an isotropic distribution, that at that stage
was demonstrated by upper limits on the dipole moment of the distribution (see
e.g. Fishman et al.1994; Hakkila et al.1994) and the paucity of weak bursts,
which at that stage was not demonstrated by a full log N − log S distribution
(see Fig. 3) but by the very low hV /Vmaxi = 0.32 value (see e.g Fishman et
al.1994; Hakkila et al.1994) forced a cosmological distribution. The argument
was surprisingly simple.
If the sources are galactic their angular distribution
must show a significant dipole moment, unless (i) Their radial distribution is
extremely local (and we don't see the galactic structure) or (ii) The radial dis-
tances are very large so that the typical distance compared to our distance from
the galactic center is large. Possibility (i) is ruled out by the hV /Vmaxi value.
A small industry begum by proponents of the galactic models that threw GRB
sources to larger and larger distances in the galactic halo (e.g. Podsiadlowski
et al.1995). As the limits on the dipole moment become tighter and tighter
the average distance needed for the sources continuously increased. I vividly re-
member Bohdan joking that with a typical distance of 200kpc one might call the
model cosmological... Additionally, as the distances increased so did the energy
budget until even the galactic models suffered from the Compactness problem
(Shemi & Piran, 1993).
Bohdan was very excited during this period. He enjoyed the lively (at
times too lively) debate that went on. He told me that in order to prevent
misunderstanding or misrepresentations of his ideas he was signing, at that time,
all GRB referee reports that we wrote. He was most amused when Nature
published a galactic origin paper (Lingenfelter & Higdon 1992) and refused to
publish his rebuttal because "our readers will be confused if we published two
contradicting opinions". Both of us had a great time making fun of some of
the models. Surprisingly the fight went on for quite some time. At the second
Huntsville meeting that took place at the fall of 1993 more than half of the
audience still believed in galactic origin and this situation persisted even in the
"Great Debate" that took place in 1995.
The debate culminated in the "Great Debate" that too place in April 1995
at the Smithsonian in Washington, D.C., between Bohdan (Paczy´nski, 1995) and
Don Lamb (Lamb 1995). Bohdan's abstract presents his argument best: "The
positions of over 1000 gamma-ray bursts detected with the BATSE experiment
on board of the Compton Gamma Ray Observatory are uniformly and randomly
distributed in the sky, with no significant concentration to the galactic plane or
to the galactic center. The strong gamma-ray bursts have an intensity distri-
bution consistent with a number density independent of distance in Euclidean
space. Weak gamma-ray bursts are relatively rare, indicating that either their
number density is reduced at large distances or that the space in which they are
distributed is non-Euclidean. In other words, we appear to be at the center of
a spherical and bounded distribution of bursters. This is consistent with the
Bohdan's GRBs
7
Figure 3.
Right: The distribution of BATSE bursts on the sky, in galactic
coordinates. The isotropy is self evident now. Left: the log(N ) − log(S) plot.
The number of bursts with N with a peak flux larger than S. In a homogenous
distribution we should see a straight line with a slope of −3/2.
distribution of all objects that are known to be at cosmological distances (like
galaxies and quasars), but inconsistent with the distribution of any objects which
are known to be in our galaxy (like stars and globular clusters). If the bursters
are at cosmological distances then the weakest bursts should be redshifted, i.e.,
on average their durations should be longer and their spectra should be softer
than the corresponding quantities for the strong bursts. There is some evidence
for both effects in the BATSE data. At this time, the cosmological distance scale
is strongly favored over the galactic one, but is not proven. A definite proof (or
dis-proof ) could be provided with the results of a search for very weak bursts in
the Andromeda galaxy (M31) with an instrument 10 times more sensitive than
BATSE. If the bursters are indeed at cosmological distances then they are the
most luminous sources of electromagnetic radiation known in the universe. At
this time we have no clue as to their nature, even though well over a hundred
suggestions were published in the scientific journals. An experiment providing
1 arc second positions would greatly improve the likelihood that counterparts of
gamma-ray bursters are finally found. A new interplanetary network would offer
the best opportunity."
8
Tsvi Piran
It was the 75th anniversary of the Curtis and Shapley 'Great Debate' (where
the size of the Universe was contested just three years before Edwin Hubble made
his seminal discovery about its expansion and thus the size of the Universe). The
parallel of the Curtis/Shapley and GRB debates could not be more poignant:
just as few changed their minds about the size of the Universe after the 1920
debate as have changed their mind about the distances of GRBs after this one.
The audience was split roughly 50% 50% at the end of this discussion. I must
say that when I heard the results of the vote at this debate I was shocked. To
me at that time the question what are the distances of GRBs was not an open
question (Piran 1995).
Bohdan was very happy about this work. In a report of his achievements in
a NASA grant that he posts (in a very unusual manner) on astro-ph (Paczy´nski,
1996) he writes: "The research project: 'Models and Scenarios for Gamma-Ray
Bursts' resulted in a total of 20 published research papers. The central issue was
the distance scale to gamma-ray bursters, the issue brought up by the remarkable
discovery of the distribution properties of gamma-ray bursts by the BATSE on
Compton GRO. The last paper on the reference list is the PI's contribution to
the debate on the distance scale to gamma-ray bursts held in Washington DC on
22 Apr. 22 1995. When this project got started, right after the announcement
of the BATSE results at the conference in Annapolis in the fall of 1991, only a
small fraction of astrophysicists seriously considered the possibility that gamma-
ray bursts are at cosmological distances. By now the cosmological distance scale
has become a majority view, to a large extent because of the publications listed
in this final report. PI considers this to be the most important and lasting result
from the research supported by this grant."
The debate ended with the measurement of the redshift of the optical af-
terglow of GRB 970508 (Metzger et al., 1997). It is amusing to recall that even
after this observation "conspiracy theories" suggesting coincidence between this
GRB and the observed afterglow was put forwards for a while.
5. Building a theoretical model - how GRBs work
In 1986 Bohdan contributed to the early ideas on a thermal fireball. However,
it was clear that this idea is rather preliminary. The resulting emission from
this simple fireball has a quasi-thermal spectrum, quite unlike what is observed.
Additionally,it was clear that this picture of a pure radiation (and pairs) fireball
was too idealized.
5.1. From Radiation and Baryons to Ultra relativistic Baryonic out-
flow
The next question was what would be the effects of baryons on such a pure
radiation fireball. Here, once more we have worked in parallel on the same prob-
lem and the results were published more or less simultaneously (Abramowicz,
Novikov & Paczy´nski, 1990; Shemi & Piran, 1990). Both works have shown that
the addition of even a small baryonic load would change drastically the outcome.
The baryons will be dragged along and accelerated by the radiation field. If the
total baryonic load is not too small eventually all the initial thermal energy will
be transferred to the kinetic energy of the baryons. If the baryonic load is not
Bohdan's GRBs
9
too large, that is m < E0/c2 the final outcome will be a relativistic baryonic
outflow with Γ = E0/M (E0 being the total energy and M the baryonic mass).
This was a step in the right direction but it was not good enough. Relativistic
baryons can serve as a source of cosmic rays. But what we have to find a way
to convert their kinetic bulk energy back to radiation.
5.2. Astro-ph/9204001
In the spring of 1992 Bohdan gave a colloquium, on the cosmological origin
of GRBs at Harvard.
I was amused by the fact that talk was on March 5th
- the anniversary of the famous 1979 March 5th GRB. This was was the only
well localized GRB at that time, and it was in the LMC - giving credence to
the Galactic origin. By now we know that it was not a regular GRB but a
soft Gamma repeater (SGR) but this is another story. Bohdan stayed for a
whole week during which we wrote, together with Ramesh Narayan, "Gamma-
Ray Bursts as the death throes of massive binary stars" (Narayan, Paczy´nski &
Piran, 1992- NPP92).
This paper, astro-ph/9204001, is the first astro-ph.
It is a well known,
highly cited(more than 400 citations) paper. Most remember it because of the
serious attempt to build a consistent merger model for GRBs. However, in
addition to discussing neutron star mergers that it contained several new ideas
that helped shape our theoretical understanding of GRBs:
• Internal shocks
• GRBs are powered by accretion onto a newborn black holes.
• The duration of the GRB is determined by the activity of the inner en-
gine (accretion onto the black hole) while the duration of the pulses is
determined by the dynamical time scale of the source
• GRBs should be followed by a long lasting Afterglow
• Binary neutron stars has a long spiraling-in phase before they merge. As
they receive a large kick when the binary is born they can escape during
this phase from their host galaxy.
• 1015Gauss magnetic fields may exist at GRBs' inner engines.
Somewhat surprising, or maybe not, the paper was not easily accepted. The
first referee simply rejected it as (i) too speculative and (ii) irrelevant because
GRBs are galactic. The second one accepted it reluctantly stating: "It three
well respected well known scientists what to make fools of themselves who am I
to stop them. I was very happy to be referred to as a "respectable and well known
scientist" but I suspected that I gained this title because of my co-authors. I
was even happier that the paper was accepted. In retrospect given the "anti-
cosmological" atmosphere at that time we should have been grateful to have had
such a broad minded referee.
10
Tsvi Piran
5.3. From Matter to Light Internal Shocks
The most urgent question was how to convert the relativistic baryonic outflow
to radiation. The solution proposed in NPP92 was internal shocks: "These
ejecta should, through collisions with one another at large radii and low optical
depth, give a non Planckian spectrum by various nonthermal mechanisms. For
instance given the strong magnetic fields, synchrotron processes might naturally
produce the observed power-law spectrum.". This was of course just a short
exposition of the idea and Bohdan continues to elaborate on it later in 1993
together with Ghuohong Xu (Paczy´nski & Xu 1993). The basic idea is that
the source is irregular and it emits a wind with a variable Lorentz factor. An
idealized picture of such a wind is a situation when the outflow is in the form
of shells that moves with different Lorentz factors. Faster shells collide with
slower ones. Paczy´nski & Xu (1993) consider within this context proton proton
collisions that produce pions that decay producing Gamma-Rays. However this
won't be very efficient. Later on Rees & M´esz´aros (1994)suggested the picture
accepted now according to which the collisionless shocks that form accelerate
particles to very high energies and possibly also generate strong magnetic fields.
These shocks are the source of the observed prompt emission. The schematic
picture is depicted as a central part of Fig. 1 and in Fig. 4. It is central now
to our current model. However at the time the simpler external shocks picture
proposed a few years earlier by M´esz´aros and Rees (1992) - namely interaction
of the outflow with the surrounding matter- was considered as the source of
the prompt emission. Only in 1997 we (Sari & Piran, 1997) have shown that
external shocks cannot produced the observed variable light curves and as of
today internal shocks are the only known viable way to convert the kinetic
energy of the outflow to radiation and produce a highly variable light curve!
It should be stressed that the same ideas hold for Poynting flux dominated
outflow, a possibility that is considered seriously now as an alternative to the
baryonic outflow. The kinematic arguments that require internal process within
the outflow are applicable regardless of the exact nature of the outflow. The
main difference of course is that in the Poynting flux case instabilities such as
reconnection replace the collisionless shocks. But the basic feature that external
interaction cannot produce the variable light curves remains.
The idea of internal shocks is linked directly to another important issue.
In internal shocks the temporal structure of the burst is determined by the
inner engine. The duration is determined by the time that the source is active
while the dynamical time scale of the source dictates the short term fluctuations
time scale seen in the individual pulses. The need of a prolonged activity of
the inner has pointed out towards accretion onto a newborn black hole as the
maincandidate for the activity
5.4. From Matter to Light Afterglow
Clearly, not all the kinetic energy can be dissipated in internal shocks. Only
the relative kinetic energy can be dissipated there. The bulk center of mass
kinetic will remain. This leads immediately to the idea of an afterglow: "The
ejecta should much later also produce something similar to a supernova rem-
nant." (NPP92). With James Rhodas, (Paczy´nski & Rhodas, 1993) Bohdan
continues to elaborate on this idea. They use the analogy with supernova to es-
Bohdan's GRBs
11
Figure 4.
Internal shocks are produced from a variably flow (schematically
depicted as shells). The faster shells catch up with the slower ones and produce
the shocks. The system is, most likely, powered by an accretion disk and 1015
Gauss field could arise in this disk.
timate that GRBs will be followed by a radio transients. More detailed models
focusing on higher energies were suggested later by numerous authors, (M´esz´aros
& Rees, 1997; Wijers, Rees & M´esz´aros, 1997; Waxman, 1997; Sari, Piran &
Narayan, 1998). Still this was the first suggestion that GRBs will be followed
by a long duration afterglow.
The afterglow prediction was verified in 1997 with the discovery of X-ray
afterglow by BeppoSAX (Costa et al., 1997). This discovery was followed by
detection of optical (van Paradijs et al., 1997) and radio (Frail et al., 1997)
afterglows as well.
6. GRB progenitors - mergers and hypernovae
The question what makes GRBs is of course the central one for the whole prob-
lem. Once it was realized that the bursts are cosmological and the energy budget
was set to be around 1051 − 1052ergs it was clear that the sources involve a com-
pact object. Most likely the formation of the object and the release of its binding
energy, and if not that at least a significant catastrophic event. The neutron
star merger model, which is currently the most likely to work for short GRBs,
12
Tsvi Piran
was proposed already in the eighties (Paczy´nski, 1986; Eichler et al., 1999). For
a while it was thought that this is the model for all GRBs.
However, when BeppoSAX begun detecting X-ray afterglows and host galax-
ies were revealed, it turned out that GRBs (actually long GRBs, as BeppoSAX
detected afterglow only from long GRBs) are located within small irregular star
forming galaxies. Now we know that within these galaxies GRBs are located in
the highest star forming regions and that the rate of GRBs is roughly propor-
tional to the square of the rate of star formation (Fruchter et al., 2006). Bohdan
did not need much data to make a far reaching conclusion. At the fall of 1997
he concluded on the basis of just three well localized GRBs, 970228, 0970508
and 980828 that GRBs arise in the vicinity of star forming regions (Paczy´nski,
1998). He concludes that the merger model must be abandoned and GRBs must
be linked to death of massive stars. "There is tentative evidence that the GRBs
970228, 970508, and 970828 were close to star-forming regions. If this case is
strengthened with future afterglows, then the popular model in which GRBs are
caused by merging neutron stars will have to be abandoned, and a model linking
GRBs to cataclysmic deaths of massive stars will be favored."
In a typical manner Bohdan ignored the theoretical prejudice against this
idea. At the time baryonic contamination was a major concern. Within the
fireball model discussed above it was clear that a significant baryonic load will
result in a non-relativistic outflow which won't be able to drive a GRB. Collaps-
ing stars have large envelopes that could be a strong source of contamination.
However, observations are more important than theory and if the bursts are near
star forming regions they must involve stellar death. Bohdan outlines a model
based on Woosley's (1993) "failed supernova" on accretion and Blandford-Znajek
mechanism (1976) and suggests that GRBs operate like microquasars.
Once again Bohdan's ability to grasp the basic point from a dismal amount
of data leads to a great success. A few month later Galama et al., (1998)
discovered that a very powerful type Ic SN 1998bw is associated with GRB
980425! At the same time MacFadyen & Woosley (1999) demonstrate that a
sufficiently powerful jet can punch a hole in a stellar atmosphere, (provided that
the later is sufficiently small as would be the case if the Hydrogen envelope has
escaped). As more and more SN like bumps were discovered on GRB afterglow
light curves (Bloom et al., 1999) the model gained credibility. It was eventually
proven with GBR 030339/SN 2003dh where an 98bw like SN spectrum arose
just as expected from the GRB optical light curve (Staneck et al., 2003; Hjorth,
J., et al. 2003).
In the process Bohdan coined the name Hypernovae (Paczy´nski, 1998).
Bohdan anticipated that GRB associated collapse events are more powerful than
a regular supernovae and hence they should be given a name indicating that.
Hyper is clearly more powerful than Super. It turned out that the first GRB
associated supernova SN 1998bw was indeed much more powerful than average
and indeed GRB associated supernovae do show higher velocity ejecta and are
more powerful than other type Ic SNe.
Bohdan's GRBs
13
7. Some ideas for the future: Neutrinos, GRB remnants and optical
flashes preceding GRBs
Bohdan has left many ideas to be tested in the future.
7.1. GRB neutrinos
Already in 1994 Bohdan noticed (Paczy´nski & Xu, 1993) that internal shocks
within GRBs can produce high energy neutrinos. In this specific model proton-
proton collisions between protons from different shells produce pions,which in
turn produce the observed gamma-rays as well as 30 GeV neutrinos. In fact,
more energy is released in this case in neutrinos than in gamma-rays.
Proton-proton collisions are not very efficient and they can produce only
a weak GRB signal. To overcome this problem Paczy´nski and Xu proposed
that the bursts are very narrow beamed. As far as we understand GRBs today
both ideas are invalid. GRBs are beamed but the jets are much wider. Proton-
proton collisions are indeed not efficient enough to produce the observed prompt
emission. However, what we do take from this paper today is first the basic
concept of internal shocks within the outflows and second the idea that GRBs
are prime candidates for being sources of high energy neutrinos - an idea that is
generally accepted today (see e.g. Achterberg et al.(2007) for a description of a
recent search using AMANDA).
7.2. GRB remnants
Several times we have worked in parallel on related idea. The last one was in the
early 2000. It was realized at that time that GRBs are beamed (Rhodas, 1999;
Sari et al., 1999; Kulkarni et al., 1999). It was also realized that the afterglow, in
which the Lorentz factor is much lower is essentially less beamed. One expects
therefore, orphan afterglows. Afterglows whose GRB prompt emission does not
points towards us. An intriguing question was how to search for such orphan
afterglows. The late radio phase is naturally the longest one. But this phase is
the weakest. Additionally, if one waits too long the afterglow becomes spherical
and indistinguishable from a regular Supernova remnant.
However, for a period of 5000 years it is possible to see within the GRB
remnant a bipolar structure that is induced by the original nonspherical GRB
jets (Ayal & Piran, 2001). This will enable us to distinguish GRB remnants
from SNRs. Bohdan (Paczy´nski, 2001a) build on these ideas and estimated that
at any time there should be several dozen relatively nearby GRB radio remnants
which can be resolved with VLBA as being bipolar rather than spherical. He
suggested to combine this idea with other methods to detect imprints of GRBs
such as the effect of the original gamma-rays on the interstellar medium (Draine,
2000) and perform a systematic search for GRB remnants. Such a search was
not done yet. It is something that Bohdan has left for the future.
7.3. Optical Flashes
In recent years Bohdan was fascinated with the transient Universe (Paczy´nski,
2001). This was not surprising. After all both microlensing events and GRBs
to whom he devoted his research in recent years are transient phenomenon. He
kept constantly looking for new possible "flashes" that will shine in the night sky.
14
Tsvi Piran
Figure 5. Density profile of a GRB remnant as a function of µ ≡ M c2/E0,
where M is the accumulatd mass and E0 is the initial energy (from Ayal &
Piran, 2001).
Along this line Bohdan noticed an exciting idea of Beloborodov (2002) that the
prompt emission can cause a cascade of pairs at a distance of ∼ 1018cm from the
origin. The idea is simple. Some gamma-rays will interact with the surrounding
matter and will be reflected backwards. Each photon that is reflected backwards
will certainly interact with the outgoing radiation and produce a pair. The
produced pairs will increase the number of back scattered photons and will
accelerate the process causing a possible runaway. Bohdan suggested that with
the right conditions (Kumar & Panaiteschu, 2004) a cloud of pairs will form and
this cloud will be opaque to gamma-rays. Until this cloud is cleared away only
lower energy photons will be seen. Bohdan suggested, therefore, that an optical
flash will appear in such a case and that this flash will precede the gamma-rays
(Paczy´nski, 2001). He was worried, however, that such flashes might be missed,
as they appear before the GRB trigger, and he searched for method to detect
them.
8. Epilogue
Bohdan was dominant in convincing the community that GRBs are cosmolog-
ical. However, he also had far reaching contributions that shaped our current
Bohdan's GRBs
15
understanding of how GRBs operate. The scope of these contribution is best
realized by presenting once again Fig. 1, that describes the basic theory of
GRBs, but now with Bohdan's contributions superimposed on it. His ideas on
hypernova, links with star formation, internal shocks, afterglow that looks like
a SNR, and above all the cosmological origin of GRBs all paved the ways to the
present "standard" model.
Figure 6.
Bohdan's major contributions to the current GRB model.
The fast response satellite Swift that was launched three years ago have
changed to a large extend the simple picture that we had before. It turns out
that, like in other cases in Astronomy, the afterglow picture is more complicated
than what was originally thought. In particular the early X-ray light curve, as
seen by Swift is rather different from the previous expectations, showing unex-
pected rapid decline that is followed by a shallow phase before joining the more
familiar light curve at about 104sec (Nousek et al., 2006). At present it is not
clear what are the processes that control this light curve. One widely discussed
possibility is that energy is added to the blast wave during the shallow phase.
It is interesting to note that already in 1993 Bohdan discussed the possibility
that a significant amount of energy can come out from the inner engine in a
low Lorentz factor material (Paczy´nski & Xu, 1993). "The slower material with
a low Lorentz factor will be gradually added to the blast wave". This kind of
"energy injection" is in fact the leading interpretation today of this phase (see
e.g. Zhang et al., 2006; Granot, Konigl & Piran, 2006).
Another interesting issue is the fact that Swift has observed bursts that are
further out than what is expected if the bursts simply follow the SFR (Natarajan
et al., 2005,, Daigne et al., 2006; Guetta & Piran, 2007). There are more distance
bursts than the simple model suggests. This result is consistent with the fact
16
Tsvi Piran
that in space GRBs are more concentrated in high SFR regions, as if they follow
a higher power of the SFR. It could be that the relevant factor is low metalicity
(Fynbo, 2003) but the possibility of evolution of the luminosity function or
another unexpected factor cannot be ignored.
Other long standing open questions have existed before Swift and are still
with us: What is the exact working of the inner engine, what is the nature of
the relativistic outflow and how do collisionless shock accelerate particles and
generate magnetic fields? It is illuminating to read what were Bohdan's thoughts
about all that: "It is not likely that the concept of a GRB as a microquasar
powered by the Blandford & Znajek (1977) mechanism can be proven or disproven
on purely theoretical grounds. It is useful to realize, that while we have plenty
of sound evidence that Type II supernovae explode as a result of some 'bounce',
or whatever process following the formation of a hot neutron star, there is no
generally accepted physical process which would be efficient enough to make this
happen. The theoretical problem with the SN II explosions persists in spite of 2 or
3 decades of intense effort by a large number researchers. The problem is vastly
worse with the GRBs as they are 104 − 105 times less common than supernovae.
This might imply that a very special set of circumstances is necessary to generate
the suitable energetic explosion".
Bohdan was probably right. It will take time, a lot of detailed observations
and ingenious theoretical insight to figure out all those details. However, regard-
less of the different variants of the current model and the current observational
puzzles I am sure that Bohdan's basic ideas on how GRBs work will be with us
forever to stay.
Acknowledgments. This work was partially supported by an ISF center
for Excellence in High Energy Astrophysics.
References
Bohdan Paczy´nski publications on Gamma-Ray Bursts
Li, L.-X., & Paczy´nski, B. 2006, Improved correlation between the variability and peak
luminosity of gamma-ray bursts, MNRAS, 366, 219
Paczy´nski, B., & Haensel, P. 2005, Gamma-ray bursts from quark stars, MNRAS, 362,
L4
Gehrels, N., et al. 2005, Erratum: "The Swift Gamma-Ray Burst Mission" ApJ, 611,
1005 [2004]), ApJ, 621, 558
Stockdale, C. J., Van Dyk, S. D., Weiler, K. W., Sramek, R. A., Panagia, N., Rupen,
M. P., Paczynski, B., & Weiler, K. W. 2004, VLBA Observations of SN 2001em:
Supernova, Misdirected Gamma-Ray Burster, or Both?, Bulletin of the American
Astronomical Society, 36, 1464
Gehrels, N., et al. 2004, The Swift Gamma-Ray Burst Mission, ApJ, 611, 1005
Paczynski, B. 2001, Gamma-Ray Bursts at Low Redshift, Acta Astronomica, 51, 1
Paczynski, B. 2001, Gamma-ray burst - supernova relation, Supernovae and Gamma-
Ray Bursts: the Greatest Explosions since the Big Bang, 1
Paczy´nski, B. 2001, Monitoring Variability of the Sky, IAU Colloq. 183: Small Telescope
Astronomy on Global Scales, 246, 45
Paczy´nski, B. 2000, Monitoring All Sky for Variability, PASP, 112, 1281
Paczynski, B. 1999, Gamma-Ray Burst - Supernova Relation, ArXiv Astrophysics e-
prints, arXiv:astro-ph/9909048
Li, L.-X., & Paczy´nski, B. 1998, Transient Events from Neutron Star Mergers, ApJ,
507, L59
Bohdan's GRBs
17
Paczynski, B. 1998, Are Gamma-Ray Bursts in Star-Forming Regions?, ApJ, 494, L45
Paczynski, B. 1998, Gamma-Ray Bursts as Hypernovae, American Institute of Physics
Conference Series, 428, 783
Paczynski, B. 1996, Studies of Gamma-Ray Bursts Time Variability, Princeton Univ. Re-
port,
Paczynski, B. 1995, How Far Away Are Gamma-Ray Bursters?, PASP, 107, 1167
Wijers, R. A. M. J., & Paczynski, B. 1994, On the nature of gamma-ray burst time
dilations, ApJ, 437, L107
Paczy´nski, B. 1994, Cosmic Gamma-ray bursts, 26th Meeting of the Polish Astronomical
Society, 59
Paczynski, B., & Xu, G. 1994, Neutrino bursts from gamma-ray bursts, ApJ, 427, 708
Paczy´nski, B. 1994, Cosmic bursts of gamma radiation., Urania, 65, 41
Paczy´nski, B., Rhoads, J., & Xu, G. 1994, Radio and Neutrino Emission from Theoret-
ical Gamma-Ray Bursters, Gamma-Ray Bursts, 307, 542
Rhoads, J. E., & Paczynski, B. 1993, Radio Transients from Gamma-Ray Bursters,
BAAS, 25, 1296
Paczynski, B., & Rhoads, J. E. 1993, Radio Transients from Gamma-Ray Bursters,
ApJ, 418, L5
Stanek, K. Z., Paczynski, B., & Goodman, J. 1993, Features in the spectra of gamma-ray
bursts, ApJ, 413, L7
Paczy´nski, B. 1993, Gamma-Ray Bursts, Texas/PASCOS '92: Relativistic Astrophysics
and Particle Cosmology, 688, 321
Paczy´nski, B. 1993, Gamma-ray bursts., American Institute of Physics Conference Se-
ries, 280, 981
Narayan, R., Paczynski, B., & Piran, T. 1992, Gamma-ray bursts as the death throes
of massive binary stars, ApJ, 395, L83
Paczynski, B. 1992, Gamma-Ray Bursts: Facts and Fantasies, BAAS, 24, 751
Mao, S., & Paczynski, B. 1992, On the Galactic disk and halo models of gamma-ray
bursts, ApJ, 389, L13
Mao, S., & Paczynski, B. 1992, On the cosmological origin of gamma-ray bursts, ApJ,
388, L45
Paczynski, B. 1992, Estimating redshifts for gamma-ray bursts, Nat, 355, 521
Paczynski, B. 1992, On the galactic origin of gamma ray bursts, The Compton Obser-
vatory Science Workshop, 287
Paczy´nski, B. 1992, Gamma-ray bursts from colliding neutron stars., Los Alamos Work-
shop on Gamma-Ray Bursts, p. 67 - 74, 67
Paczynski, B. 1992, Gamma-ray bursts from colliding neutron stars, Gamma-Ray Bursts
- Observations, Analyses and Theories, 67
Paczynski, B. 1992, How Far Away are Gamma-Ray Bursters, Comments on Astro-
physics, 16, 241
Paczynski, B. 1992, Extragalactic scenarios for gamma-ray bursts, American Institute
of Physics Conference Series, 265, 144
Paczynski, B. 1992, GB 790305 as a very strongly magnetized neutron star, Acta As-
tronomica, 42, 145
Paczynski, B. 1992, On the two population model for gamma-ray bursts, Acta Astro-
nomica, 42, 1
Haensel, P., Amsterdamski, P., & Paczy´nski, B. 1991, Gamma-ray bursts from coalesc-
ing strange stars., Nuclear Physics B Proceedings Supplements, 24, 162
Haensel, P., Paczynski, B., & Amsterdamski, P. 1991, Gamma-ray bursts from colliding
strange stars, ApJ, 375, 209
Abramowicz, M. A., Novikov, I. D., & Paczynski, B. 1991, The appearance of highly
relativistic, spherically symmetric stellar winds, ApJ, 369, 175
Zdziarski, A. A., Svensson, R., & Paczynski, B. 1991, Bursts of gamma rays from
Compton scattering at cosmological distances, ApJ, 366, 343
Paczynski, B. 1991, Cosmological gamma-ray bursts, Acta Astronomica, 41, 257
18
Tsvi Piran
Paczynski, B. 1991, On the Galactic origin of gamma-ray bursts, Acta Astronomica, 41,
157
Paczynski, B. 1990, A test of the galactic origin of gamma-ray bursts, ApJ, 348, 485
Paczy´nski, B. 1989, Electromagnetic fireball for gamma-ray bursts., ST ScI-GSFC Work-
shop on Ultra-Hot Plasmas and Electron-Positron Pairs in Astrophysics, p. 47,
47
Paczynski, B. 1989, Comet Showers and Gamma-Ray Bursts, Nat, 337, 689
Paczynski, B. 1988, Gamma-ray bursts from cusps on superconducting cosmic strings
at large redshifts, ApJ, 335, 525
Paczynski, B., & Long, K. 1988, Distribution of intensities of gamma-ray bursts, ApJ,
333, 694
Paczynski, B. 1987, Gravitational microlensing and gamma-ray bursts, ApJ, 317, L51
Babul, A., Paczynski, B., & Spergel, D. 1987, Gamma-ray bursts from superconducting
cosmic strings at large redshifts, ApJ, 316, L49
Paczynski, B. 1986, Gamma-ray bursters at cosmological distances, ApJ, 308, L43
Other references:
Achterberg, A., et al. 2007, ApJ, 664, 397
Ayal, S., & Piran, T. 2001, ApJ, 555, 23
Beloborodov, A., 2002, ApJ, 565, 808
Blandford, R. D., & Znajek, R. L. 1977, MNRAS, 179, 433
Bloom, J. S., et al. 1999, Nat, 401, 453
Costa, E., et al. 1997, Nat, 387, 783
Daigne, F., Rossi, E., & Mochkovitch, R., 2006, MNRAS, 372, 1034
Djorgovski, S. G., et al. 2003, Proc. SPIE, 4834, 238
Draine, B. T. 2000, ApJ, 532, 273
Eichler, D., Livio, M., Piran, T., & Schramm, D. N. 1989, Nat, 340, 126
Fishman, G. J., et al. 1994, ApJS, 92, 229
Fishman, G. J., & Meegan, C. A. 1995, ARA&A, 33, 415
Frail, D. A., Kulkarni, S. R., Nicastro, L., Feroci, M., & Taylor, G. B. 1997, Nat, 389,
261
Fruchter A.S., et al., 2006, Nature, 441, 463
Goodman, J. 1986, ApJ, 308, L47
Fynbo, J. et al., 2003, A&A 406, L63.
Galama, T. J., et al. 1998, Nat, 395, 670
Goodman, J. 1997, New Astronomy, 2, 449
Hjorth, J., et al. 2003, Nature, 423, 847
Granot J., Konigl A. & Piran T., 2006, MNRAS, 370, 1946
Hakkila, J., Meegan, C. A., Pendleton, G. N., Fishman, G. J., Wilson, R. B., Paciesas,
W. S., Brock, M. N., & Horack, J. M. 1994, ApJ, 422, 659
Klebesadel, R. W., Strong, I. B., & Olson, R. A. 1973, ApJ, 182, L85
Kouveliotou, C., Meegan, C. A., Fishman, G. J., Bhat, N. P., Briggs, M. S., Koshut,
T. M., Paciesas, W. S., & Pendleton, G. N. 1993, ApJ, 413, L101
Kulkarni S. R., et al., 1999, Nature, 398, 389
Kumar, P., & Panaitescu, A. 2004, MNRAS, 354, 252
Lamb, D. Q. 1995, PASP, 107, 1152
Lingenfelter, R. E., & Higdon, J. C. 1992, Nat, 356, 132
Meegan, C. A., Fishman, G. J., Wilson, R. B., Horack, J. M., Brock, M. N., Paciesas,
W. S., Pendleton, G. N., & Kouveliotou, C. 1992, Nat, 355, 143
M´esz´aros P., Rees M.J., 1992, MNRAS, 257, 29P
Meszaros, P., & Rees, M. J. 1997, ApJ, 476, 232
M´esz´aros P., 2002, ARA&A, 40, 137
M´esz´aros P., 2006, Rep. Prog. Phys, 69, 2259
Metzger, M. R., Djorgovski, S. G., Kulkarni, S. R., Steidel, C. C., Adelberger, K. L.,
Frail, D. A., Costa, E., & Frontera, F. 1997, Nat, 387, 878
MacFadyen, A. I., & Woosley, S. E. 1999, ApJ, 524, 262
Bohdan's GRBs
19
Nakar E., et al., 2006, ApJ, 650, 281
Nakar E., 2007, PhR, 442, 166
Natarajan, P. et al., 2005, MNRAS 364, L8
Nousek J. A. et al., 2006, ApJ, 642, 389
Panaitescu A., Kumar P., 2002, ApJ, 571, 779
Piran, T. 1992, ApJ, 389, L45
Piran, T., & Shemi, A. 1993, ApJ, 403, L67
Piran T., 1995, in the Proceedings of "Some Unsoved Problems in Astrophysics", Prince-
ton April 1995, Eds. J. Bahcall and J. Osriker, Princeton University Press
Piran T., 1999, Phys Rep., 314, 575
Piran T., 2005, RMP, 76, 1143
Piran, T., Narayan, R., & Shemi, A. 1992, American Institute of Physics Conference
Series, 265, 149
Podsiadlowski, P., Rees, M. J., & Ruderman, M. 1995, MNRAS, 273, 755
Rees, M.J., & M´esz´aros P.,1994, ApJ, 430, 93
Rhoads, J.E., 1999, ApJ, 525, 737
Sari R., & Piran T., 1997, ApJ, 485, 270
Sari R., Piran T., & Narayan R., 1998, ApJ, 497, L17
Sari R., Piran T., Halpern J.P., 1999, ApJ, 519,
Shemi A., & Piran T., 1990, ApJ, 365, L55
Stanek, K. Z., et al. 2003, ApJ, 591, L17
Taylor, G. B., Frail, D. A., Berger, E., & Kulkarni, S. R. 2004, ApJ, 609, L1
van Paradijs, J., et al. 1997, Nat, 386, 686
.
Waxman, E. 1997, ApJ, 485, L5
Wijers R.A.M.J., Rees M.J., & M´esz´aros P., 1997, MNRAS, 288, L51
Wijers R.A.M.J., & Galama T.J., 1999, ApJ, 523, 177
Woosley S. E., 1993, ApJ, 405, 273
Woosley S. E., Bloom J. S., 2006, ARA&A, 44, 507
Yost S.A., et al., 2003, ApJ, 597, 459
Zhang, B., & M´esz´aros, P. 2004, IJMPA, 19, 2385
Zhang, B., Fan, Y. Z., Dyks, J., Kobayashi, S., M´esz´aros, P., Burrows, D. N., Nousek,
J. A., & Gehrels, N. 2006, ApJ, 642, 354
Zhang B., 2007, ChJAA, 7, 1
|
astro-ph/0203464 | 1 | 0203 | 2002-03-26T17:02:27 | The luminosity functions and stellar masses of galactic disks and spheroids | [
"astro-ph"
] | We present a method to obtain quantitative measures of galaxy morphology and apply it to a spectroscopic sample of field galaxies in order to determine the luminosity and stellar mass functions of galactic disks and spheroids. We estimate, for each galaxy, the bulge-to-disk luminosity ratio in the I-band using a two-dimensional image fitting procedure. Monte Carlo simulations indicate that reliable determinations are only possible for galaxies approximately two magnitudes brighter than the photometric completeness limit, leaving a sample of 90 galaxies with well determined bulge-to-total light ratios. We construct the luminosity functions of disks and spheroids and, using a stellar population synthesis model, we estimate the stellar mass functions of each of these components. The disk and spheroid luminosity functions are remarkably similar. We do, however, find evidence in the bi-variate luminosity function that spheroid-dominated galaxies occur only among the brightest spheroids, while disk-dominated galaxies span a much wider range of disk luminosities. Remarkably, the total stellar mass residing in disks and spheroids is approximately the same. For our sample, we find the ratio of stellar masses in disks and spheroids to be 1.3+/-0.2. Ongoing large photometric and redshift surveys will lead to a large increase in the number of galaxies to which our techniques can be applied and thus to an improvement in the current estimates. (abridged) | astro-ph | astro-ph |
The luminosity functions and stellar masses of galactic disks and
spheroids
A. J. Benson1 C. S. Frenk2 & R. M. Sharples2
1. California Institute of Technology, MC 105-24, Pasadena, CA 91125, U.S.A.
(e-mail: [email protected])
2. Physics Department, University of Durham, Durham, DH1 3LE, England
ABSTRACT
We present a method to obtain quantitative measures of galaxy morphol-
ogy and apply it to a spectroscopic sample of field galaxies in order to deter-
mine the luminosity and stellar mass functions of galactic disks and spheroids.
For our sample of approximately 600 galaxies we estimate, for each galaxy, the
bulge-to-disk luminosity ratio in the I-band using a two-dimensional image fit-
ting procedure. Monte Carlo simulations indicate that reliable determinations
are only possible for galaxies approximately two magnitudes brighter than the
photometric completeness limit, leaving a sample of 90 galaxies with well de-
termined bulge-to-total light ratios. Using our measurements of individual disk
and bulge luminosities for these 90 galaxies, we construct the luminosity func-
tions of disks and spheroids and, using a stellar population synthesis model, we
estimate the stellar mass functions of each of these components. The disk and
spheroid luminosity functions are remarkably similar, although our rather small
sample size precludes a detailed analysis. We do, however, find evidence in the bi-
variate luminosity function that spheroid-dominated galaxies occur only among
the brightest spheroids, while disk-dominated galaxies span a much wider range
of disk luminosities. Remarkably, the total stellar mass residing in disks and
spheroids is approximately the same. For our sample (which includes galaxies
brighter than M∗ + 2, where M∗ is the magnitude corresponding to the char-
acteristic luminosity), we find the ratio of stellar masses in disks and spheroids
to be 1.3 ± 0.2. This agrees with the earlier estimates of Schechter & Dressler,
but differs significantly from that of Fukugita, Hogan & Peebles. Ongoing large
photometric and redshift surveys will lead to a large increase in the number of
galaxies to which our techniques can be applied and thus to an improvement in
the current estimates.
-- 2 --
1.
Introduction
It has long been known that galaxies in the nearby Universe display a range of morpho-
logical characteristics that distinguish between disk-dominated (e.g. spiral) and spheroid-
dominated (e.g.
elliptical) galaxies, with many fine subdivisions within each class (e.g.
Hubble 1926; de Vaucouleurs 1959). The origins of these different types of galaxy and the
evolutionary connections, if any, between them are still unclear, although there is a wealth
of observational data and several proposed theories.
Traditionally, morphology has been assigned by human classifiers directly from galaxy
images, a process that is accurate only to within about two T-types (Naim et al. 1995).
More recently, computer-based algorithms have been developed which show a reasonable
correlation with "eyeball" estimates, at least for bright galaxies (e.g. Abraham et al. 1996).
Unfortunately, all these classifications tend to give considerable weight to detailed mor-
phological features, such as spiral arms or asymmetries in the image, and are difficult to
compare with current theoretical predictions which focus on simpler quantities such as the
total stellar mass or luminosity in the disk and spheroidal components. Here, we consider a
morphological quantifier that is more easily related to theoretical models.
It is widely believed that disks form by slow accretion of gas which acquired angular
momentum through tidal torques (Hoyle 1949; Peebles 1969), although whether this pic-
ture works in detail remains an open question (Navarro, Frenk & White 1995; Navarro &
Steinmetz 1997, 2000; van den Bosch, Burkert & Swaters 2001). Spheroids, on the other
hand, are thought to form either by a "monolithic collapse" (Eggen, Lyden-Bell & Sandage
1962; Jimenez et al. 1999), or as a result of mergers of pre-existing galaxies (Toomre 1977;
Barnes & Hernquist 1992 and references therein). Detailed theoretical predictions for the
statistical morphological properties of galaxies and their evolution have been calculated for
the hierarchical merging formation mechanism appropriate to cold dark matter cosmolo-
gies (Kauffmann, White & Guiderdoni 1993; Kauffmann 1995, 1996; Baugh, Cole & Frenk
1996a,b; Somerville, Primack & Faber 2001). Among other things, these models give the
relative luminosities and stellar masses of the spheroids and disks of galaxies.
In this work, we measure the I-band bulge-to-total light ratio, B/T, for a large sample
of galaxies with spectroscopic redshifts by fitting two-dimensional models to the observed
galaxy images. We use this information to estimate the spheroid and disk luminosity func-
tions, as well as the total stellar mass that resides in disks and spheroids. An earlier attempt
to estimate the relative contributions to the luminosity from spheroids and disks was carried
out by Schechter & Dressler (1987), based on "eyeball" estimates of the bulge-to-total ratio.
The remainder of this paper is arranged as follows. In §2 we describe our basic dataset.
-- 3 --
In §3 we introduce the method used to fit model images to the data and thereby extract
B/T (along with other interesting parameters) and describe how we estimate errors. In §4
we examine the accuracy of our technique and determine how well the B/T ratio can be
measured as a function of the apparent magnitude of a galaxy. We then compute luminosity
functions and total stellar masses for spheroids and disks. Finally, in §5 we present our
conclusions.
2. Data
The galaxy sample used in this work is that of Gardner et al. (1997). The reader
is referred to that work for a full description of the data. Here we summarize the most
important features of the dataset.
Imaging of two fields of total area 10 deg2 was carried out in the B, V and I bands
using the T2KA camera on the Kitt Peak National Observatory (KPNO) 0.9-m telescope,
resulting in images with 0.68-arcsec per pixel. Exposures of 300s reached 5σ detection depths
of B=21.1, V=20.9 and I=19.6 in 10-arcsecond circular apertures. Imaging was also carried
out in the K-band using the IRIM camera on the KPNO 1.3-m telescope resulting in 1.96-
arcsecond per pixel images, and a 5σ detection depth of K=15.6 in a 10-arcsecond circular
aperture. The positions of the fields were chosen randomly (the field centers are RA 14h15m,
Dec. +00◦ and RA 18h0m, Dec. +66◦). The I-band images, which we will use in this work,
were bias-subtracted, flattened using twilight flats and with median sky flats. Objects were
identified with the sextractor program (Bertin & Arnouts 1996), using a 3σ threshold.
The seeing in the optical images varied in the range 1.3 < FWHM < 2.0 arcsec. One field
contained a nearby rich galaxy cluster.
Spectroscopic follow-up was obtained for a K-selected sample in sub-regions of total
area 4.4 deg2, using the Autofib-2 fiber positioner and WYFFOS spectrograph on the 4.2m
William Herschel Telescope on La Palma. Spectra were obtained for 567 galaxies with K< 15,
which allowed redshifts to be measured for 510 galaxies (a redshift completeness of 90%).
Although the spectroscopic sample is K-selected, this does not introduce any incompleteness
in the I< 16 sample used extensively in this work (i.e. there are no galaxies with I−K< 1
in the sample). We also briefly consider an I< 18 sample, for which the spectroscopic
completeness falls to around 50% because of the K-band selection (which also introduces a
bias in this fainter sample against objects that are blue in I−K). For the I< 16 and I< 18
samples the median redshift is z = 0.08 and 0.14 respectively.
-- 4 --
3. Method
Wadadekar, Robbason & Kembhavi (1999) have proposed a two-dimensional galaxy
decomposition technique that can efficiently recover B/T ratios (and other parameters) of
model galaxy images with high accuracy (see also Byun & Freeman 1995; de Jong 1996).
They present a detailed study of the effects of uncertainties in the point spread function
(PSF), the presence of nearby stars and the stability of the B/T estimates as a function
of signal-to-noise. Our approach is similar to theirs, but we apply the technique to a large
photometric sample of real galaxies1. We consider the "real-world" problems of automated
masking of nearby galaxies and stars and make a thorough assessment of the errors in
the measured parameters. We also present, in an Appendix, estimators for the luminosity
functions of disks and spheroids.
For this analysis, we have used the sample of 636 I < 18 galaxies of Gardner et al. (1997)
imaged in B, V, I and K, as described in §3. Postage stamp images of 33×33 pixels (0.68′′/pix)
around each galaxy were extracted from the I-band data. This was the best observed band
by Gardner et al. (1997) and is particularly well-suited for our purposes because it minimizes
the effects of young blue stars. This image size is large enough to include the entire region of
the galaxy for which reasonable signal-to-noise is achieved (and in most cases extends well
beyond it.) To determine the bulge-to-total ratio, B/T, we fit the two-dimensional surface
brightness profile of each galaxy using a combination of an exponential disk,
and an r1/4-law spheroid,
Σd(θ) = Σd,0 exp(−θ/θd),
Σs(θ) = Σs,e exp(−7.67[(θ/θe)1/4 − 1]),
(1)
(2)
where θ is the angular distance from the galaxy center. We will also consider a more general
r1/n spheroid profile, as Wadadekar, Robbason & Kembhavi (1999) did. The disk is allowed
to be inclined, and to have arbitrary position angle. The spheroid is allowed an ellipticity
(defined as the ratio of semi-major to semi-minor axes) in the range 1 to 6, and can also
have an arbitrary position angle. To mimic seeing, we construct mock images using these
profiles which we then smooth with a Gaussian filter (integrated over each pixel to account
for the variation of the PSF across the pixel),
p(θ) = exp[−(θ/σ)2/2]/2πσ2.
(3)
1Wadadekar, Robbason & Kembhavi (1999) applied their technique to three galaxies for which previous
estimates of B/T were available and found reasonable agreement.
-- 5 --
The width of the Gaussian is treated as a free parameter, to account for variations in seeing
between the images. (We will examine in §4.1 the effect of using a more realistic PSF.) The
postage stamp images were centered on the galaxy of interest, but we allow the position of
the image center to vary since in many cases the resulting sub-pixel variations lead to lower
values of χ2. We also allow a small contribution from a faint, constant surface brightness
background in order to take into account small inaccuracies in sky subtraction. The best
fitting parameters for each galaxy were then obtained by minimizing χ2 using Powell's algo-
rithm (Brent 1973). There are a total of 12 fit parameters (13 if we include n when fitting
r1/n spheroid profiles) summarized in Table 1. In §4.1 we will consider how accurately this
procedure recovers the B/T ratio of the galaxies.
A significant fraction of the postage stamp images was contaminated by a secondary
galaxy (and occasionally by more than one). We use a simple algorithm to identify such
contaminants and mask them from the image. The aim is to remove objects which are
physically distinct from the galaxy of interest without masking any pixels of the galaxy
itself. We first rank the pixels in the image by surface brightness and then proceed to
find groups of bright pixels. The brightest pixel is assigned to the first group. Successive
pixels are assigned to a pre-existing group if they touch it (i.e.
if they are adjacent either
horizontally, vertically or diagonally), or else are assigned to a new group. In the case where
a pixel touches more than one group, the two groups are merged. This process is continued
until pixels of a fixed signal-to-noise are reached (specifically, we consider only pixels more
than 3σ above the sky background). If more than one group exists at this point, the group
at the center is deemed to be the galaxy of interest and the pixels of all secondary groups are
marked as being contaminated and are not included in the χ2 sum. This simple algorithm
works well in the majority of cases, but it fails in a few (28 out of 636 galaxies), either by
not removing a contaminating galaxy, or by removing a significant fraction of the primary
galaxy. Rather than attempting to use a more complex algorithm in these cases, we resorted
to cleaning the image by hand (i.e. we view the image and manually mark the contaminated
pixels).
-- 6 --
Table 1: The parameters used to construct mock galaxy images in the fitting procedure.
Each parameter is described in the text.
Σd,0 Disk central surface brightness
Σs,e
Spheroid surface brightness at the ef-
fective radius
Disk angular scalelength
Spheroid effective radius
Disk position angle
Spheroid position angle
Disk inclination angle
Spheroid ellipticity
θd
θe
Pd
Ps
i
e
(x, y) Center of image
B
σ
n
Excess background surface brightness
Seeing (c.f. eqn. 3)
Spheroid profile index (fixed at n = 4
unless otherwise stated)
-- 7 --
Fig. 1. -- (cont.) Postage stamp images of three representative galaxies from our sample.
The top row shows a galaxy that is fit very well by our procedure, the bottom row shows
one that is poorly fit and the middle row shows a more typical result. The left hand column
is the original 33 × 33 pixel galaxy image with contours indicating the pixel value in ADUs.
The right hand column is the residual image after subtracting the best fit model galaxy.
Contours show the absolute value of the residual in units of σ, the rms uncertainty on
each pixel value. Hatched regions contained contaminant galaxies and were removed by our
automated cleaning procedure before fitting. (Where an entire row or column is hatched,
the postage stamp image was recentered prior to fitting.) Between the original image and
the residual maps we quote the value of χ2 per degree of freedom, Q (the probability that
a random fluctuation exceeds this value of χ2) and the I-band apparent magnitude. Also
shown is a histogram of dP/d(B/T), the distribution of bulge-to-total ratios found from the
Monte Carlo simulations described in the text, with a vertical dashed line indicating the
best-fit B/T value for the original image.
-- 8 --
The majority of galaxies (363 out of 626) are fit reasonably well by this procedure.
We regard a galaxy as being reasonably well fit when Q, the probability that the measured
value of χ2 is exceeded by random fluctuations, is greater than 5%. Not surprisingly though,
many galaxies are not well fit. These typically show signs of strong morphological disturbance
(perhaps due to a recent or imminent merger) or other inhomogeneities. Examples of well-fit
and poorly-fit galaxies are given in Figure 1. Poorly-fit galaxies are easily identified by their
large χ2 values and so may be excluded from further analysis if desired. It should be noted
that a poor-fit does not indicate a failure of our fitting procedure per se, rather it signals
that the galaxy is not well described by a combination of a spheroid and a disk. We choose
to show results computed using the entire sample, regardless of how well a galaxy was fit,
but we will comment on how our results change if badly fit galaxies are excluded from the
analysis.
Errors on the fitted parameters could, in principle, be determined using a ∆χ2 approach,
but this would require mapping χ2 in the 12 dimensional parameter space of the fit --
an exceedingly time consuming exercise -- and, in any case, the errors are unlikely to be
normally distributed given that the model is highly nonlinear in the parameters. We therefore
adopt a Monte Carlo approach to error estimation. Using the best fitting model for each
galaxy, we generate 30 realizations of that model, add random noise at the same level as in
the real image, and mask out any pixels which were masked out in the original. We then
find the best fitting parameters for each realization and take their distribution as indicative
of the uncertainties in the actual fit. It should be noted that this is only a valid procedure
if the original image is well fit by the model. In cases where this is not the case, there is no
reason to expect the Monte Carlo distributions to give an estimate of the true errors. The
B/T distributions for the three galaxies illustrated in Figure 1 are shown in that figure.
4. Results
4.1. Accuracy Checks
We begin by assessing the reliability of our procedure for recovering the true B/T ratio of
a galaxy (assuming, of course, that real galaxies are well described by our model). Our Monte
Carlo procedure for error estimation allows a determination of the accuracy of our technique.
For each galaxy, the value of B/T input into the Monte Carlo simulations may be compared
to the mean and standard deviation of the distribution of 30 recovered B/T values. Figure 2
gives the results of these accuracy tests. The left-hand panel shows the standard deviation of
the recovered B/T ratio, as a function of the I-band apparent magnitude of the mock image.
For bright galaxies (mI <∼16), σMC is fairly small, typically less than about 0.1. However,
-- 9 --
for fainter galaxies, σMC increases very rapidly, resulting in rather poorly constrained B/T
values. (In reality, σMC depends also upon the other parameters that describe the mock
image, but the correlation with apparent magnitude is the most important.)
In the right-hand panel of Figure 2, we plot the mean value of B/T recovered from the
Monte Carlo simulations against the true value for the mock image. The large solid circles
indicate those images for which σMC ≤ 0.1. Evidently, for these galaxies the value of B/T
is recovered accurately and without any strong systematic bias. The small dots show the
results for all other galaxies. Now the scatter is much larger and, more importantly, there
are systematic biases in the mean recovered B/T, such that very low and very high values
are avoided. This effect is not surprising: the values of B/T for these faint galaxies are
almost entirely unconstrained. (Note that the standard deviation for a completely uniform
distribution of B/T is approximately 0.3.) As a result, the distribution of B/T from the
Monte Carlo simulations becomes close to uniform, with the mean tending towards 0.5 as
σMC increases. For the 90 galaxies in our sample with mI ≤ 16, there is a tight correlation
between (B/T )true and the mean value recovered from the Monte Carlo simulations and it is
this subsample that we will use below to compute luminosity functions. Unfortunately, its
relatively small size limits the statistical accuracy of our estimates quite considerably.
-- 10 --
Fig. 2. -- Left-hand panel: the standard deviation, σMC, of the distribution of recovered
B/T ratios from 30 Monte Carlo realizations of a model galaxy image, as a function of the
apparent I-band magnitude of the model image. Right-hand panel: the mean recovered B/T
ratio from 30 Monte Carlo realizations of a model galaxy image plotted against the true B/T
value. Large, filled circles show those galaxies for which σMC ≤ 0.1, while small open circles
show all other galaxies.
-- 11 --
To find the best fit solution we must choose initial values for the parameters to be fit and
then use Powell's algorithm to search for values producing a better fit. For parameters such
as the position angles, disk inclination and spheroid/disk sizes we make initial guesses based
on the image being fitted. Other parameters are initially assigned "typical" values. We have
checked the effect of altering these initial values. For galaxies with σMC ≤ 0.1, the choice of
initial value makes almost no difference to the recovered values of the parameters, indicating
that our technique is finding the true minimum χ2. As σMC becomes larger, however, the
recovered parameters begin to depend strongly upon the initial values chosen. For these
images, the χ2 surface in the 12 dimensional parameter space does not possess an obvious
minimum (i.e.
it is very noisy). This is just another way of saying that the values of the
fitted parameters for these faint images are highly uncertain.
Andredakis, Peletier & Balcells (1995) have demonstrated that the bulges of spiral
galaxies are more accurately fit by an r1/n, rather than by the more usual r1/4, surface
brightness profile, with values of n ranging from around 1 to 6 (similar variations in n are
seen for elliptical galaxies; Bingelli & Cameron 1991; Caon, Capaccioli & D'Onofrio 1993).
They show that the value of n is strongly correlated with morphological type. Although
our present data provide only rather poor constraints on the value of n (due to limited
angular resolution and signal-to-noise), we have nevertheless repeated the fitting procedure
using r1/n profiles for the spheroids, treating n as a free parameter. For galaxies where the
B/T ratio is well determined, we find that there is a strong correlation between the B/T
values obtained with r1/4 and r1/n profiles, although inevitably some scatter is present. The
disk and spheroid luminosity functions computed using B/T ratios from r1/n fits show no
statistically significant difference from those using r1/4 fits.
Finally, we remind the reader that our analysis makes use of a Gaussian PSF to mimic
the effects of seeing in the data. A Gaussian accurately describes the core of the PSF
measured from bright stars in the images. However, a profile consisting of a Gaussian
core plus power-law wings provides a better match to many of the stellar profiles.
(The
variation of Gaussian and core components from night to night in the imaging data is not
so well characterized, however, and this is why we make use of a simple Gaussian for our
main analysis.) Fitting the images using such a profile (keeping the relative proportions of
Gaussian and power-law wings fixed, but allowing the overall radial scale of the PSF to be a
free parameter) results in small changes in the B/T ratio, typically significantly smaller than
the error in the best fit value. Thus, the luminosity functions presented below are unaffected
by the exact choice of PSF. However, it is clear that a good characterization of the PSF and
its variation will be crucial to obtain accurate disk and spheroid luminosity functions from
larger, higher quality datasets.
-- 12 --
4.2. Luminosity Functions
Using the I< 16 sample of approximately 90 galaxies for which we have good estimates
of the B/T ratio we now proceed to estimate the disk and spheroid luminosity functions.
Our aim here is to develop the techniques required for this measurement and demonstrate
them using a particular dataset. Given the small size of the dataset we must expect that
both statistical (due to the small number of galaxies) and systematic (due, for example, to
the lack of rich clusters in the dataset) errors will be present. These issues are considered
further in §5.
To determine the present-day luminosity functions, we need to apply k+e corrections to
the galaxy luminosities. We use the type-dependent k+e corrections obtained by Gardner
et al. (1997). Briefly, a set of model galaxy colors was computed using an updated version
of the Bruzual & Charlot (1993) stellar population models with a range of star formation
histories. The observed colors of each galaxy were matched to one of the models and that
particular model was then used to extrapolate the observed galaxy luminosity to z = 0. Note
that our type-dependent k+e corrections are based on the total (i.e. disk plus spheroid) color
of each galaxy. In principle, k+e corrections could be applied to each component separately
if spheroid/disk decompositions were carried out in several bands. Given the uncertainties
in our present estimates of B/T, we refrain from this degree of complexity in this analysis.
We use the stepwise maximum likelihood (SWML) estimator proposed by (Efstathiou,
Ellis & Peterson 1988, hereafter EEP) and also the parametric maximum likelihood method
proposed by (Sandage, Tammann & Yahil 1979, hereafter STY) to compute disk and spheroid
luminosity functions. The detectability of a spheroid depends on both its apparent magnitude
and the B/T ratio, and we must account for this in constructing the likelihood function.
This leads us to define a two-dimensional function, Φ(M, B), such that Φ(M, B)dMdB is
the number of galaxies per unit volume with B/T ratio in the range B to B + dB and
spheroid absolute magnitude in the range M to M + dM (with an equivalent definition for
disks). The application of the maximum likelihood estimator to this function is discussed
in detail in Appendix A. The normal luminosity function of spheroids is readily derived
0 Φ(M, B)dB (and similarly for disks). For the STY method we must
assume some parametric form for the luminosity function. We have tried fitting the disk and
spheroid luminosity functions with a "Schechter⊗exponential" form, namely Φ(M, B) =
φ(M) exp(βB), where φ(M) is the normal Schechter function and β is a parameter to be fit,
motivated by the shape of the SWML estimate of these luminosity functions.
using φ(M) = R 1
-- 13 --
Fig. 3. -- I-band luminosity functions. Triangles, circles and stars show the SWML estimates
of the total, spheroid and disk luminosity functions respectively. Only galaxies brighter than
I = 16 have been used, k+e corrections have been applied to all galaxies, and distances
have been calculated assuming (Ω0, Λ0) = (0.3, 0.7). Errorbars are the sum in quadrature
of the standard SWML errors and the variance in estimates of the luminosity function from
30 Monte Carlo realizations of the spheroid/disk decomposition procedure. The very heavy
solid line shows the best fit Schechter function to the total luminosity function, while the
inset shows the values of α and M⋆ for this fit, together with their 1 and 2σ error ellipses.
The heavy and thin solid lines show the best-fit "Schechter⊗exponential" functions to the
spheroid and disk luminosity functions respectively, and confidence regions for α and M⋆ for
these fits are given in the inset (the remaining parameter of the fits was β = 0.0 ± 0.37 and
2.1 ± 0.37 for spheroids and disks respectively).
-- 14 --
Figure 3 shows the resulting I-band luminosity functions with distances computed as-
suming (Ω0, Λ0) = (0.3, 0.7)2 and H0 = 100h km/s/Mpc. Triangles show the total luminosity
function; circles and stars show the spheroid and disk luminosity functions separately. These
SWML luminosity functions are normalized to the I-band number counts in the Sloan Dig-
ital Sky Survey using the procedure described in the Appendix. For the total luminosity
function, we plot the standard SWML errorbars (obtained from the covariance matrix of
the luminosity function as described by EEP), but for the spheroid and disk luminosity
functions, the errorbars are the sum in quadrature of the standard SWML errors and the
variance in the luminosity function estimated from the 30 Monte Carlo realizations of the
spheroid/disk decomposition process. The errors from each source are of comparable mag-
nitude (although the variance from the Monte Carlo realizations is the smaller of the two).
The very heavy solid line shows the best-fitting Schechter function to the total luminosity
function (determined using the STY method); the inset shows the values of α and M⋆ for this
fit, together with their 1 and 2σ error contours. (The small sample size is reflected in rather
large and correlated uncertainties in M⋆ and α.) Heavy and thin solid lines show the best fit
STY "Schechter⊗exponential" luminosity function fits to the spheroid and disk luminosity
functions respectively (with the corresponding confidence ellipses for α and M⋆ shown in the
inset, and the values of β given in the figure caption). A likelihood ratio test (Efstathiou,
Ellis & Peterson 1988) shows that the Schechter⊗exponential luminosity function is not a
particularly good fit to the data. With the present small dataset we have been unable to
find a better functional form. This situation will be rectified with a larger dataset (assuming
that some suitable functional form does actually exist).
The I-band luminosity functions of disks and spheroids are remarkably similar. The
only significant difference is that the spheroid luminosity function is somewhat lower at faint
magnitudes. However, given the small size of the present sample, this difference may not be
robust. The luminosity densities in disks and spheroids obtained by integrating the SWML
luminosity functions over the range of absolute magnitudes shown in Fig. 3 are 5.8 ± 0.8
and 4.7 ± 0.7 × 107hL⊙/Mpc3 respectively (where we have taken M⊙ = 4.14 in the I-band;
Cox 2000). In principle, we can use our Schechter⊗exponential fits to estimate the total
luminosity density, extrapolating to include the contribution from arbitrarily faint spheroids
and disks. Doing so yields results which agree with the SWML estimates within the quoted
errors, suggesting that our determination may have suitably converged. However, it must
be kept in mind that the Schechter⊗exponential form is not a particularly good fit to the
current datasets.
2Assuming (Ω0 , Λ0) = (1, 0) instead changes our results only slightly, shifting the luminosity function
faintwards due to the smaller luminosity distance in this model.
-- 15 --
Fig. 4. -- Slices through the SWML estimate of the bi-variate luminosity function, Φ(M, B),
for different absolute magnitudes, M, as indicated in the figure. Points were computed in bins
of size (∆M, ∆B) = (0.48, 0.33) and errors obtained as described by EEP. The upper panel
shows the spheroid luminosity function, and the lower panel the disk luminosity function.
The points without errorbars above the dotted line in each panel indicate the mean B/T
(and D/T) for spheroids (and disks) in the corresponding absolute magnitude bins. Solid
lines indicate the best-fit "Schechter⊗exponential" parametric luminosity function for the
MI − 5 log h = −19.82 bin.
-- 16 --
In Fig. 4, we show slices through the bi-variate luminosity function, Φ(M, B), at constant
M for several values of M (as indicated in the figure, and in bins of width ∆M = 0.48). It
is evident that Φ(M, B) is not independent of B and that, in fact, it may not be separable
into a simpler form Φ(M, B) = φ(M)g(B). This is particularly noticeable for the spheroid
luminosity function. Figure 4 shows that spheroid-dominated systems (i.e. B/T> 2/3) are
found only in the brightest spheroids, while disk-dominated systems (i.e. D/T> 2/3) have
disks with a much broader range of luminosities. This point is made more clearly in Fig. 5
where we show the spheroid luminosity function of spheroid-dominated systems and the
disk luminosity function of disk-dominated systems. We find spheroid-dominated systems in
abundance only brightwards of MI − 5 log h ≈ −21, but disk-dominated systems across the
whole range of luminosities.
-- 17 --
Fig. 5. -- SWML estimates of the I-band luminosity functions of spheroids in spheroid-
dominated galaxies (i.e. B> 2/3; circles) and disks in disk-dominated galaxies (i.e. D/T>
2/3; stars). The samples include 13 and 36 galaxies respectively. Errors are the sum in
quadrature of the standard SWML errors and the variance found in luminosity functions
estimated from 30 Monte Carlo realizations of the spheroid/disk decomposition procedure.
-- 18 --
4.3. The stellar mass in disks and spheroids
The stellar mass associated with the luminosity of each galaxy is easily obtained by
a similar procedure to that employed to calculate the k-corrections (namely fitting their
BVIK colors to a set of template galaxies). We use these estimates to construct the stellar
mass functions of spheroids and disks in the local Universe and thereby estimate the total
mass content in each component. The method that we adopt is the same that Cole et al.
(2001) used to measure the total stellar mass density in the Universe from a combination
of 2dFGRS and 2MASS data. Cole et al. (2001) found the stellar mass density3 in units
of the critical density to be Ωstars = 0.0016 ± 0.00024h−1 or Ωstars = 0.0029 ± 0.00043h−1,
depending on whether a Kennicutt (1983) or a Salpeter (1955) stellar initial mass function
(IMF) was assumed.
(Both estimates include the effects of dust on galaxy luminosities
as described by Cole et al. 2001.) We normalize our stellar mass functions by requiring
them to produce the same number density of galaxies more massive than M∗ as the Cole
et al. (2001) stellar mass function. The total stellar mass density inferred from our I < 18
sample is then Ωstars = 0.0009 ± 0.00007h−1 or Ωstars = 0.0017 ± 0.00012h−1 for the same
two IMFs respectively and the same prescription for dust-extinction. Errors on the stellar
mass density were found by summing in quadrature the error from each individual bin in the
SWML mass function, together with the error in the overall normalization. Our estimates
include contributions from galaxies with stellar masses greater than 109h−2M⊙ below which
the SWML stellar mass function is not well determined. We can check our result using the
STY stellar mass function. For this, we fit a Schechter function convolved with a Gaussian
of width 0.1 in log10 Mstars to account for the scatter in the relation between stellar mass and
I-band absolute magnitude. We find that our estimates of Ωstars using the SWML and STY
mass functions agree within the errors, suggesting that the result has already converged to
sufficient accuracy. Our estimates, however, are lower than those of Cole et al. (2001), a
reflection of the flatter faint end slope of our mass functions which may well be due to the
small size of our sample.
For our I < 16 sample, for which the bulge-to-total ratio is well measured, we find a
slightly higher total stellar mass density of Ωstars = 0.0012 ± 0.00014h−1 for the Kennicutt
(1983) IMF. Splitting into spheroidal and disk components, we find Ωstars,spheroids = 0.00039±
0.00006h−1 and Ωstars,disks = 0.00051 ± 0.00008h−1 for this same IMF. (Note that with the
SWML method the stellar mass densities of disks and spheroids are not guaranteed to sum
to give the total stellar mass density.) If, instead, we assume a Salpeter IMF, the ratio of
3Specifically, Cole et al. (2001) estimated the mass locked up in stars and stellar remnants which differs
from the time integral of the star formation rate due to recycling of material by massive stars. We adopt
the same definition of stellar mass here.
-- 19 --
disk to spheroid stellar mass densities increases slightly from 1.31 to 1.37, but this change is
negligible given the current errors in these quantities. Although the small size of our sample is
clearly a significant limitation, this initial result suggests that spheroids and disks contribute
about equally to the stellar mass density of the Universe. The techniques developed in this
paper, when applied to a much larger dataset, should allow their contributions to be more
accurately determined.
5. Discussion
We have presented a detailed method to determine the bulge-to-total ratios of galaxies
by fitting to two-dimensional photometry, and have applied this technique to determine the
I-band bulge-to-total luminosity ratios of a sample of approximately 600 galaxies brighter
than I= 18 with spectroscopic redshifts. Our approach is designed to work with realistic
galaxy images, dealing automatically with contamination by nearby objects, a varying PSF
and small changes in the background from image to image. A crucial part of the fitting
procedure is a Monte Carlo determination of the errors on the fitted parameters, an approach
which is favored since it is fast and automatically accounts for the highly non-linear nature
of the model parameters. For the current sample of galaxies, around 60% are well fit by a
combination of an exponential disk and an r1/4-law spheroid. Those that are not well fit
frequently show signs of morphological disturbance. We find that bulge-to-total ratios are
determined accurately (i.e. with errors of around 10%) only for galaxies brighter than I≈ 16.
For the 90 galaxies brighter than I= 16 in this sample we measure the B/T ratio with
reasonable accuracy. We have used the resulting disk/spheroid decomposition of these bright
galaxies to construct separate luminosity functions for disks and spheroids. We find no
significant differences between them when considered purely in terms of luminosity, although
the statistical uncertainties associated with the small sample size make the detection of any
differences difficult. However, when we consider the bi-variate distributions of luminosity and
bulge-to-total or disk-to-total light, we find that spheroid dominated systems (B/T> 2/3)
only occur for the brightest spheroids, while disk-dominated systems (D/T> 2/3) occur for
a much broader range of disk luminosities.
The relative contributions of disks and spheroids to the total stellar mass density in the
Universe is a very important constraint on theories of galaxy formation which attempt to
describe the assembly of galaxies as a function of time. We find, perhaps surprisingly, that
the disks and spheroids in our sample contribute almost equally to the stellar mass density
today (in a ratio of 1.3 ± 0.2). Since the stellar populations in disks are generally younger
than those in spheroids, it is an interesting coincidence that the total stellar mass in the two
-- 20 --
kinds of structural components should be so similar at the present time.
Schechter & Dressler (1987) reached a similar conclusion to ours using a photometric
comparison technique to estimate the B-band bulge-to-disk ratios of galaxies. While this
technique may not be as accurate as our own on a galaxy-by-galaxy basis, it should provide
a good estimate of the total contribution of each component to the stellar mass density. It
is therefore reassuring that our results agree well with those of Schechter & Dressler (1987).
A different result was obtained by Fukugita, Hogan & Peebles (1998), who derived a ratio
of disk to spheroid stellar mass density of 0.33 ± 0.23. Although they found comparable B-
band luminosity density in spheroids and disks, they adopted a spheroid mass-to-light ratio
around four times greater than that for disks, resulting in spheroids making a significantly
greater contribution to the stellar mass density. While we use a more accurate technique for
converting from luminosity to stellar mass (a technique which could be improved further if
B/T ratios were measured for each galaxy in several bands), the small size of our sample
limits the accuracy of our results. In particular, our sample may not contain enough rich
clusters which are known to contain higher fractions of spheroid dominated galaxies than
the field (e.g. Dressler 1980) and this could introduce a small bias in our results.
Clearly the greatest limitation of this work is the small size of the sample of galaxies for
which accurate disk/spheroid decompositions can be performed. Fortunately, this problem
should be remedied in the near future with the advent of high quality, large area photometric
surveys such as that being carried out by the SDSS project.
Acknowledgments
We thank Jon Gardner and Carlton Baugh for supplying data used in this work and for
valuable discussions, and the referee, Alan Dressler, for valuable suggestions. We also thank
Istvan Szapudi for his assistance in the early stages of this work.
REFERENCES
Abraham R. G., van den Bergh S., Glazebrook K., Ellis R. S., Santiago B. X., Surma P.,
Griffiths R. E., 1996, ApJS, 107, 1
Andredakis Y. C., Peletier R. F., Balells M., 1995, MNRAS, 275, 874
Arnaud M., Rothenflug R., 1985, A&AS, 60, 425
Barnes J. E., Hernquist L., 1992, ARA&A, 30, 705
-- 21 --
Baugh C. M., Cole S., Frenk C. S., 1996a, MNRAS, 282, L27
Baugh C. M., Cole S., Frenk C. S., 1996b, MNRAS, 283, 1361
Bertin E., Arnouts S., 1996, A&AS, 117, 393
Bingelli B., Cameron L. M., 1991, A&A, 252, 27
Brent R. P., 1973, Algorithms for minimization without derivatives (Englewood Cliffs, New
Jersey: Prentice Hall), Chapter 7
Bruzual & Charlot, 1993, ApJ, 405, 538
Byun Y. I., Freeman K. C., 1995, ApJ, 448, 563
Caon N., Capaccioli M., D'Onofrio M., 1993, MNRAS, 265, 1013
Cole S. et al. (The 2dFGRS Team), 2001, astro-ph/0012429 (submitted to MNRAS)
Cox A. N., 2000, Allen's astrophysical quantities (4th edition; New York, Springer)
de Jong R. S., 1996, A&AS, 118, 557
de Vaucouleurs G., 1959, in Flugge S., ed., Handbuch der Physik 53, Springer-Verlag, Berlin,
p. 275
Dressler A., ApJ, 236, 351
Efstathiou G., Ellis R. S., Peterson B. A., 1988, MNRAS, 232, 431 (EEP)
Eggen O. J., Lynden-Bell D., Sandage A. R., 1962, ApJ, 136, 748
Fukugita M., Hogan C. J., Peebles P. J. E., 1998, ApJ, 503, 518
Gardner J. P., Sharples R. M., Frenk C. S., Carrasco B. E., 1997, ApJ, L99
Hoyle F., 1949, in Burgers J. M., van de Hulst H. C., eds., Problems of Cosmical Aerody-
namics, Central Air Documents Office, Dayton, p. 195
Hubble E., 1926, ApJ, 64, 321
Jimenez R., Friaca A. C. S., Dunlop J. S., Terlevich R. J., Peacock J. A., Nolan L. A., 1999,
MNRAS, 305, 16
Kauffmann G., White S. D. M., Guiderdoni B., 1993, MNRAS, 264, 201
-- 22 --
Kauffmann G., 1995, MNRAS, 274, 161
Kauffmann G., 1996, MNRAS, 281, 487
Kennicutt R. C., 1983, ApJ, 272, 54
Loveday J., Peterson B. A., Efstathiou G., Maddox S. J., 1992, ApJ, 390, 338
Menanteau F., Ellis R. S., Abraham R. G., Barger A. J., Cowie L. L., 1999, MNRAS, 309,
208
Naim A., Lahav O., Buta R. J., Corwin H. G., de Vaucoulers G., Dressler A., Huchra J. P.,
van den Bergh S., Raychaudhury S., Sodr´e L., Storrie-Lombardi M. C., 1995, MNRAS,
274, 1107
Navarro J. F., Frenk C. S., White S. D. M., 1995, MNRAS, 275, 56
Navarro J. F., Steinmetz M., 1997, ApJ, 478, 13
Navarro J. F., Steinmetz M., 2000, ApJ, 538, 477
Peebles P. J. E., 1969, ApJ, 155, 393
Sandage A., Tammann G. A., Yahil A., 1979, ApJ, 232, 352
Salpeter E. E., 1955, ApJ, 121, 61
Schechter P., 1976, ApJ, 203, 557
Schechter P. L., Dressler A., 1987, AJ, 94, 563
Somerville R. S., Primack J. R., Faber S. M., 2001, MNRAS, 320, 504
Toomre A., 1977, in Tinsley B. M., Larson R. B., eds., The Evolution of Galaxies and Stellar
Populations. Yale Univ. Press, New Haven, p. 401
van den Bosch F. C., Burkert A., Swaters R. A., 2001, astro-ph/0105158
Wadadekar Y., Robbason B., Kembhavi A., 1999, AJ, 117, 1219
This preprint was prepared with the AAS LATEX macros v5.0.
-- 23 --
Appendix A. Estimators for Spheroid and Disk Luminosity Functions
The traditional 1/Vmax estimator is trivially adapted to the case of disk and spheroid
luminosity functions. The estimator is applied just as in the case of the standard luminosity
function, except that the total luminosity of the galaxy (i.e. disk plus spheroid luminosity)
is used to compute Vmax, since it is this total luminosity that determines the volume within
which the galaxy could have been detected.
The maximum likelihood estimator of Efstathiou, Ellis & Peterson (1988, hereafter EEP)
is also easily generalized to the case of spheroid and disk luminosity functions. Consider the
case of the spheroid luminosity function (the same arguments apply to disks). As noted
earlier, the detectability of a spheroid depends upon both its absolute magnitude, M, and
on the bulge-to-total ratio which we denote by B in this Appendix. We begin therefore by
defining a two-dimensional function, Φ(M, B), such that Φ(M, B)dMdB is the number of
galaxies with bulge-to-total ratio in the range B to B + dB and spheroid absolute magnitude
M to M + dM per unit volume. The normal luminosity function of spheroids is easily recov-
0 Φ(M, B)dB. The probability that galaxy i with spheroid magnitude
Mi and bulge-to-total ratio Bi is seen in a magnitude limited survey is
ered using φ(M) = R 1
pi ∝ Φ(Mi, Bi),Z 1
0 Z M ′
−∞
lim(zi,B)
Φ(M, B)dMdB ,
(4)
where M ′
lim(zi, B) = Mlim(zi) − 2.5 log10 B and Mlim(zi) is the limiting absolute magnitude
of the survey at redshift zi. The use of M ′
lim is necessary since arbitrarily faint spheroids
will make it into the survey provided that they have a sufficiently low bulge-to-total ratio
(corresponding to sufficiently bright disks).
From this definition we can construct the usual likelihood function
ln L =
N
Xi=1
ln Φ(Mi, Bi) −
N
Xi=1
lim(zi,B)
ln(Z 1
0 Z M ′
−∞
Φ(M, B)dMdB) + const,
(5)
where N is the total number of galaxies. There are now two ways to proceed. In the first
we assume a simple parametric form for Φ(M, B) and maximize the likelihood with respect
to the parameters. This is analogous to fitting a Schechter function (Schechter 1976) to the
normal luminosity function (e.g. Sandage, Tammann & Yahil 1979). A simple parametric
form which we have tried to fit our data is
Φ(M, B) = φ(M) exp(βB),
(6)
where φ(M) is the usual Schechter function and β is a parameter to be estimated from the
fit. With this method, the likelihood function of eqn. (5) can be evaluated for each value
-- 24 --
of the three parameters α, β and M⋆, and hence the parameter values which maximize the
likelihood are readily obtained.
The second approach involves splitting Φ(M, B) into bins in M and B and treating each
as a parameter. This is equivalent to the SWML method of EEP for estimating the standard
luminosity function.
We represent Φ(M, B) as follows:
Φ(M, B) = Φk,h,(cid:26) Mk − ∆M/2 < M < Mk + ∆M/2, k = 1, . . . , Np
Bh − ∆B/2 < B < Bh + ∆B/2,
h = 1, . . . , Nq.
(7)
The likelihood function may then be written as
N
Xi=1
ln L =
where
W (Mi−Mk, Bi−Bh) ln Φk,h−
Xi=1
N
ln( Nq
Xh=1
Np
Xk=1
Φk,h∆M∆BH[Mk, Bh, Mlim(zi)])+const,
(8)
Wk,h(Mi, Bi) =(cid:26) 1 if Mk − ∆M/2 < Mi < Mk + ∆M/2, and Bh − ∆B/2 < Bi < Bh + ∆B/2,
0 otherwise
(9)
and
H[Mk, Bh, Mlim(zi)] =
1
∆M∆B Z Bh+∆B/2
Bh−∆B/2 Z Mk+∆M/2
Mk−∆M/2
Q(M, B)dMdB,
(10)
where Q(M, B) = 0 if M > Mlim(zi)−2.5 log10 B and Q(M, B) = 1 otherwise. Since only the
shape of the luminosity function is constrained by the above likelihood function, we introduce
an additional constraint, g =PkPh Φk,h(Lk,h/Lf)β∆M∆B − 1 = 0, where Lk,h is the total
luminosity of a galaxy with spheroid magnitude Mk and bulge-to-total ratio Bh and Lf is a
fiducial luminosity (which we will take to be that corresponding to MI − 5 log h = −20.5),
using a Lagrangian multiplier λ as did EEP. Maximizing ln L′ = ln L + λg then yields
Φk,h =
i=1 H[Mk, Bh, Mlim(zi)]/PNp
PN
l=1PNq
PN
i=1 W (Mi − Mk, Bi − Bk)
m=1 Φl,mH[Ml, Bm, Mlim(zi)]
(11)
which are easily solved with an iterative procedure. The covariance matrix for the parameters
is obtained in a manner entirely analogous to that outlined by EEP.
Normalization of the maximum likelihood luminosity function can be achieved using the
actual redshift data as described by Loveday et al. (1992), but using M ′
lim in the selection
function to account for the effects of the bulge-to-total ratio. A better approach is to nor-
malize by performing a least squares fit to the number counts of galaxies from a wide area
-- 25 --
survey. The cumulative number count to apparent magnitude m is given by
0 Z 1
n(m) =Z ∞
0 Z M ′
−∞
lim
Φ(m − D(z) − K(z) − 2.5 log10 B, B)
dV
dz
dMdBdz,
(12)
where D(z) and K(z) are the distance modulus and k+e correction respectively at redshift
z, which we can compute from the SWML estimate of Φ.
|
astro-ph/0412210 | 1 | 0412 | 2004-12-09T13:46:30 | Generation of Gravitational Radiation in Dusty Plasmas and Supernovae | [
"astro-ph",
"gr-qc"
] | We present a novel nonlinear mechanism for exciting a gravitational radiation pulse (or a gravitational wave) by dust magnetohydrodynamic (DMHD) waves in dusty astrophysical plasmas. We derive the relevant equations governing the dynamics of nonlinearly coupled DMHD waves and a gravitational wave (GW). The system of equations is used to investigate the generation of a GW by compressional Alfv\'{e}n waves in a type II supernova. The growth rate of our nonlinear process is estimated, and the results are discussed in the context of the gravitational radiation accompanying supernova explosions. | astro-ph | astro-ph |
Generation of Gravitational Radiation in Dusty Plasmas and Supernovae
Gert Brodin,1, ∗ Mattias Marklund,1, ∗ and Padma K. Shukla1, 2, ∗
1Department of Physics, Umea University, SE -- 901 87 Umea, Sweden
2Institut fur Theoretische Physik IV, Fakultat fur Physik und Astronomie,
Ruhr-Universitat Bochum, D -- 44780 Bochum, Germany
(Dated: February 4, 2019)
We present a novel nonlinear mechanism for exciting a gravitational radiation pulse (or a gravi-
tational wave) by dust magnetohydrodynamic (DMHD) waves in dusty astrophysical plasmas. We
derive the relevant equations governing the dynamics of nonlinearly coupled DMHD waves and a
gravitational wave (GW). The system of equations is used to investigate the generation of a GW
by compressional Alfv´en waves in a type II supernova. The growth rate of our nonlinear process is
estimated, and the results are discussed in the context of the gravitational radiation accompanying
supernova explosions.
PACS numbers: 04.30.Db, 97.60.Bw, 98.62.En
It is well known that there exist numerous mechanisms
for the conversion between gravitational waves (GWs)
and electromagnetic waves [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11,
12, 13, 14, 15, 16, 17, 18, 19]. For example, the propaga-
tion of GWs across an external magnetic field gives rise to
a linear coupling to the electromagnetic field [1] , which
may lead to the gravitational wave excitation of ordinary
electromagnetic waves in vacuum, or of magnetohydrody-
namic (MHD) waves in a plasma [2, 3, 4]. Furthermore,
various nonlinear coupling mechanisms give rise to three-
wave couplings between GWs and electromagnetic waves
in matters. We also note that four-wave processes may
cause graviton-photon conversion even in the absence of
external matters or fields [5]. Moreover, GWs can cou-
ple to other types of waves, e.g. sound waves, also in
neutral media [6]. There are numerous motives for con-
sidering wave couplings involving GWs. In some cases,
the emphasis is on the basic theory [5, 6, 7, 8, 9, 10]. In
other works, the focus is on GW detectors [11, 12, 13],
on cosmology [14, 15, 16], or on astrophysical applica-
tions such as binary mergers [17], gamma ray bursts [18]
or pulsars [19]. Many of the previous works have con-
centrated on the conversion from GWs to electromag-
netic waves, which can be analysed within a test matter
approach which neglects the back reaction on the grav-
itational field. We note that such an approach can be
justified if the background energy density is low.
In this Letter, we consider the three-wave coupling be-
tween two DMHD waves and a GW, including the ef-
fects of dust particles [20] in a dense medium such as
the supernova where electrons, protons, and charged dust
macroparticles are abundant. For this purpose, we derive
the dust Hall MHD equations [20], i.e. equations describ-
ing the dust MHD waves, including the effect of a GW.
We emphasize that for a low-beta plasma, the system
of equations has a structure which can describe both a
dust-dominated plasma, as well as an ordinary Hall-MHD
plasma (if we replace the dust mass density by the ion
mass density). Using the normal mode approach [21],
the three-wave coupling equations are derived, includ-
ing the back reaction on a GW from the Einstein equa-
tions. The system is shown to fulfil the Manley-Rowe
relations [21] (which means that the interaction process
can be viewed quantum mechanically) and to be energy
conserving. The three-wave equations are then used to
analyse the generation of a GW by the compressional
Alfv´en waves in the iron core of the type II supernova
[22, 23]. It turns out that the characteristic timescale for
the Alfv´en-GW conversion can be less than a millisecond,
which implies that the mechanism is potentially relevant
for the high-frequency part (> 1MHz) of the supernova
GW spectrum.
The plasma dynamics, due to the response to a gravi-
tational wave
ds2 = −dt2+(1 + h+) dx2+(1 − h+) dy2+2h×dx dy+dz2
(1)
can be formulated according to
∂tns + ∇ · (nsvs) = 0,
(2)
and
msns(∂t + v · ∇)v = −∇ps + qsns(E + v × B) + msnsgs,
(3)
e1 =
(1 + h+/2) ∂y −
where we have introduced the tetrad e0 = ∂t,
(1 − h+/2) ∂x − h×/2∂y, e2
h×/2∂x, e3 = ∂z, and ∇ = (e1, e2, e3). Moreover,
=
gs = − 1
− 1
− 1
2(cid:2)(v2
2 (1 − vsz)(vsx∂th+ + vsy∂th×)e1
2 (1 − vsz)(vsx∂th× − vsy∂th+)e2
sx − v2
sy)∂th+ + 2vsxvsy∂th×(cid:3) e3
(4)
represents the gravitational acceleration of the particle
species s due to the GWs. We have assumed that ∂z ≈
−∂t holds for the GWs.
The electromagnetic field is determined through the
gravity modified Maxwell's equations. Using the same
notation as above, they take the form
Again using Eqs. (10) and (11) to eliminate the electric
2
∂tE = ∇ × B −Xs
∂tB = −∇ × E − jB,
qsnsvs − jE,
(5)
(6)
with the constraints ∇ · E = Ps qsns and ∇ · B = 0.
Here, the effects of the GWs (1) are represented by the
effective currents
jE = − 1
− 1
2 [(Ex − By)∂th+ + (Ey + Bx)∂th×] e1
2 [−(Ey + Bx)∂th+ − (Ex − By)∂th×] e2, (7)
and
jB = − 1
− 1
2 [(Ey + Bx)∂th+ − (Ex − By)∂th×] e1
2 [(Ex − By)∂th+ + (Ey + Bx)∂th×] e2. (8)
With the general setting established above, we will
from now on focus on the case of a three-component
dusty plasma, for which we have the equation of state
ps = kBTsns. Thus, the plasma is composed of electrons
(e), ions (i), and dust particles (d). The mass md of
the dust particles is assumed to be much larger than the
electron and ion masses, viz. me and mi, respectively.
We will assume that the plasma is approximately quasi-
neutral, i.e. qini = ene − qdnd. Moreover, the waves un-
der consideration are supposed to propagate with phase
velocities much smaller than the speed of light c. Thus,
we may neglect the displacement current in Amp´ere's law
(5), i.e.
∇ × B =Xs
qsnsvs + jE.
(9)
Due to the constraint me, mi ≪ md the momentum con-
servation equation (3) for the inertialess electrons and
ions becomes
0 = −kBTe∇ne − ene(E + ve × B) + menege,
(10)
and
0 = −kBTi∇ni + qini(E + vi × B) + minigi,
(11)
respectively. Adding Eqs. (10) and (11), using the
quasineutrality condition, assuming that the number
densities of the electrons and ions are not much larger
than the number density of the dust, and using the heavy
dust approximation, the dust momentum equation takes
the form
ρd (∂t + vd · ∇) vd = −kB(cid:18)Td −
qd
qi
+(∇ × B) × B − jE × B + ρdgd
Ti(cid:19) ∇nd
(12)
field, Faraday's law (6) becomes
md
qd
∂tB = ∇×(vd × B)−
∇×[(∂t + vd · ∇) vd − gd]− jB,
(13)
where we have used the dust momentum equation (12).
Thus, Eqs. (12) and (13) together with the dust conti-
nuity equation
∂tρd + ∇ · (ρdvd) = 0,
(14)
constitute the dust MHD equations in the presence of a
GW. For a low-beta plasma, the pressure term in (12) is
negligible, which means that the structure of Equations
(12)-(14) is the same as in an ordinary Hall-MHD plasma
without dust. Henceforth, we will consider a low-beta
plasma, drop the index d on all quantities and thus let
q/m be either the ion charge to mass ratio or the consid-
erably smaller charge to mass ratio of the dust particles.
As a result, our mathematical analysis below will then
apply either to a dust Hall-MHD plasma, or to an ordi-
nary Hall-MHD plasma without dust.
To simplify the problem, we consider the case when
the dust-acoustic speed cs = kBT /m is much smaller
such
that the pressure term in (12) can be neglected. As a
prerequisite for the nonlinear calculations, we first study
the linear modes of the system (12)-(14) omitting the
than the dust Alfv´en velocity CA = (cid:0)B2
0/µ0ρ(cid:1)1/2
gravitational contributions. Letting B = B0bz + B1, ρ =
ρ0 + ρ1, where the index 1 denotes the perturbation of
the equilibrium part, and linearizing Eqs. (12)-(14) and
Fourier analysing, we readily obtain the dispersion rela-
tion
(cid:0)ω2 − k2
z C 2
A(cid:1) −
A(cid:1)(cid:0)ω2 − k2C 2
ω2k2
z k2C 2
A
ω2
c
= 0,
(15)
where ωc = qB0/m is the gyrofrequency. For frequencies
much smaller than the gyrofrequency, we note that the
z C 2
modes separate into the shear Alfv´en wave, ω2 − k2
A ≈
0, and the compressional Alfv´en wave, ω2 − k2C 2
A ≈ 0.
Below we will consider the more general case described
by (15), however. For later applications it is convenient
to use the linear equations to express all quantities in
terms of a single variable. Thus, we let the wavevector
of the dust MHD waves lie in the x − z-plane, and write
vy = i
ω
ωc
k2
z C 2
A
z C 2
(ω2 − k2
A)
vx, vz = 0, ρ1 = ρ0
kxvx
ω
(16)
Bx = −B0
vx
By = −iB0
kz
(ω2 − k2
z C 2
A)
where ρd = mdnd.
approximation [Te + (e/qi)Ti]ne ≪ [Td − (qd/qi)Ti]nd.
In Eq. (12) we have also used the
Bz = B0
ωkz
k2C 2
A
ω2
ωc
ωkx
k2C 2
A
vx.
vx
(17)
(18)
(19)
Next, we consider a system of three weakly inter-
acting waves. Two dust MHD waves with frequencies
and wave-numbers (ω1, k1) and (ω2, k2) respectively, and
an arbitrarily polarized gravitational wave propagating
along the z-direction with the frequency and wavenum-
ber (ωg, kgbz). Noting that the gravitational dispersion
relation reads ωg = kgc and that CA ≪ c, the frequency
and wavenumber matchings (energy and momentum con-
servation) can be approximated
ωg = ω1 + ω2, kg = k1 + k2 ⇒ 0 ≈ k1 + k2.
(20)
time, i.e. ρ = ρ0 +P2
We will thus use k1 ≈ −k2 below, and we define
k1x = −k2x ≡ kx as well as k1z = −k2z ≡ kz (let-
ting ky = 0 for convenience). All quantities are now
assumed to be superpositions of two dust MHD waves
waves whose amplitudes are weakly varying functions of
j=1 ρj(t) exp[i(kj · r − ωt)] + c.c.,
where c.c stands for the complex conjugate, In princi-
ple, the gravitational wave should also contribute, but we
note that within a fluid model the gravitational wave con-
tribution to all plasma perturbations (velocity, magnetic
field and density) are second order in the gravitational
wave amplitudes, provided that the GW propagates par-
allel to B0, as we have assumed. Thus, the only linear
perturbations due to the gravitational wave are those of
the metric as described by Eq. (1).
Next, in order to simplify the algebra, we introduce
where
3
ω2
1,2
k2
1,2C 2
A
z1,2C 2
A
ρ0
W1,2 =
2"1 +
c(cid:0)ω2
V× = i(cid:20) ω1k2
1,2k4
1,2 − k2
z C 2
ωc (ω2
ω2
ω2
+
z1,2C 2
Ak2 − ω2ω2
1 − k2
z1C 2
and
V+ = i"1 +
ω1ω2
ω2
c
k2
z C 2
(ω2
1 − k2
A(cid:0)k2
z C 2
1,2
1k2
z
ω2
z1,2C 2
k2
A(cid:1)2 1 +
A) k2 + 1 ↔ 2(cid:21) ,
A + ω1ω2(cid:1)
2 − k2
z C 2
A)
z C 2
A) (ω2
+
A!# (25)
(26)
ω1ω2k2
z
k4C 2
A # .
(27)
Next using the Einstein equations, linearized in h+, h×,
keeping only the resonantly varying part of T µν we obtain
for the × and +-polarization, respectively
iωg
dh×
dt
= κ(cid:20)ρ0vx1vy2 +
2(cid:20)ρ0 (vx1vx2 − vy1vy2) +
κ
Bx2By1
µ0
(cid:21) ,
(28)
Bx1Bx2 − By1By2
µ0
(cid:21) ,
(29)
and
iωg
dh×
dt
=
which is reduced to
vyj −
ω2
j
k2
j
kzj
kxjB0
Bxj
and
dh×
dt
= −
ρ0ωg
Wg
V×vx1vx2,
(30)
the normal mode aj defined by
zjC 2
A
zj C 2
ω2
j k2
j − k2
aj =
vxj − i
ωj
kxj
zj C 2
j k2
ω2
A
zjC 2
j − k2
j C 2
j − k2
ωc(cid:0)ω2
A(cid:1)
A(cid:1) ω2
k2
j ωj
Bz
C 2
A
B0
c − k2
+
ω2
c(cid:0)ω2
+i(cid:0)ω2
j C 2
Aω2
j
A(cid:1) kxj
kzj
ωckxjB0
By.
(21)
Returning to Eqs.
(12)-(14) and including the nonlin-
ear terms [26], we can, by keeping the part varying as
exp[i(k1 · r − ω1t)], derive
2xh+ + v∗
∂a1
∂t
ω2
1
2k2
kz
ω1
= −
2kx(cid:0)v∗
kxB0(cid:0)B ∗
kz(cid:2)(cid:0)ω2
1 − k2C 2
−
+i
2xh+ − B ∗
A(cid:1) ω2
k2ω1ωckxB0
2yh×(cid:1) − i
2yh×(cid:1)
1(cid:3)
Aω2
c − k2C 2
ω2
1k2
z C 2
2ωc (ω2
A(cid:0)v∗
2xh× − v∗
1 − k2
z C 2
A) kx
2yh+(cid:1)
(cid:0)B ∗
2xh× − B ∗
(22)
2yh+(cid:1) ,
and we obtain a similar result for ∂a2/∂t by letting 1 ↔ 2.
After some algebra, using Eqs. (16)-(19) and (21), we
find that Eq. (22) reduces to
dvx1
dt
=
ρ0ω1
W1
v∗
x2 (V+h+ + V×h×) ,
(23)
and similarly for mode 2
dvx2
dt
=
ρ0ω2
W2
v∗
x1 (V+h+ + V×h×) ,
(24)
dh+
dt
= −
ρ0ωg
Wg
V+vx1vx2,
(31)
where Wg = ω2
g/2κ. The total wave energy is Wtot =
W1 vx12 + W2 vx22 + Wg(h×2 + h+2) and it is eas-
ily verified from (23)-(24) together with (30)-(31) that
Wtot is conserved. Furthermore, the appearance of the
same coupling coefficients V×, V+ in Eqs.
(23)-(24)
as well as in (30)-(31) assures that the Manley-Rowe
relations are fulfilled, which implies that each mode
changes energy in direct proportion to its frequency, i.e.
(dW1/dt)/(dW2/dt) = ω1/ω2 etc. The system of (23)-
(24) together with (30)-(31) describing the energy conver-
sion between DMHD waves and GWs is one of the main
results of the present letter. A more elaborate calcula-
tion scheme, including effects such as inhomogeneity and
background curvature, is a project for future research.
We now apply our results to the gravitational radia-
tion arising from the iron core of a type II supernova
where the densities can be of the order 1017 kg/m3 [23].
The large neutrino outflow (which can reach powers of
1033 W/cm2 see e.g. Ref. [22]) can generate MHD waves
described by (15). From the flux conservation, we expect
the iron core to be strongly magnetized (comparable to
pulsars), and for magnetic field strengths B0 ∼ 108 T,
the gyrofrequency will be much larger than all other fre-
quencies of the problem. The dispersion relation then
separate into the shear Alfv´en waves ω2 − k2
A ≈ 0 and
the compressional Alfv´en waves ω2 − k2C 2
A ≈ 0. Assum-
ing that the pump MHD wave is a compressional mode
with a frequency 5 MHz, the matching conditions (20)
can be fulfilled for a GW with a typical frequency 3MHz
and a shear Alfv ´en wave with the frequency 2 MHz [27].
For the assumed geometry, the MHD waves couple only
to the h×-polarization (to a good approximation), and
by combining (24) and (30), we obtain
z C 2
d2h×
dt2 = −h×
ω2
ωg
V×2
W2
16πG
c2 ρ0 vx12 .
(32)
Thus, noting that the factor ω2 V×2 /ωgW2 is negative
[27] and has a magnitude of order unity [28] for the given
parameters, the growth rate is
γ ∼p16πGρ0
vx1
c
,
(33)
which for a weakly relativistic pump quiver speed,
vx1 /c ∼ 1/10, implies γ ∼ 10 kHz. Thus, we deduce
that excitation of a GW by MHD waves is a reasonably
fast process in a dense medium such as the supernova iron
core. Furthermore, the present process can give rise to
GWs of higher frequencies than many of the previously
considered excitation mechanisms, see e.g.
[24]. Thus,
our model contributes to the understanding of gravita-
tional radiation emissions accompanying supernova ex-
plosions [25].
∗ Also at: Centre for Fundamental Physics, Ruther-
ford Appleton Laboratory, Chilton Didcot, Oxfordshire,
OX11 0QX, UK
[1] L. P. Grishchuk and A. G. Polnarev, General Relativity
and Gravitation Vol. 2 ed. A. Held (Plenum Press, New
York, 1980) pp. 416-430.
[2] D. Papadopoulos et al., A & A 377, 701 (2001).
[3] J. Moortgat and J. Kuijpers, A & A 402, 905 (2003).
[4] J. Moortgat and J. Kuijpers, Phys. Rev. D, 70, 023001
(2004).
[5] A. Kallberg, G. Brodin, and M. Marklund, submitted
(2004) (gr-qc/0410005).
[6] A. M. Anile, J. K. Hunter and B. Turong, J. Math. Phys.
40, 4474 (1999)
4
[7] M. Servin and G. Brodin, Phys. Rev. D 68, 044017
(2003).
[8] M. Servin et al., Phys. Rev. D 67, 087501 (2003).
[9] J. T. Mendon¸ca, Plasma Phys. Control. Fus., 44, B225,
(2002).
[10] A. B. Balakin et al., J. Math. Phys., 44, 5120 (2003)
[11] F.Y. Li., M. X. Tang., Int. J. Mod. Phys. D, 11, 1049
(2002)
[12] G. Brodin and M. Marklund, Class. Quantum Grav. 20,
45 (2003).
[13] R. Ballantini et al., Class. Quantum Grav. 20, 3505
(2003).
[14] D. Papadopoulos, Class Quantum Grav. 19, 2939 (2002).
[15] M. Marklund et al., Phys. Rev. D 62, 101501 (2000).
[16] P. A. Hogan and E. M. O'Shea, Phys. Rev D 65, 124017
(2002).
[17] G. Brodin et al., Phys. Rev. D 63, 124003 (2001).
[18] L. Vlahos et al., Astrophys. J., 604, 297 (2004).
[19] H. J. M. Cuesta, Phys. Rev. D 65, 64009 (2002).
[20] P. K. Shukla and A. A. Mamun, Introduction to Dusty
Plasma Physics (Institute of Physics, Bristol, 2002).
[21] J. Weiland and H. Wilhelmsson, Coherent Nonlinear In-
teraction of Waves in Plasmas, (Pergamon Press, New
York, 1977).
[22] G.G. Raffelt, Stars as Laboratories for Fundamental
Physics (The University of Chicago Press, Chicago,
1996).
[23] S. E. Woosley et al., Rev. Mod. Phys. 74 1015 (2002).
[24] C. Cutler and K. Thorne, in Proceeding of the 16th Inter-
national Conference on General Relativity and Gravita-
tion, Durban, South Africa, 2001, eds. N.T. Bishop and
S.D. Maharaj (World Scientific, 2002).
[25] M. V. Sazhin et al., JETP Lett. 64, 871 (1996).
[26] When including the nonlinear GW-coupling,
it is in
principle important to separate the coordinate compo-
nents (indices x, y, z) from the tetrad components (in-
dices 1, 2, 3), since the difference is first order in h+, h×.
However, for notational convenience we let indices x, y, z
denote tetrad components 1, 2, 3 in equation (22) and
henceforth.
[27] Note that from the Manley-Rowe relations the pump
wave must have the highest frequency for the growth
rates to become real and positive, Thus using the match-
ing condition (20) we see that the shear Alfv´en mode
formally has a negative frequency. Furthermore, we note
that the dispersion relations together with the match-
ing conditions means that the pump mode cannot prop-
agate purely along the magnetic field. Specifically, the
wavevector matching implies kxCA ≈ 4.6MHZ, where
CA ≈ 300m/s for the given parameters.
[28] This can be seen by taking the limit ωc → ∞ of the
c from
2 − k2
expression, noting that(cid:0)ω2
the dispersion relation (15).
z2C 2
A(cid:1) scales as 1/ω2
|
astro-ph/0202343 | 1 | 0202 | 2002-02-19T00:19:44 | Atomic Processes in Planetary Nebulae | [
"astro-ph"
] | A hot central star illuminating the surrounding ionized H II region usually produces very rich atomic spectra resulting from basic atomic processes: photoionization, electron-ion recombination, bound-bound radiative transitions, and collisional excitation of ions. Precise diagnostics of nebular spectra depend on accurate atomic parameters for these processes. Latest developments in theoretical computations are described, especially under two international collaborations known as the Opacity Project (OP) and the Iron Project (IP), that have yielded accurate and large-scale data for photoionization cross sections, transition probabilities, and collision strengths for electron impact excitation of most astrophysically abundant ions. As an extension of the two projects, a self-consistent and unified theoretical treatment of photoionization and electron-ion recombination has been developed where both the radiative and the dielectronic recombination processes are considered in an unified manner. Results from the Ohio State atomic-astrophysics group, and from the OP and IP collaborations, are presented. A description of the electronic web-interactive database, TIPTOPBASE, with the OP and the IP data, and a compilation of recommended data for effective collision strengths, is given. | astro-ph | astro-ph |
**TITLE**
ASP Conference Series, Vol. **VOLUME**, **PUBLICATION YEAR**
**EDITORS**
Atomic Processes in Planetary Nebulae
Sultana N. Nahar
Department of Astronomy, The Ohio State University, Columbus, OH
43210, USA
Abstract. A hot central star illuminating the surrounding ionized H
II region usually produces very rich atomic spectra resulting from basic
atomic processes: photoionization, electron-ion recombination, bound-
bound radiative transitions, and collisional excitation of ions. Precise
diagnostics of nebular spectra depend on accurate atomic parameters
for these processes. Latest developments in theoretical computations
are described, especially under two international collaborations known
as the Opacity Project (OP) and the Iron Project (IP), that have yielded
accurate and large-scale data for photoionization cross sections, transi-
tion probabilities, and collision strengths for electron impact excitation of
most astrophysically abundant ions. As an extension of the two projects,
a self-consistent and unified theoretical treatment of photoionization and
electron-ion recombination has been developed where both the radiative
and the dielectronic recombination processes are considered in an uni-
fied manner. Results from the Ohio State atomic-astrophysics group,
and from the OP and IP collaborations, are presented. A description
of the electronic web-interactive database, TIPTOPBASE, with the OP
and the IP data, and a compilation of recommended data for effective
collision strengths, is given.
1.
Introduction
A planetary nebula may be thought of as an 'astrophysical laboratory' of atomic
emission-line spectra. The spectra enable determination of its temperature,
density, and abundances of elements. The precise spectral analyis require accu-
rate atomic parameters for the radiative are collisional processes in the nebular
plasma. The four basic atomic processes that dominant the nebular plasma are:
i) Radiative bound-bound transition for excitation or de-excitation:
X +Z + hν ⇀↽ X +Z ∗,
where the ion, X, is of charge, Z.
ii) Photoionization (PI) by absorption of a photon:
X +Z + hν ⇀↽ X +Z+1 + ǫ,
The inverse process is the electron-ion radiative recombination (RR).
iii) Autoionization (AI) and dielectronic recombination (DR):
1
2
Author & Co-author
e + X +Z → (X +Z−1)∗∗ → (cid:26) e + X +Z
AI
X +Z−1 + hν DR
where the intermediate doubly excited state, autoionizing state, introduces res-
onances in the atomic processes. The inverse process of DR is photoionization
via the autoionizing state.
iv) The collisional process of electron-impact excitation (EIE)
e + X +Z → e′ + X +Z ∗∗
is one of the primary processes for spectral formation in astrophysical plasmas.
Each process needs to be treated separately to obtain the relevant atomic
parameters. Sample results for each process, obtained mainly under the OP
(1995, 1996) and the IP (1993) for accurate atomic data and stellar opacities
will be presented later. Astrophysical model applications, such as for plasma
opacities, need a huge amount of atomic data as these deal with a large number
of atomic levels over wide energy ranges. Consistent sets of atomic parameters
can be obtained if the atom or the ion is described by the same wavefunction
expansion in each process. This also reduces the uncertainty in applications
involving different processes and approximations. The Close Coupling (CC) R-
matrix mehtodology employed by the OP and IP enable the computation of such
self-consistent sets of atomic parameters.
The large amount of atomic data and opacities obtained under the OP
and the IP are available electronically through the existing database, TOPbase
(Cunto et al. 1993), and through its planned extension TIPTOPBASE (C.
Mendoza and the OP/IP team).
2. Theory
All calculations for the various atomic parameters of the dominant atomic pro-
cesses are carried out using the accurate and powerful R-matrix method in the
close-coupling approximation (e.g. Burke & Robb 1975, Seaton 1987, Berring-
ton et al. 1987, Berrington et al. 1995). The total wavefunction for a (N+1)
electron system in the CC approximation is described as:
ΨE(e + ion) = A
N
Xi
χi(ion)θi +Xj
cjΦj(e + ion),
(1)
where χi is the target ion or core wavefunction in a specific state SiLiπi or level
Jiπi, θi is the wavefunction of the interacting (N+1)th electron in a channel
labeled as SiLi(Ji)πi k2
i is the incident kinetic energy. Φj is
the correlation functions of (e+ion) system that compensates the orthogonality
condition and short range correlation interations. The complex resonant struc-
tures in photoionization, recombination, and in electron impact excitation are
included through channel couplings.
i ℓi(SLπ or Jπ); k2
Relativistic effects are included through Breit-Pauli approximation in inter-
mediate coupling. The (N+1)-electron Hamiltonian in the Breit-Pauli approxi-
mation, as adopted in the Iron Project, is
APS Conf. Ser. Style
H BP
N +1 = H N R
N +1 + H mass
N +1 + H Dar
N +1 + H so
N +1,
where non-relativisitc Hamiltonian is
H N R
N +1 =
N +1
Xi=1
−∇2
i −
2Z
ri
+
N +1
Xj>i
.
2
rij
3
(2)
(3)
N +1 is the mass correction term, H Dar
H mass
N +1 is the
spin-orbit interaction term. Spin-orbit interaction splits the LS terms into fine-
structure levels labeled by Jπ where J is the total angular momentum. Solutions
of the Schrodinger equation, H BP
N +1Ψ = EΨ which becomes a set of coupled equa-
tions with the CC expansion, give the bound wavefunctions, ΨB, for negative
energies, E < 0, and continuun wavefunction, ΨF , for positive energies, E ≥ 0.
N +1 is the Darwin term, H so
The transition matrix elements for various atomic processes are:
< ΨBDΨF > for photoionization and recombination,
< ΨBDΨB ′ > for oscillator strength,
< ΨF H(e + ion)ΨF ′ > for electron impact excitation,
where D is the dipole operator.
The transition matrix element with the dipole operator can be reduced to
the generalized line strength defined, in either length or velocity form, as
*Ψf
N +1
Xj=1
SL = (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)
2
rjΨi+(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)
, SV = ω−2(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)
2
.
(4)
*Ψf
N +1
Xj=1
∂
∂rj
Ψi+(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)
where ω is the incident photon energy in Rydberg units, and Ψi and Ψf are the
wave functions representing the initial and final states, respectively.
The oscillator strength fij and the transition probability Aji for the bound-
bound transition can be obtained from S as
fij =
Eji
3gi
S, Aji(a.u.) =
1
2
α3 gi
gj
E2
jifij,
(5)
where Eji is the transition energy, α is the fine structure constant, and gi, gj
are the statistical weight factors of the initial and final states. The lifetime,
τj, of a level j can be obtained from the transition probabilities decaying to
and τ0 =
the lower levels as τj = (Pi Aji(s−1))−1, where Aji(s−1) = Aji(a.u.)
2.4191 × 10−17s.
The photoionization cross section (σP I ) is proportional to the generalized
τ0
line strength (S),
σP I =
4π
3c
1
gi
ωS.
(6)
The calculations of total electron-ion recombination employs a new unified
treatment (Nahar & Pradhan 1994,1995). The method considers the infinite
number of recombining states and incorporates the radiative and dielectronic
recombinations, RR and DR, in a unified manner. The contributions to re-
combination from states with n ≤ 10 are obtained from σP I using principle of
4
Author & Co-author
detailed balance, while the resonant contributions from states with 10 < n ≤ ∞
are obtained from an extension of the DR theory of Bell & Seaton (1985).
The recombination cross section, σRC, is related to σP I through the princi-
ple of detailed balance,
σRC = σP I
gi
gj
h2ω2
4π2m2c2v2 .
The recombination rate coefficient, αRC , is obtained as
αRC (T ) = Z ∞
0
vf (v)σRC dv,
(7)
(8)
where f (v) is the Maxwellian velocity distribution function. The total αRC is
obtained from contributions from infinite number of recombined states.
The collision strength for transition by electron impact excitation from the
initial state of the target ion SiLi to the final state SjLj is given by
Ω(SiLi − SjLj) =
1
2 XSLπXlilj
(2S + 1)(2L + 1)SSLπ(SiLili − SjLjlj)2,
(9)
where S is the scattering matrix. The effective collision strength or the Maxwellian
averaged collision strength can be obtained as
Υ(T ) = Z ∞
0
Ωij(ǫj)e
−ǫj
kT d(ǫj/kT ).
(10)
which give the excitation rate coefficient, qij(T ) = (8.63×10−6/ωiT 1/2)e−Eij /kT Υ(T )
in cm3s−1. where T is in K, Eij = Ej − Ei, Ei < Ej are in Rydbergs (1/kT =
157885/T), and j is the excited upper state.
3. Results and Discussions
Sample results for each atomic process are presented, e.g., for f -values, σP I ,
αR(T ), and Υ(T ), in the following subsections.
3.1. Bound-Bound transitions
Due to fine structure, extensive sets of data for f - and A-values can be obtained
for a large number of bound-bound transitions using the Breit-Pauli R-matrix
method. In contrast to dipole allowed LS multiplets, BPRM method includes
relativistic effects, and consideration of both the dipole allowed and intercombi-
nation transitions. The fine structure energy levels are analysed with quantum
defects to obtain the spectroscopic identification, (Ct St Lt Jt πtnℓ)SLJ π where
(Ct St Lt Jt πt) denotes the core configuration, spin, orbital, total angular mo-
menta and parity, nl is configuration of the outer or valence electron, SL are
the possible spin and orbital angular momenta, and Jπ are the total angular
momentum and parity of the (N+1)-electron system. The fobidden quadrupole
(E2) and magnetic dipole (M1) transitions are treated with atomic structure
calculations using codes such as SUPERSTUCTURE (Eissner et al. 1974).
APS Conf. Ser. Style
5
Accurate fine structure transition probabilities have been obtained for a
number of ions such as:
Li-like: C IV, N V, O VI, F VII, Ne VIII, Na IX, Mg X, Al XI, Si XII, S XIV,
Ar XVI, Ca XVIII, Ti XX, Cr XXII, Ni XXVI
Fe ions: Fe V, Fe XVII, Fe XXI, Fe XXIII, Fe XXIV, Fe XXV
Other ions: C II, C III, O IV, S II, Ar XIII
The first large-scale application of the BPRM method for a complex ion,
Fe V, resulted in 3865 fine structure energy levels, compared to 179 observed, and
about 1.46 million f - and A−values for dipole allowed and intercombination (E1)
transitions and 362 forbidden (E2,M1) transitions (Nahar et al. 2000). Table
I presents a sample with complete spectroscopic identification of the transition
array 3d4(4D) − 3d3(4F )4p(5F o) in Fe V.
Table I: Transition probabilities of Fe V. g=2J+1.
leveli
3d4(5D)
3d4(5D)
3d4(5D)
3d4(5D)
3d4(5D)
3d4(5D)
3d4(5D)
3d4(5D)
3d4(5D)
3d4(5D)
3d4(5D)
3d4(5D)
levelj
gi
gj Ei(Ry) Ej(Ry)
f
3d3(4F )4p(5F o)
3d3(4F )4p(5F o)
3d3(4F )4p(5F o)
3d3(4F )4p(5F o)
3d3(4F )4p(5F o)
3d3(4F )4p(5F o)
3d3(4F )4p(5F o)
3d3(4F )4p(5F o)
3d3(4F )4p(5F o)
3d3(4F )4p(5F o)
3d3(4F )4p(5F o)
3d3(4F )4p(5F o)
1
3
3
5
5
5
7
7
9
7
9
9
3
3
5
3
5
7
5
7
7
9
9
11
5.5132
5.5119
5.5119
5.5094
5.5094
5.5094
5.5058
5.5058
5.5015
5.5058
5.5015
5.5015
3.1644
3.1644
3.1496
3.1644
3.1496
3.1443
3.1496
3.1443
3.1443
3.1391
3.1391
3.1343
2.154E-01
3.790E-04
1.358E-03
4.617E-02
5.967E-02
1.462E-02
6.895E-03
5.889E-02
1.966E-03
3.262E-02
5.139E-02
7.548E-02
A(sec−1)
3.18E+09
1.68E+07
3.65E+07
3.40E+09
2.67E+09
4.69E+08
4.30E+08
2.64E+09
1.13E+08
1.14E+09
2.30E+09
2.78E+09
3d4(5D)
3d3(4F )4p(5F o)
25
35
5.5055
3.1451
1.068E-01
3.42E+09
3.2. Photoionization and Recombination
The photoionization cross sections, σP I, are calculated including autoionizing
resonances that can enhance the background cross sections considerably. Fig. 1
shows the photoionization cross sections of the gound states of Fe I to Fe V
(Bautista & Pradhan 1998). Extensive resonances dominate the cross sections
for these complex ions. The enhancement in the backgound is up to three orders
of magnitude for Fe I, over an order or magnitude for Fe II, and ∼ 50% for Fe
III.
For simpler systems the accuracy of theoretical cross sections can be tested
against experimental data. Very precise measurements of σP I are now being
carried out at a few places, such as the Advanced Light Source at Berkeley (e.g.
Covington et al. 2001), the merged photon-ion beam set-up at Aahus University
(Kjeldsen et al. 1999), and a synchroton based experiment at University of Paris-
Sud (Wuilleumier et al., private communication). An example of the comparison
of ground state photoionization cross sections of C II, with very good agreement
6
Author & Co-author
Fe I
Fe IV
]
b
M
[
n
o
i
t
c
e
s
s
s
o
r
c
n
o
i
t
i
i
a
z
n
o
o
o
h
p
t
0
1
g
o
L
2
1
0
−1
2
1
0
−1
1
2
1
0
−1
2
.6
.8
1
1.2
Fe II
1.4
1.6
1.8
2
2
1.5
1
.5
0
−.5
3
2
1
4
6
8
Fe V
10
12
1.5
2
Fe III
2.5
3
0
4
5
6
7
8
9
10
11
12
2.2
2.4
2.6
2.8
3
Energy [Ry]
Figure 1.
of Fe I - Fe V.
Photoionization cross sections, σP I, of the ground states
)
b
M
(
I
P
σ
30
20
10
0
24
30
)
b
M
(
I
P
σ
20
10
0
24
C II + hν -> C III + e
(b) Expt
25
26
27
28
29
30
31
(b) Theory: BPRM
25
26
27
28
29
30
31
Photon Energy (eV)
Comparison of theoretical and experimental photoion-
Figure 2.
ization cross sections, σP I, of the ground state fine structure levels
2s22p(2P o
1/2,3/2) of C II (Nahar 2002).
with experiment, is shown in Fig. 2. Theoretical calculations include relativistic
fine structure.
An example of the total unified recombination rate coefficient αR(T ) for
O III is presented in Fig. 3 (solid) (Nahar 1998). The earlier results are RR rates
(dashed, Pequignot et al. 1991), low temperature DR rates (dotted, Nussbaumer
& Storey 1983), high temperature DR rates by Badnell & Pindzolla (1990, short-
dash-long-dash) and by Shull & Steenberg (1982, dot-desh). Differences can be
noticed between the present unified values and the sum of the earlier (RR+DR)
results, especially in the low and intermediate temperature. Compared to the
earlier Shull & Steenberg results, autoionization into excited levels in the high
temperature region reduces the recombination rates significantly.
The unified treatment for the total recombination provides rates that are
valid over a wide range of temperatures for all practical purposes in contrast to
addition of RR and DR rates obtained using different approximations for dif-
APS Conf. Ser. Style
7
e + O IV -> O III
10-9
10-10
10-11
10-12
)
1
-
c
e
s
3
m
c
(
R
α
10
100
1000
10000
T (K)
105
106
107
Figure 3.
cients (αR(T ) (solid) with the previous calculations.
Comparison of the unified total recombination rate coeffi-
ferent temperature ranges. The method enables obtaining self-consistent sets of
photoionization and recombination cross sections by using the identical wave-
function expansion for both the inverse processes. Following is the list of atoms
and ions for which self-consistent sets of σP I and αR(T ) have so far been obtained
for over 45 ions (e.g. Nahar & Pradhan 1997, http : //www.astronomy.ohio − state.edu/ ∼pradhan):
Carbon: C I, C II, C III, C IV, C V, C VI
Nitrogen: N I, N II, N II, N IV, N V, N VI, N VI
Oxygen: O I, O II, O III, O IV, O V, O VI, O VII, O VII
Iron: Fe I, Fe II, Fe III, Fe IV, Fe V, Fe XIII, Fe XVII, Fe XXI, Fe XXIV, Fe
XXV, Fe XXVI
C-like: F IV, Ne V, Na VI, Mg VII, Al VIII, Si IX, S XI, Ar XIII, Ca XV
Other ions: Si I, Si II, S II, S III, Ar V, Ca VII, Ni II
The data include the state specific recombination rates for hundreds of bound
levels with n≤ 10 for each ion.
The self-consistent sets of photoionization/recombination data include new
photoionization cross sections that are generally an improvement over the OP
data since more extensive and accurate eigenfunction expansions are employed.
3.3. Electron Impact Excitation
Reviews and compilations of available theoretical data sources are: An evaluated
compilation of theoretical data sources for electron-impact excitation of atomic
ions, (A.K. Pradhan & J.W. Gallagher, Atomic Data and Nuclear Data Tables,
52, 227, 1992), and Electron Collisions with Atomic Ions (A.K. Pradhan & H.L.
Zhang, in LAND OLT-BORNSTEIN Volume Atomic Collisions, Ed. Y. Itikawa,
Springer-Verlag, in press).
A table of recommended data for effective collision strengths and A-values
for nebular ions is available on-line from www.astronomy.ohio-state.edu/∼pradhan
The aim of the IP is to compute collisional data for the iron-peak elements
in various ionization stages. Of the 50 publications in the "Atomic data from
the IRON Project" series in Astronomy and Astrophysics journal, most are on
8
Author & Co-author
Figure 4.
Collision strength for the first transition in Fe VI
Ω(4F3/2 − 4F5/2) from the Iron Project CC R-matrix calculation (Chen
& Pradhan 1999); previous results (• - Nussbaumer & Storey 1978, ×
- Garstang et al. 1978) neglect resonances.
collisional strengths. At present the Iron Project website is maintained by Keith
Butler at: www.usm.uni − muenchen.de/people/ip/iron − project.html
An example of IP calculations for iron ions is the recent work on Fe VI (Chen
& Pradhan 1999). Fig. 4 presents the detailed collision strength Ω(E) showing
extensive resonance structure. Previous calculations neglecting resonances seri-
ously underestimate the Maxwellian averaged effective collision strength Υ(T )
(Eq. 10) by several factors.
4. TIPTOPBASE: atomic radiative and collisional data
The existing electronic archive for the OP data, TOPbase (http : //heasarc.gsf c.nasa.gov
at Goddard, NASA and http : //vizier.u − strasbg.f r/OP.html at CDS) con-
tain (i) photoionization cross sections of bound LS terms for ∼ 200 ions with Z
= 1 - 14, 16, 18, 20, 26, (ii) energy levels and transition probabilities (∼ 107 f -
values), (iii) monochromatic and Rosseland mean opacities (at CDS only)
The new database, TIPTOPbase (under development, C. Mendoza & the
OP/IP team) will have radiative and collisional data from the OP and the IP.
The data include:
(i) All data from TOPbase (and replacement of improved data), (ii) col-
lisional data for iron and iron peak elements, (iii) new elements through all
ionization stages: P, Cl, K, as well as selected ions such as Ni II and Ni III, (iv)
updated sets of radiative data from new CC calculations, (v) "Tail" photoion-
ization cross sections at high energies that include inner-shell ionization, (vi)
total and level specific recombination rate coefficients, (vi) additional data for
f -values including inner-shell excitations in iron ions Fe VIII - XIII ("PLUS"
data) calculated with SUPERSTRUCTURE, (vii) radiative data (σP I and f -
values) for fine structure levels including relativistic effects, (viii) on-line com-
APS Conf. Ser. Style
9
putational facilities for opacities and radiative accelerations for user-specified
mixtures of elements ("customized" opacities and radiative forces) (see IP in
Extended Abstracts).
5. Conclusion
The current status of large-scale ab initio close coupling R-matrix calculations for
radiative and collisional processes is reported. The Iron Project Breit Pauli R-
matrix radiative calculations include large numbers of dipole allowed and inter-
combination transitions. Self-consistent sets of atomic data for photoionization
and unified (electron-ion) recombination (including RR and DR) are obtained,
and should yield more accurate photoionization models. Work is in progress for
heavy ions of the iron group elements.
Acknowledgments. This work is supported partially by the U.S. National
Science Foundation and NASA.
References
Bautista, M.A. & Pradhan, A.K. 1998, ApJ, 492, 650
Berrington, K.A., Burke, P.G., Butler, K., Seaton, M.J., Storey, P.J., Taylor,
K.T., & Yu, Yan, 1987, J.Phys. B, 20, 6379
Berrington, K.A., Eissner, W.B., & Norrington, P.H. 1995, CPC, 92, 290
Burke, P.G. & Robb, W.D. 1975, Adv. At. Mol. Phys., 11, 143
Chen, G.X. & Pradhan, A.K. 2000, ApJS, 147, 111
Covingtion et al. 2001, Phys.Rev.Lett. 87, 243002
Cunto, W., Mendoza, C., Ochsenbein, F., Zeippen, C.J, 1993, A&A, 275, L5
Eissner, W., Jone, S. W., & Nussbaumer, N. 1974, CPC, 8, 270
Hummer, D.G., Berrington, K.A., Eissner, W., Pradhan, A.K., Saraph, H.E., &
Tully, J.A. 1993, A&A, 279, 298
Kjeldsen, H., Folkmann, F., Hensen, J.E., Knudsen, H., Rasmussen, M.S., West,
J.B., & Andersen, T. 1999, ApJ. 524, L143
Nahar S.N. 1999, ApJS, 120, 131
Nahar S.N. 2002, Phys.Rev.A (in press)
Nahar S.N., Delahaye F.,Pradhan A.K., & Zieppen C.J. 2000, A&AS144, 141
Nahar, S.N. & Pradhan, A.K. 1994, Phys.Rev.A, 49, 1816; 1995, ApJ, 447, 966;
Nahar, S.N 1996, Phys.Rev.A, 53, 1545
Nahar, S.N., & Pradhan, A.K. 1997, ApJS, 111, 339
Pequignot, D., Petitjean, P., & Boisson, C. 1991, A&A, 251, 680; Nussbaumer,
H. & Storey, P.J. 1983, A&A, 126, 75; Badnell, N.R. & Pindzola, M.S.
1989, Phys.Rev.A, 39, 1690; Shull, J.M. & van Steenberg, M. 1982, ApJS,
48, 95
Seaton, M.J. 1987, J.Phys.B, 20, 6363
The Opacity Project 1 and 2, compiled by the Opacity Project team (Institute
of Physices, London, UK, 1995, 1996)
|
astro-ph/9308044 | 1 | 9308 | 1993-08-30T18:35:00 | Reconstructing the Inflaton Potential---Perturbative Reconstruction to Second Order | [
"astro-ph",
"hep-ph"
] | One method to reconstruct the scalar field potential of inflation is a perturbative approach, where the values of the potential and its derivatives are calculated as an expansion in departures from the slow-roll approximation. They can then be expressed in terms of observable quantities, such as the square of the ratio of the gravitational wave amplitude to the density perturbation amplitude, the deviation of the spectral index from the Harrison--Zel'dovich value, etc. Here, we calculate complete expressions for the second-order contributions to the coefficients of the expansion by including for the first time corrections to the standard expressions for the perturbation spectra. As well as offering an improved result, these corrections indicate the expected accuracy of the reconstruction. Typically the corrections are only a few percent. | astro-ph | astro-ph | SUSSEX -- AST 93/8-2
FNAL -- PUB -- 93/258 -- A
astro-ph/9308044
(August 1993)
3
9
9
1
g
u
A
0
3
1
v
4
4
0
8
0
3
9
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
Reconstructing the inflaton potential --
perturbative reconstruction to second-order
Edmund J. Copeland∗
School of Mathematical and Physical Sciences,
University of Sussex, Brighton BN1 9QH, U. K.
Edward W. Kolb†
NASA/Fermilab Astrophysics Center
Fermi National Accelerator Laboratory, Batavia, IL 60510, and
Department of Astronomy and Astrophysics, Enrico Fermi Institute
The University of Chicago, Chicago, IL 60637
Andrew R. Liddle‡
Astronomy Centre, School of Mathematical and Physical Sciences,
University of Sussex, Brighton BN1 9QH, U. K.
James E. Lidsey§
NASA/Fermilab Astrophysics Center
Fermi National Accelerator Laboratory, Batavia, IL 60510
One method to reconstruct the scalar field potential of inflation is a
perturbative approach, where the values of the potential and its derivatives
are calculated as an expansion in departures from the slow-roll approxima-
tion. They can then be expressed in terms of observable quantities, such
as the square of the ratio of the gravitational wave amplitude to the den-
sity perturbation amplitude, the deviation of the spectral index from the
Harrison -- Zel'dovich value, etc. Here, we calculate complete expressions
for the second-order contributions to the coefficients of the expansion by
including for the first time corrections to the standard expressions for the
perturbation spectra. As well as offering an improved result, these cor-
rections indicate the expected accuracy of the reconstruction. Typically
the corrections are only a few percent.
PACS numbers: 98.80.-k, 98.80.Cq, 12.10.Dm
∗Electronic mail: [email protected]
†Electronic mail: [email protected]
‡Electronic mail: [email protected]
§Electronic mail: [email protected]
I. INTRODUCTION
An intriguing prospect raised by recent large-scale structure observations, par-
ticularly those of the Cosmic Background Explorer (COBE) satellite [1], is that ob-
servations may soon provide rather detailed information regarding the nature of the
vacuum energy driving inflation [2,3,4]. In most models this vacuum energy is iden-
tified as the self-interaction potential of a scalar inflaton field, and its precise form is
determined by some particle physics model. Since there currently exist many possible
models [5], it is of great interest to investigate whether observations can select which,
if any, of these models is correct. As well as having detailed implications for the
initial conditions for structure formation in the Universe, the energy scale of inflation
provides a link with particle physics at high energies, and may be a useful method of
probing models of unification.
The prime observational consequences of inflation derive from the stochastic spec-
tra of density (scalar) perturbations and gravitational wave (tensor) modes generated
during inflation. Each stretches from scales of order centimeters to scales well in ex-
cess of the size of the presently observable Universe. Once within the Hubble radius,
gravitational waves redshift away and so their main influence is on the large-scale
microwave background anisotropies, such as those probed by COBE [6]. Advanced
gravitational wave detectors such as the proposed beam-in-space experiments may be
able to detect the gravitational waves on a much shorter (about 1014cm) wavelength
range [7]. The density perturbations are thought to lead to structure formation in
the Universe. They produce microwave background anisotropies across a much wider
range of angular scales than do the tensor modes, and constraints on the scalar spec-
trum are also available from the clustering of galaxies and galaxy clusters, peculiar
velocity flows and a host of other measurable quantities [4].
Recently, we provided a formalism which allows one to reconstruct the inflaton
potential V (φ) directly from a knowledge of these spectra [8,9]. This developed an
original but incomplete analysis by Hodges and Blumenthal [10]. An important result
that follows from our formalism is that knowledge of the scalar spectrum alone is in-
sufficient for a unique reconstruction. Reconstruction from only the scalar spectrum
leaves an arbitrary integration constant, and since the reconstruction is nonlinear, dif-
ferent choices of this constant lead to different functional forms for the potential. A
minimal knowledge of the tensor spectrum, say its amplitude at a single wavelength,
is sufficient to lift this degeneracy. With further information the problem becomes
overdetermined, providing powerful consistency relations which would exclude infla-
tion if not satisfied.
The most ambitious aim of reconstruction is to employ observational data to
deduce the complete functional form of the inflaton potential over the range cor-
responding to large-scale structure. The observational situation is some way from
providing the quality of data that this would require, and at present a more realistic
approach is to attempt a reconstruction of the potential about a single point φ0 [9].
1
For this one requires such information as the amplitudes of both scalar and tensor
modes and also the spectral index of the scalar perturbations at a single scale. It
is possible that such information can be deduced from a combination of microwave
background experiments that span a range of angular scales [11]. In this paper it is
our aim to provide an improved calculation of the coefficients of such a perturbative
reconstruction.
To some extent all inflationary calculations rely on the use of the slow-roll approx-
imation. In the form we present here, the slow-roll approximation is an expansion in
terms of quantities defined from derivatives of the Hubble parameter H. In general
there are an infinite hierarchy of these which can in principle all enter at the same
order in an expansion. However, to calculate V (φ0) one needs only the first deriva-
tive of H, for V ′(φ0) one needs up to the second derivative of H, and so on. These
parameters can be converted into more observationally related quantities, as we shall
see.
The slow-roll approximation arises in two separate places. The first is in simpli-
fying the classical inflationary dynamics of expansion, and the lowest-order approxi-
mation ignores the contribution of the inflaton's kinetic energy to the expansion rate.
The second is in the calculation of the perturbation spectra; the standard expres-
sions are true only to lowest-order in slow-roll. In our earlier work [8,9], we utilized
the Hamilton-Jacobi approach [12] to treat the dynamical evolution exactly, but were
forced for analytic tractability to retain the lowest-order approximation for the pertur-
bation spectra. Technically therefore, the results were accurate only to lowest-order,
though in models close to the power-law inflation limit this hybrid approach offers
substantial improvements for certain quantities.
Until recently further improvements have not been possible, but a very elegant cal-
culation of the perturbation spectra to next order in slow-roll has now been provided
by Stewart and Lyth [13]. This does not permit analytic progress in functional recon-
struction, but their results can be combined with the Hamilton-Jacobi approach to
generate the complete second-order term in perturbative reconstruction. The purpose
of this work is to calculate this correction. This serves two useful purposes. Firstly,
the results allow a more accurate reconstruction to be performed, and secondly the
relative size of lowest-order and second-order contributions provides a useful (though
not rigorous) measure of the theoretical error in reconstruction. As we shall see, even
the lowest-order results are typically accurate to within a few percent.
II. TO SECOND-ORDER IN SLOW-ROLL
We shall use the notation and philosophy of our earlier paper [9], except that the
perturbation spectra shall be given by expressions improving on Eq. (3.4) of that
work. The Hamilton-Jacobi equations arise when one uses the scalar field φ as a time
2
variable, and writes the Hubble parameter H = a/a, where a is the scale factor, as a
function of φ. The field equations are [12]
[H ′(φ)]2 −
3
2
κ2H 2(φ) = −
1
2
κ4V (φ) ,
κ2 φ = −2H ′ ,
(2.1)
(2.2)
where dots are time derivatives, primes are φ derivatives, κ2 = 8π/m2
P l and mP l is the
Planck mass. Without loss of generality we may assume φ > 0, so that H ′(φ) < 0.
Where square roots appear later this choice is used to fix the sign of the prefactor.
The slow-roll approximation can be specified by parameters defined from deriva-
tives of H(φ). There are in general an infinite number of these as each derivative
is independent, but usually only the first few enter into any expressions. We shall
require the first three, which are all of the same order when defined by1
ǫ(φ) =
η(φ) =
ξ(φ) =
2
H(φ)#2
κ2 " H ′(φ)
,
2
κ2
2
κ2
H ′′(φ)
H(φ)
H ′′′(φ)
H ′(φ)
= ǫ −
= η −
ǫ′
,
√2κ2ǫ
2η′
√2κ2ǫ
.
(2.3)
The slow-roll approximation applies when these slow-roll parameters are small in
comparison to unity. The condition for inflation, a > 0, is precisely equivalent to
ǫ < 1.
The lowest-order expressions for the scalar (AS) and tensor (AG) amplitudes as-
sume {ǫ, η, ξ} are negligible compared to unity. Improved expressions for the scalar
and tensor amplitudes for finite but small {ǫ, η, ξ} were found by Stewart and Lyth
[13]:
H 2
H ′
√2κ2
8π3/2
κ
4π3/2 H [1 − (C + 1)ǫ] ,
AS ≃ −
AG ≃
[1 − (2C + 1)ǫ + Cη] ,
(2.4)
(2.5)
where C = −2 + ln 2 + γ ≃ −0.73 is a numerical constant, γ ≈ 0.577 being the Euler
constant. The right hand sides of these expressions are evaluated when the scale in
question crosses the Hubble radius during inflation, 2π/λ = aH. The spectra can
equally well be considered to be functions of wavelength or of the scalar field value.
Eq. (2.7) below allows one to move from one to the other.
The standard results to lowest-order are given by setting the square brackets to
unity. Historically it has been common even for this result to be written as only
1Let us stress that our choice φ > 0 implies √ǫ = −p2/κ2 H ′/H; one needs to be careful with
the signs to reproduce our results.
3
an approximate equality (the ambiguity arising primarily because of a vagueness in
defining the precise meaning of the density perturbation), though the precise nor-
malization to lowest-order was established some time ago by Lyth [14] (see also the
discussion in [4]).
The improved expressions for the spectra in Eqs. (2.4) and (2.5) are accurate in
so far as ǫ and η are sufficiently slowly varying functions that they can be treated
adiabatically as constants while a given scale crosses outside the Hubble radius. Cor-
rections to this would enter at next order. This differs from the usual situation in
which H is treated adiabatically. For the standard calculation to be strictly valid H
must be constant, but provided it varies sufficiently slowly (characterized by small ǫ
and η), it can be evaluated separately at each epoch. This injects a scale depen-
dence into the spectra. There is a special case corresponding to power-law inflation
for which ǫ and η are precisely constant and equal to each other. In this case the
above expressions for the perturbation spectra are exact [13,15]. Furthermore, the
corrections to each spectrum are the same and they cancel when the ratio is taken.
In the general case ǫ and η may be treated as different constants if it is assumed that
the timescale for their evolution is much longer than the timescale for perturbations
to be imprinted on a given scale. This assumption worsens as η is removed from ǫ,
which would be characterized by the next order terms becoming large.
Throughout we shall be quoting results which feature a leading term and a cor-
rection term linear in the slow-roll parameters. We shall utilize the symbol "≃"
to indicate this level of accuracy throughout. The correction terms shall be placed
in square brackets, so the lowest-order equations can always be obtained by setting
the square brackets equal to one. A useful relationship can be obtained from Eqs.
(2.3) -- (2.5):
ǫ ≃
A2
G
A2
S
[1 − 2C(ǫ − η)] .
(2.6)
As we shall see, η is encoded in the spectral index of the scalar perturbations, whose
deviation from unity must also be small for slow-roll to apply.
A key equation in Refs. [8,9] is the consistency equation, which connects the scalar
spectrum to the tensor spectrum and its derivative. The spectra as given in Eqs. (2.4)
and (2.5) are functions of the value of φ when the fluctuations crossed the Hubble
radius during inflation. This is converted into a dependence on wavelength λ with
the relation [9]
dλ
dφ
= λ
H
H ′
κ2
2
[1 − ǫ] .
Differentiation of Eq. (2.5) with respect to φ implies that
d ln AG
dφ ≃ −s κ2
2
AG
AS
[1 + (C + 2)ǫ − (C + 2)η] ,
4
(2.7)
(2.8)
and it follows that
λ
AG
dAG
dλ
=
A2
G
A2
S
[1 + 3ǫ − 2η] .
(2.9)
As expected this agrees with the expansion of the corresponding expression in Ref.
[9] for the special case ǫ = η. This equation is interesting in its own right, but for
perturbative reconstruction its use is restricted to the removal of derivatives of the
tensor spectrum.
III. PERTURBATIVE RECONSTRUCTION TO SECOND ORDER
The aim now is to obtain expressions for the potential and its derivatives about
a single point φ0, given information regarding the spectra at the scale λ0 which left
the horizon at φ = φ0. The four main quantities of observational interest are the am-
plitudes and spectral indices of the two spectra. However, in view of the consistency
equation, Eq. (2.9), only three of these are independent. We shall concentrate on the
two amplitudes and the scalar spectral index, since these are probably the easiest to
measure. The scalar spectral index n is defined by
1 − n =
S(λ)
d ln A2
d ln λ
.
To lowest-order in slow-roll one can show that the spectral index is given by [9]
1 − n = 4ǫ − 2η ,
(3.1)
(3.2)
which provides the route to determining η. Conceptually we are passing from these
three observables to the three parameters that describe the potential, which are the
slow-roll parameters ǫ and η, and the overall normalization. In terms of the observ-
ables, therefore, the slow-roll approximation amounts to an expansion in both A2
G/A2
S
and (1−n), which give corrections to the same order. We shall see that the correction
term for V ′′(φ0) requires the introduction of the third slow-roll parameter ξ, requiring
a new independent observable to determine it.
One obtains directly from the field equation Eq. (2.1) and the definitions of the
spectra in Eqs. (2.4) and (2.5) an expression for the amplitude of the potential:
V (φ0) ≃
≃
48π3
κ4 A2
48π3
κ4 A2
G(λ0)"1 +(cid:18)5
G(λ0)"1 + 0.21
3
G(λ0)
S(λ0)# ,
A2
+ 2C(cid:19) A2
S(λ0)# .
A2
A2
G(λ0)
(3.3)
(3.4)
In Refs. [8,9], we gave the numerical factor on the second-order term as −1/3, which
incorporated only the dynamical slow-roll corrections. In fact, the spectral corrections
5
to V (φ0) dominate the dynamical ones for any inflation model, reversing the sign
of the correction, which may be significant if the tensors are important. However
the relative contribution of the scalar and tensor modes to large angle microwave
anisotropies with our spectral normalization is given approximately by R [9], where
R =
2A2
S
25A2
G
.
(3.5)
Even if the contributions to the anisotropies from scalar and tensor modes are equal,
the correction term in the potential is only 2%. This is a powerful indication that
even the lowest-order perturbative reconstruction promises to be very accurate.
To obtain V ′(φ0) we need the scalar spectral index at λ0, denoted by n0. Differ-
entiation of the potential with respect to φ, followed by some straightforward algebra
gives
V ′(φ0) ≃ −
≃ −
≃ −
96π3
√2κ3
96π3
√2κ3
96π3
√2κ3
A3
G(λ0)
AS(λ0)
A3
G(λ0)
AS(λ0)
A3
G(λ0)
[1 + (C + 2)ǫ + (C − 1/3)η] ,
[1 + 1.27ǫ − 1.06η] ,
AS(λ0) "1 − 0.85
A2
A2
G(λ0)
S(λ0)
+ 0.53(1 − n0)# .
Note that for power-law inflation, which has [6]
(1 − n0) ≃
25
4π
A2
A2
G(λ0)
S(λ0)
,
(3.6)
(3.7)
the corrections in the square brackets nearly cancel, but other models [16,17] can
feature larger corrections, e.g. 16 % for natural inflation with n0 = 0.7.
The calculation for V ′′(φ0) is much more involved. One can show a precise rela-
tionship
V ′′(φ0)
H 2(φ0)
= 3(ǫ + η) −(cid:16)η2 + ǫξ(cid:17) .
(3.8)
A new observable will be needed to determine ξ, the easiest example being the rate of
change of the scalar spectral index. This would be substantially harder to measure,
and it is fortunate that it only enters at second-order.
[It would however enter at
leading order in V ′′′(φ0)]. From Eqs. (2.5) and (3.8), we can obtain the second-order
correction to V ′′(φ0) in terms of the slow-roll parameters
V ′′(φ0) ≃
48π3
κ2 A2
G(λ0) (ǫ + η)"1 + (2C + 2)ǫ −
η2 + ǫξ
3(ǫ + η)# .
(3.9)
It is however harder to convert the prefactor into observables, because there are now
lowest-order terms in both ǫ and η. To generate the correct second-order term, it is
6
not enough to use the first-order expression for η in terms of the spectral index. One
must instead use the second-order result, as given by Stewart and Lyth [13]
1 − n ≃ 4ǫ − 2η + 8(1 + C)ǫ2 − (6 + 10C)ǫη + 2Cǫξ ,
(our η being the negative of their δ), which leads to
(3.10)
4C + 4
5C + 3
ǫ + η ≃ 3ǫ(cid:20)1 +
S (cid:20)1 +
≃ 3
A2
G
A2
ǫ −
3
4 − 2C
3
η +
3
C − 3
3
ǫ +
η +
C
3
ξ(cid:21) −
ξ(cid:21) −
C
3
,
1 − n0
2
1 − n0
2
.
(3.11)
Substituting this into Eq. (3.9) yields
V ′′(φ0) ≃
144π3
A4
A2
G(λ0)
4C + 10
C − 3
S(λ0) (cid:20)1 +
G(λ0) (1 − n0) [1 + (2C + 2)ǫ] −
ǫ +
3
3
η +
C
3
ξ(cid:21)
16π3
κ2 A2
κ2
24π3
κ2 A2
−
G(η2 + ǫξ) ,
(3.12)
where the last term is entirely second-order. Note that there are two lowest-order
terms. An interesting case is η = −ǫ, corresponding to H ∝ φ1/2, for which the
lowest-order term vanishes identically and the final term of Eq. (3.9) is the only one
to contribute. The second derivative of the potential is the lowest derivative at which
it is possible for the expected lowest-order term to vanish.
The final step is to convert the second-order terms into the observables. As they
are already second-order, one only needs the lowest term in their expansion to convert.
From the expression for the spectral index, one finds to lowest-order that
+ 5η − 4ǫ .
(3.13)
1
2ǫ
ξ ≃
dn
d ln λ(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)λ0
Note that the derivative of the spectral index is of order ǫ2. To lowest-order we also
have
η ≃ 2ǫ −
1
2
(1 − n0);
ǫ ≃
A2
G
A2
S
.
Progressively substituting all these into Eq. (3.12) yields
3
(1 − n0) + (36C + 2)
2
A2
G(λ0)
A2
S(λ0)
(1 − n0)
A4
A4
G(λ0)
S(λ0)
V ′′(φ0) ≃
G(λ0)"9
16π3
A2
G(λ0)
κ2 A2
S(λ0) −
A2
1
(1 − n0)2 − (12C − 6)
4
3C − 1
−
+
2
dn
d ln λ(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)λ0# ,
7
(3.14)
(3.15)
where the first two terms are lowest-order and the remainder are second-order.2
For power-law models, the last term is zero and the remaining correction terms
nearly cancel each other, though they are not small individually. For natural inflation
models the correction terms are all individually small, with an overall correction of
about 4% at n0 = 0.7. (In natural inflation, (dn/d ln λ)λ0 ≃ η2 ≃ (1 − n0)2/16 [13].)
IV. DISCUSSION AND CONCLUSIONS
To conclude, we have calculated the full second-order corrections to the perturba-
tive reconstruction of the inflaton potential. The first-order terms agree with those
we found previously [8,9], while the second-order terms offer an improvement. They
serve to quantify the expected errors in the perturbative reconstruction, and in general
these errors are small. Even in cases where tensors provide a substantial contribu-
tion to the large angle microwave background anisotropies and/or the spectral index
deviates significantly from unity, the corrections are typically only a few percent.
Consequently, example figures based on plausible data sets that we presented in our
earlier papers remain valid. One then has some degree of confidence that one can use
our lowest-order expressions as was done in [18].
ACKNOWLEDGMENTS
EJC, ARL and JEL are supported by the Science and Engineering Research Coun-
cil (SERC) UK. EWK and JEL are supported at Fermilab by the DOE and NASA
under Grant NAGW -- 2381. ARL acknowledges the use of the Starlink computer sys-
tem at the University of Sussex. EJC and ARL thank the Aspen Center for Physics
for hospitality during the completion of this work. We would like to thank David
Lyth for helpful discussions.
[1] G. F. Smoot et al., Astrophys. J. Lett. 396, L1 (1992); E. L. Wright et al.,
Astrophys. J. Lett. 396, L13 (1992).
[2] A. Guth, Phys. Rev. D23, 347 (1981); A. Albrecht and P. J. Steinhardt, Phys.
Rev. Lett. 48, 1220 (1982); A. D. Linde, Phys. Lett. 108B, 389 (1982); A. D.
Linde, Phys. Lett. 129B, 177 (1983).
2Factoring out the lowest-order terms as in previous expressions leads to a very complicated
result, so we break with our convention regarding the use of square brackets.
8
[3] E. W. Kolb and M. S. Turner, The Early Universe, (Addison-Wesley, New York,
1990).
[4] A. R. Liddle and D. H. Lyth, "The Cold Dark Matter Density Perturbation,"
to be published, Phys. Rep. (1993).
[5] See Refs. 3 and 4 for surveys of different models.
[6] L. M. Krauss and M. White, Phys. Rev. Lett. 69, 869 (1992); R. L. Davis, H.
M. Hodges, G. F. Smoot, P. J. Steinhardt and M. S. Turner, Phys. Rev. Lett.
69, 1856 (1992); A. R. Liddle and D. H. Lyth, Phys. Lett. 291B, 391 (1992); J.
E. Lidsey and P. Coles, Mon. Not. R. astr. Soc. 258, 57P (1992); D. S. Salopek,
Phys. Rev. Lett. 69, 3602 (1992); F. Lucchin, S. Matarrese, and S. Mollerach,
Astrophys. J. Lett. 401, 49 (1992).
[7] A. R. Liddle, "Can the Gravitational Wave Background from Inflation be De-
tected Locally?" Sussex report SUSSEX-AST 93/7-3 (1993).
[8] E. J. Copeland, E. W. Kolb, A. R. Liddle and J. E. Lidsey, Phys. Rev. Lett.
71, 219 (1993).
[9] E. J. Copeland, E. W. Kolb, A. R. Liddle and J. E. Lidsey, to appear, Phys.
Rev. D, (Sept 15th 1993).
[10] H. M. Hodges and G. R. Blumenthal, Phys. Rev. D42, 3329 (1990).
[11] R. Crittenden, J. R. Bond, R. L. Davis, G. P. Efstathiou and P. J. Steinhardt,
Phys. Rev. Lett. 71, 324 (1993).
[12] D. S. Salopek and J. R. Bond, Phys. Rev. D42, 3936 (1990); J. E. Lidsey, Phys.
Lett. 273B, 42 (1991).
[13] E. D. Stewart and D. H. Lyth, Phys. Lett. 302B, 171 (1993).
[14] D. H. Lyth, Phys. Rev. D31, 1792 (1985).
[15] D. H. Lyth and E. D. Stewart, Phys. Lett. 274B, 168 (1992).
[16] F. C. Adams, J. R. Bond, K. Freese, J. A. Frieman and A. V. Olinto, Phys.
Rev. D47, 426 (1993).
[17] J. E. Lidsey, "Tilting the Primordial Power Spectrum with Bulk Viscosity,"
Fermilab report 93/099 -- A (1993).
[18] M. S. Turner, "Recovering the Inflationary Potential," Fermilab report 93/182-
A (1993).
9
|
astro-ph/9802154 | 1 | 9802 | 1998-02-12T08:49:21 | Gamma-Ray Burst Spectra and Time Histories from 2 to 400 keV | [
"astro-ph"
] | The Gamma-Ray burst detector on Ginga consisted of a proportional counter to observe the x-rays and a scintillation counter to observe the gamma-rays. It was ideally suited to study the x-rays associated with gamma-ray bursts (GRBs). Ginga detected 120 GRBs and 22 of them had sufficient statistics to determine spectra from 2 to 400 keV. Although the Ginga and BATSE trigger criteria were very similar, the distribution of spectral parameters was different. Ginga observed bend energies in the spectra down to 2 keV and had a larger fraction of bursts with low energy power law indexes greater than zero. The average ratio of energy in the x-ray band (2 to 10 keV) compared to the gamma-ray band (50 to 300 keV) was 24%. Some events had more energy in the x-ray band than in the gamma-ray band. One Ginga event had a period of time preceding the gamma rays that was effectively pure x-ray emission. This x-ray ``preactivity'' might be due to the penchant for the GRB time structure to be broader at lower energy rather than a different physical process. The x-rays tend to rise and fall slower than the gamma rays but they both tend to peak at about the same time. This argues against models involving the injection of relativistic electrons that cool by synchrotron radiation. | astro-ph | astro-ph |
Gamma-Ray Burst Spectra and Time Histories From 2 to 400 keV
E. E. Fenimore
MS D436, Los Alamos National Laboratory, Los Alamos, NM 87544
The Gamma-Ray burst detector on Ginga consisted of a proportional counter to observe the x-rays and a
scintillation counter to observe the gamma-rays. It was ideally suited to study the x-rays associated with gamma-
ray bursts (GRBs). Ginga detected ∼ 120 GRBs and 22 of them had sufficient statistics to determine spectra
from 2 to 400 keV. Although the Ginga and BATSE trigger criteria were very similar, the distribution of spectral
parameters was different. Ginga observed bend energies in the spectra down to 2 keV and had a larger fraction
of bursts with low energy power law indexes greater than zero. The average ratio of energy in the x-ray band
(2 to 10 keV) compared to the gamma-ray band (50 to 300 keV) was 24%. Some events had more energy in the
x-ray band than in the gamma-ray band. One Ginga event had a period of time preceding the gamma rays that
was effectively pure x-ray emission. This x-ray "preactivity" might be due to the penchant for the GRB time
structure to be broader at lower energy rather than a different physical process. The x-rays tend to rise and fall
slower than the gamma rays but they both tend to peak at about the same time. This argues against models
involving the injection of relativistic electrons that cool by synchrotron radiation.
1. INTRODUCTION
2. INSTRUMENTAL DETAILS
Gamma-ray bursts (GRBs) are aptly name,
most of their power is usually emitted in the 100
keV to 1 MeV energy range. Early results indi-
cated that only a few percent of the energy of a
Gamma-Ray burst (GRB) occurs in the 2-to-10
keV x-ray range [1,2], although the X-ray emis-
sion might outlast the main GRB event in some
bursts (so called X-ray tails). Based on these
intriguing results, the gamma-ray burst detec-
tor (GBD) flown aboard the Ginga satellite was
specifically designed to investigate burst spectra
in the X-ray regime [3]. Ginga was launched in
February of 1987, and the GBD was operational
from March, 1987, until the reentry of the space-
craft in October, 1991.
More recently, the BeppoSax satellite has ob-
served several bursts in both x-rays and gamma-
rays [4 -- 6] and has discovered soft x-ray afterglows
[7]. This latter discovery has opened the way for
the long-sought GRB counterparts, including one
with a measured redshift [8].
Here, we summarize x-ray results from the
Ginga experiment.
The GBD aboard Ginga consisted of a propor-
tional counter (PC) sensitive to photons in the 2-
to-25 keV range and a scintillation counter (SC)
recording photons with energies between 15-to-
400 keV. Each detector had an ≈ 60 cm2 effective
area. The PC and SC provided 16 and 32 channel
spectra, respectively, over their indicated energy
ranges. The detectors were uncollimated except
for the presence of shielding to reduce backside
illumination and the mechanical support for the
PC window. The field of view was effectively π
steradian for both the PC and the SC.
Each detector produced both spectral and time
history-data for GRBs. Photons within the en-
ergy range of each detector were accumulated
into time samples. The hardware ensured that
the time samples for both detectors started and
stopped at the same time. This facilitates mea-
suring the relative timing between the x-rays
and gamma-rays. The temporal resolution of
the time-history data depended on the teleme-
try mode. An on-board trigger system checked
the time histories for a significant increase (11σ
on either 1/8, 1/4, or 1 s time scales). Upon such
a trigger, special high resolution temporal data
(called "Memory Read Out" [MRO]) would be
2
produced. The MRO data has time resolution of
0.03125 s. The PC MRO time history extends
from 32 s before the trigger to 96 s after the trig-
ger. The SC MRO time history extends from 16
s before the trigger to 48 s after the trigger. Dur-
ing the period when both detectors have MRO
time histories, the MRO samples were in phase
with each other. In addition to the MRO data,
"Real Time" data were often continuously avail-
able. Three different telemetry modes resulted in
PC and SC time histories with either 0.125, 1.0,
or 4.0 s time resolution.
In burst mode, the GBD recorded spectral data
from the PC and SC at 0.5 s intervals for 16 s be-
fore the burst trigger time and for 48 s after the
trigger. The PC and SC had 16 and 32 energy
channels, respectively. The MRO data were used
for many of the spectral fits described here. In
the event that MRO data were not available for a
burst, we utilized the spectral data from the "real
time" telemetry modes. For these bursts, spectral
data were available with either 2, 16, or 64 s accu-
mulations. For the longer accumulations, spectral
studies were not generally feasible. See [3] and [9]
for more information concerning the instrument.
The Ginga GBD was in operation from March
1987 to October 1991. During this time ≈ 120
γ-ray bursts were identified [10,11]. From this
group, a sample of 22 events with statistically
significant spectral data were analyzed [9]. These
events occurred within the forward, π steradian
field of view of the detectors (front-side events).
Front-side events are easily identified because
they have consistent fluxes in the energy range
observed by both instruments (15 to 25 keV). Ex-
cluded bursts usually show strong rolloffs below
25 keV and spectral fits with an absorption com-
ponent due to aluminum (a principal spacecraft
material) are consistent with this interpretation.
Because the incidence angle was known for just
four of the events, we had to assume an incidence
angle for the remaining events. We selected 37
degrees for the incidence angle when the angle
was unknown. This is a typical angle consider-
ing that the mechanical support for the window
on the PC acts as a collimator limiting the field
of view to an opening angle of ∼ ±60 degrees.
Extensive simulations [9] demonstrated that our
conclusions were not affected by the uncertainty
in the incidence angle.
We have adopted the spectral model employed
by [12] because of its relative simplicity and abil-
ity to accurately characterize a wide range of
spectral continua. In addition, this choice facili-
tates direct comparison of our results with those
obtained from bursts seen by the Burst and Tran-
sient Source Experiment (BATSE). This model
has the form
N (E) = AEα exp(−E/E0)
(α − β)E0 ≥ E
N (E) = A [(α − β)E0]α−β exp(β − α)Eβ
if
and
if
(α − β)E0 ≤ E.
Here, A is an overall scale factor, α is the low-
energy spectral index, β is the high-energy spec-
tral index, and E0 is the exponential cutoff or
bend energy. The model parameters are adjusted
iteratively until a minimum in χ2 is obtained.
3. X-RAY CHARACTERISTICS OF
GINGA BURSTS
Of particular interest is the behavior of the
Ginga sample at x-ray energies. About 40% of
the bursts in the sample show a positive spectral
number index below 20 keV (i.e., α > 0), with
the suggestion of rolloff toward lower energies in
a few of the bursts (α as large as +1.5) [9]. Un-
fortunately, the lack of data below 1 keV, and the
often weak signal below 5-to-10 keV precludes es-
tablishing the physical process (photoelectric ab-
sorption, self-absorption) that may be involved
in specific bursts. Observations of the low-energy
asymptote can place serious constraints on several
GRB models, most notably the synchrotron shock
model which predicts that α should be between
-3/2 and -1/2 [13]. Crider et al. [14] uses exam-
ples from BATSE to argue that some GRB vio-
late these limits during some time-resolved sam-
ples. We find violations of these limits in the
time-integrated events.
In our sample of bursts, we find that α can be
both less than zero and greater than zero. Nega-
tive α's are often seen in time-integrated BATSE
spectra. Positive α's for which the spectrum rolls
over at low energies are usually only seen in time-
resolved BATSE spectra [14].
Our distribution of parameters (α, β, E0) is
broader than that found by BATSE. For exam-
ple, 40% of our events had α greater than 0 com-
pared to only 15% of BATSE events, and we had
bend energies that occurred as low as 1.7 keV,
wheras the lowest BATSE bend energy was 15
keV. The Ginga trigger range (50 to 400 keV)
was virtually the same as BATSE's. Thus, we
do not think we are sampling a different popu-
lation of bursts, yet we get a different range of
fit parameters. One possible explanation might
be that GRBs have two break energies, one often
in the 50 to 500 keV range and the other near 5
keV. Both BATSE and Ginga fit with only a sin-
gle break energy so BATSE tends to find breaks
near the center of its energy range, and we tend
to find breaks in our energy range. Without good
high energy observations of bursts with low E0,
it is difficult to know whether they also have a
high-energy bend.
4. X-RAY EMISSION RELATIVE TO
GAMMA-RAY EMISSION
Early measurements of the x-rays associated
with gamma-rays were fortuitous observations by
collimated x-ray detectors that just happen to
catch a GRB in their field of view. From a few
events it appeared that the amount of energy in x-
rays was only a few percent (Laros et al., 1984 [1])
confirming that GRBs were, indeed, a gamma-ray
phenomena. The Ginga experiment was designed
with a wide field of view to detect a sufficient
number of events to determine the range of x-ray
characteristics. Early reports from Ginga events
indicated that sometimes a much larger fraction
of the emitted energy was contained in the x-rays.
For example, by comparing the signal in the pro-
portional counter (roughly 2 to 25 keV) to that
of the scintillator (roughly 15 to 400 keV), we re-
ported an x-ray to gamma-ray emission ratio up
to ∼46% (Yoshida et al., 1989 [15]). Such a ratio
3
Figure 1.
The distribution of the ratio of the
energy emitted in x-rays relative to that emitted
in gamma-rays. The x-ray bandpass is defined to
be from 2 to 10 keV, and the gamma-ray bandpass
is the BATSE range of 50 to 300 keV. Note there
are some examples of equal energy in the x-rays
and the gamma-rays. The average ratio of the
energy in the x-rays to the energy in the gamma-
rays is 24%. From Strohmayer et al., 1998.
2 EN (E)dE/ R 300
depends on the bandpass for which it is evaluated.
Using the best fit α, β, and E0 from [9], one can
find the ratio of emission for a typical x-ray band-
pass (2 to 10 keV) compared to the BATSE en-
ergy range (50 to 300 keV). The ratio is defined to
be Rx/g = R 10
50 EN (E)dE. Fig-
ure 1 presents the distribution for the 22 events
analyzed in [9]. Although the ratio is often a
few percent, for some events the ratio is near
(or larger than) unity.
Some GRBs actually
have more energy in the x-ray bandpass than the
gamma-ray bandpass. The simple average of the
22 values is 24%. This large value arises because
of the few events with nearly equal energy in the
x-ray and gamma-ray bandpass. However, even
the logarithmic average is 7%.
4
5. X-RAY PREACTIVITY OR PULSE
SPREADING?
It is common for the x-ray emission to last
longer than the gamma-ray emission [4 -- 6,15 -- 17].
There is one clear example of x-ray activity pre-
ceding the GRB [16]. Figure 2 presents the time
history of GB900126 in several energy bands.
From top to bottom, one sees an x-ray hard-
ness ratio, the count rate in the PC (1-to-28
keV, 0.03125 s resolution), and five time histo-
ries based on the PC and SC pulse height analy-
sis (PHA) data (0.5 s resolution). The horizontal
arrow in the bottom panel indicates the period
that is effectively pure x-rays. During the period
marked by the arrow, there is no detectable emis-
sion above 7 keV, but there is significant emission
below 7 keV (see Murakami et al. [16]).
It would be misleading to refer to the x-ray
phase as a "precursor" because it is not a sep-
arate peak. "Preactivity" gives the connotation
of a separate cause for the emission although not
necessarily as distinct as "precursor" might imply.
However, the x-ray phase in Figure 2 could be due
to the penchant for peaks to be wider at low en-
ergy. Figure 3 shows the average autocorrelation
of GRB time histories in four energy bands. At
higher energy, the autocorrelation is wider. This
spreading of the time structure with energy is also
seen in the average rise and fall times of individ-
ual pulses [18]. On average, the peak temporal
width scales as E−0.45. The spreading in Figure
2 is larger but that could reflect that some bursts
have larger spreading than the average.
Perhaps there is not a separate cause for the x-
ray emission, but rather, the physics responsible
for the spreading as a function of energy causes
the x-rays to appear to turn on before the gamma-
rays.
The physics responsible for the spreading is un-
clear. Kazanas, Titarchuk, & Hua (1998) [19]
have suggested two processes, synchrotron cool-
ing of injected relativistic electrons and Compton
downscattering of injected photons. Synchrotron
cooling of ejected electrons can produce a time
structure that scales as E−1/2. However, such
cooling should cause the emission at lower en-
ergy to peak later. In most GRBs, the rise and
the fall are slower at low energy (causing the
E−0.45 spreading) but the peak of the emission
is not delayed substantially. In the calculations
of Kazanas, Titarchuk, & Hua, the low energy
emission (e.q., 25-to-50 keV) starts to rise after
the high energy emission (e.q., 300-t0-1000 keV)
has completely fallen. Because GRBs never seem
to do this, we think it strongly argues against syn-
chrotron cooling as the mechanism that produces
the peaks in GRBs.
REFERENCES
1. Laros, J. G., et al., 1984, ApJ, 286, 681.
2. Katoh, M., et al., 1984, in High Energy Tran-
sients in Astrophysics, AIP Conf. 115, ed. S.
E. Woosley (AIP New York), 390.
3. Murakami, T., et al., 1989, Publ. Astron. Soc.
Jap., 41, 405.
4. Piro, L., et al., 1997, A&A, in press, astro-
ph/9707215.
5. Piro, L., et al., 1998, A&A, in press, astro-
ph/9710334.
6. Frontera, F., et al., 1998, ApJ, in press, astro-
ph/9711279.
7. Costa, E., et al., 1997, Nature, 387, 783,
astro-ph/9706065.
8. Metgzer, M., et al., 1997, Nature, 387, 878.
9. Strohmayer, T., et al., 1998, ApJ, in press,
astro-ph/9712332.
10. Ogasaka, Y., et al., 1991, ApJ, 383, L61.
11. Fenimore, E. E., et al., 1993, in AIP Confer-
ence Proceeding 280, Compton Gamma-Ray
Observatory, ed. M. Friedlander, N. Gehrels,
& D. Macomb, (New York:AIP), 917.
12. Band, D., et al., 1993, ApJ, 413, 281.
13. Katz, J. L., 1994, ApJ, 432, L107.
14. Crider, A., et al., 1997, 479, L93.
15. Yoshida, A., et al., 1989, PASJ, 41, 509.
16. Murakami, T., et al., 1991, Nature, 350, 592.
Inoue, H., van Paradijs,
17. Murakami, T.,
J., Fenimore, E., & Yoshida, A., 1992,
in
Gamma-Ray Bursts, ed. C. Ho, R. I. Epstein
& E. E. Fenimore, (Cambridge: Cambridge
University Press), 239.
18. Fenimore, E. E., et al., 1995, ApJ, 448, L101.
19. Kazanas, D., Titarchuk, L. G., & Hua, X.-M.,
1998, ApJ, in press, astro-ph/9709180.
5
Figure 2. The temporal evolution of GB900126
demonstrating the x-ray "preactivity." The top
panel is the x-ray hardness and the next panel is
the total count rate in the proportional counter.
The next 5 panels use the energy-resolved PHA
data in 5 energy bands from 1 to 370 keV. Note
that the peak widths become wider with lower
energy but they are not substantially shifted in
time. Synchrotron cooling of ejected electrons is
expected to produce substantial shifts at lower en-
ergy. In the bottom panel, the time period mark
with a horizontal arrow is the period of x-ray pre-
activity during which the emission occurs only at
energies less than 7 keV. (From Murakami et al.,
1989.)
Figure 3. Average autocorrelation of 45 bright
BATSE gamma-ray bursts in four energy chan-
nels. At higher energy, GRBs have shorter
timescales. The solid curves are fits of the sum
of two exponentials to the autocorrelation his-
togram. (From Fenimore et al. 1995.)
|
astro-ph/9903375 | 1 | 9903 | 1999-03-24T17:49:15 | SNRs in the Galactic Center region observed with ASCA | [
"astro-ph"
] | We report the ASCA results of the supernova remnants (SNRs) and their candidates in the Galactic Center region. We found apparent X-ray emission from G359.1-0.5 and G0.9+0.1, and made marginal detection for G359.1+0.9, but found no significant X-ray from the other cataloged SNRs: G359.0-0.9, Sgr A East (G0.0+0.0), G0.3+0.0, Sgr D SNR (G1.0-0.1) (Green 1998). The emission from G359.1-0.5 is found to be thermal with multi temperature structures whereas that from G0.9+0.1 is quite hard, probably non-thermal. We discovered two new candidates of SNRs: G0.0-1.3 (AX J1751-29.6) and G0.56-0.01 (AX J1747.0-2828). The former, G0.0-1.3, shows the extended emission with a thin thermal plasma. The latter, G0.56-0.01, shows quite strong 6.7-keV line with the equivalent width of 2 keV, which resembles that of the Galactic Center plasma. We discuss the nature of those SNRs, relating with the origin of the Galactic Center hot plasma. | astro-ph | astro-ph |
SNRs in the Galactic Center region observed with ASCA
M. Sakano1,3, J. Yokogawa1, H. Murakami1, K. Koyama1,5 Y. Maeda2,4 and The ASCA Galactic
Plane/Center Survey team
1 Department of Physics, Kyoto University, Sakyo, Kyoto 606-8502, Japan
2 Department of Astronomy and Astrophysics, Pennsylvania State University, University Park PA 16802-6305 U.S.A.
3 Research Fellow of the Japan Society for the Promotion of Science
4 Postdoctctoral Fellow for Research Abroad of JSPS
5 CREST: Japan Science and Technology Corporation (JST)
Abstract. We report the ASCA results of the super-
nova remnants (SNRs) and their candidates in the Galac-
tic Center region. We found apparent X-ray emission
from G359.1−0.5 and G0.9+0.1, and made marginal
detection for G359.1+0.9, but found no significant X-
ray from the other cataloged SNRs: G359.0−0.9, Sgr
A East (G0.0+0.0), G0.3+0.0, Sgr D SNR (G1.0−0.1)
(Green 1998). The emission from G359.1−0.5 is found to
be thermal with multi temperature structures whereas
that from G0.9+0.1 is quite hard, probably non-thermal.
We discovered two new candidates of SNRs: G0.0−1.3
(AX J1751−29.6) and G0.56−0.01 (AX J1747.0−2828).
The former, G0.0−1.3, shows the extended emission with a
thin thermal plasma. The latter, G0.56−0.01, shows quite
strong 6.7-keV line with the equivalent width of 2 keV,
which resembles that of the Galactic Center plasma. We
discuss the nature of those SNRs, relating with the origin
of the Galactic Center hot plasma.
1. Introduction
The hard X-ray structure of the Galactic Center region is
quite unusual as well as those of the other wavelengths.
In particular, thin thermal hot plasma prevails all over
the region (Koyama et al. 1989; Yamauchi et al. 1990;
Koyama et al. 1996; Maeda 1998). It has several surpris-
ing natures. First, the temperature is quite high, about
10 keV. Second, the total energy is also high, about 1054
erg. Third, the plasma wide-spreads in the region bigger
than 1◦×1◦, which corresponds to about 150pc×150pc for
the distance of the Galactic Center. Fourth, the spectrum
of the plasma is surprisingly uniform from field to field
except for the slight change of the surface brightness. The
origin of the plasma is still an enigma.
Supernova remnants (SNRs) are one of the possi-
ble candidates for the origin of the hot plasma (e.g.,
Koyama et al. 1986), as well as the other candidates;
the past activity of the central massive black hole (e.g.,
Koyama et al. 1996), unresolved many cataclysmic vari-
ables (e.g., Mukai & Shiokawa 1993), and the global mag-
netic activity which heats the interstellar matter (e.g.,
Yokoyama et al. 1998).
If the SNRs explain all the hot plasma, first, the to-
tal energy (∼ 1054 erg) requires about 103 supernovae
in the narrow region with the scale of a few hundred pc
in the last 5 104 years, which is the age of the plasma
(Koyama et al. 1996). It would lead to the much higher
supernova rate than the common rate of a few supernovae
in a century in all the Galaxy. However, the possible star-
burst activity in the Galactic Center region has been often
discussed. In particular, recent COMPTEL observations of
26Al 1.8 MeV line suggest that 105 supernovae occurred
in the last 4 106 years (Chen et al. 1995; Hartmann 1995),
which is consistent with the required supernova rate.
Second, the supernova origin hypothesis requires the
mean temperature of about 10 keV. It is much higher than
the usual temperature of SNRs (kT ≤ a few keV). How-
ever, if SNRs densely exist in the region, the temperature
may heat up to about 10 keV by the mutual interaction
of the shocks. Thus, SNRs are still one of the candidates
for the origin of the Galactic Center plasma.
In this paper we concentrate on the X-ray natures
of the SNRs in the Galactic Center region from the ob-
servational point of view. The ASCA capability of imag-
ing spectroscopy with 0.5 -- 10 keV (Tanaka et al. 1994) is
quite suitable for the study. In fact, owing to the hard
X-ray sensitivity of ASCA, particularly above 2 keV, we
can possibly detect new SNRs which have been hidden
by the heavy interstellar absorption, hence obtain some
information whether the SNRs are valid for the origin of
the Galactic Center plasma.
Using all the available ASCA pointing observations
in this region (−1◦< l <1◦, −1◦< b <1◦) and a part
of the data of the ongoing ASCA Galactic Center sur-
vey, we studied the X-ray emission from the known SNRs
(Green 1998), and searched new SNRs. We assume the
distance to the Galactic Center to be 8.5 kpc.
2
Sakano et al.: SNRs in the Galactic Center region observed with ASCA
2. Results on the cataloged SNRs
We searched X-ray emission from the cataloged SNRs
(Green 1998), and then investigated each nature where
significant X-ray emission is detected.
2.1. G359.1−0.5
Fig. 1 (upper panel) shows the ASCA GIS image of
G359.1−0.5 with 1.6 -- 2.1 keV band. We detected center-
filled X-rays from this source, whereas the radio image
shows a clear shell-like structure (e.g., Uchida et al. 1992).
The spectrum is found to have emission lines from
highly ionized ion (Fig. 1 (lower panel)), hence the X-
rays come from thin thermal plasma. The most dis-
tinct two lines are Kα lines from helium-like silicon
and hydrogen-like sulfur. They imply that the plasma
has multi-temperature structure. The absorption column
density is estimated at NH∼ 8 1022H cm−2, suggest-
ing that G359.1−0.5 is located at near the Galactic
Center, which is consistent with the radio observations
(Uchida et al. 1992). This column density is larger by a
factor of 2 or 3 than that with the ROSAT measurement
(Egger & Sun 1998). However, we are still confident of our
estimation because the ASCA energy band is the most
suitable for measuring such heavy absorption. Details of
the analysis are given in Yokogawa et al. (1999).
2.2. G0.9+0.1
Fig. 2 (upper panel) shows the GIS image of G0.9+0.1
with 3 -- 10 keV band. We detected significant X-ray emis-
sion from this source in this hard energy band, whereas
no significant X-ray is detected in the softer energy band.
The X-ray emitting region is compact and not resolved
with ASCA GIS. The radio size of the source is 2′ in diam-
eter (Helfand & Becker 1987), hence it is consistent with
the X-ray image.
The spectrum is found to be hard; kT >2 keV
(the best-fit of 40 keV) in thermal model or Γ ∼1.5
in power-law model, with heavy absorption of NH∼
1023H cm−2, although the statistics are not good. This
hardness may imply the emission to be non-thermal ori-
gin. The flux is 2 10−12 erg s−1 cm−2 in 2 -- 10 keV
band. It is consistent with the upper limit by Einstein
(Helfand & Becker 1987), but significantly lower than the
SAX result (Mereghetti et al. 1999).
2.3. G359.1+0.9
Fig. 3 (upper panel) shows the GIS image around
G359.1+0.9. X-ray emission from the position correspond-
ing to G359.1+0.9 is detected with the significance of 9.8σ
in 0.7 -- 3 keV band (AX J1739.6−2911 in Fig. 3), whereas
the significance in 3 -- 10 keV band is 2.1σ.
Faint extended emission around AX J1739.6−2911
also can be seen in the 0.7 -- 3 keV band image. We
made the radial profile with the center at the peak
of AX J1739.6−2911 and fitted it with the model of
the point spread function (PSF) and the background
(NXB+CXB(cosmic X-ray background)) where the nor-
malizations of the PSF and the background were allowed
to be free. The model is found to be rejected with the con-
fidence of 97.4%, i.e., the slightly extended emission with
the radius of 4′∼5′ was marginally detected with the sig-
nificance of 2.2σ (Fig. 3 lower panel). The size is consistent
with the radius of the radio shell, r ∼5′.
the
central
region
The
spectrum of
of
AX J1739.6−2911 is
found to be well fitted with
the thin thermal plasma model with kT ∼0.7 keV
(χ2/d.o.f.=2.67/5) and NH∼ 0, although the statistics
are not good. The flux is 3 10−13 erg s−1 cm−2 in total
energy band, converted to the luminosity of 2 1033 erg s−1
if we assume the distance to be 8.5 kpc. Note that the soft
spectrum with no absorption may imply that this source
is not located in the Galactic Center region, but is a
foreground source, hence the above estimated luminosity
may be reduced by several factors or more.
The extended emission with positional coincidence
with the radio structure strongly supports that it is an
X-ray counter part of the SNR, if the detection of the
extended emission is true. The soft spectrum also sup-
ports it. The luminosity may be rather low for an SNR.
However, ASCA Plane survey has failed to detect many
radio SNRs, suggesting their quite dim X-ray luminosities
(Yamauchi & ASCA Galactic Plane Survey team 1999).
Therefore, the low X-ray luminosity of this source may
be acceptable for an SNR. Future observations with high
sensitivity will be encouraged.
2.4. The other cataloged SNRs
For the other cataloged SNRs (G359.0−0.9, Sgr A East
(G0.0+0.0), G0.3+0.0, Sgr D SNR (G1.0−0.1)), we de-
tected no significant X-ray emission associated with the
SNRs. It is possibly because no strong X-ray is actu-
ally emitted. However we should be conservative for no
X-ray emission from those sources because the detected
positions of those SNRs are heavily contaminated from
nearby strong X-ray sources; SLX 1744−299/300 near
G359.0−0.9, Sgr A and AX J1745.6−2901 near Sgr A
East, 1E 1743.1−2843 near G0.3+0.0, and GX3+1 near
G1.0−0.1.
3. Results on the new candidates of SNRs
We discovered two new candidates of SNRs. In this sec-
tion, we report the preliminary results on them.
Sakano et al.: SNRs in the Galactic Center region observed with ASCA
3
3.1. G0.0−1.3 (AX J1751−29.6)
Fig. 4 (upper panel) shows the GIS image of G0.0−1.3
(AX J1751−29.6) in 0.7 -- 3 keV band. We discovered the
clearly extended emission with the scale of about 40′×10′.
The spectrum is found to have the emission lines from
highly ionized ions, hence is the thermal origin (Fig. 4
lower panel). In fact, the spectrum is well fitted with a
thin thermal plasma model with kT = 0.5 ± 0.08 keV and
NH= (1.3 ± 0.2) 1022 H cm−2. The X-ray flux is ∼ 10−11
erg s−1 cm−2 in 0.5 -- 3 keV band.
The column density suggests that this source may be
in front of the Galactic Center region according to Sakano
et al. (1999), and be located at the distance of about 4 kpc
if we assume the mean interstellar density of 1 H cm−3.
Then, the obtained flux is converted to the luminosity
of 3 1035 erg s−1 under the assumption of the distance of
4 kpc. Even in the case of the quite small distance of 1 kpc,
the luminosity is larger than 1034 erg s−1.
We now consider the classification of the source. The
clearly extended emission of the thin thermal plasma with
kT ∼0.5 keV implies that this source is an SNR or a star
forming region. The luminosity is also within the typical
range of SNRs but much higher than that of a star forming
region. Therefore, AX J1751−29.6 is a strong candidate
for a new SNR.
3.2. G0.56−0.01 (AX J1747.0−2828)
Fig. 5 (upper panel) shows the GIS image of G0.56−0.01
(AX J1747.0−2828) in 6 -- 7 keV band, where this source
can be seen the most significantly. The X-ray emitting
region is compact and not resolved with GIS.
The background subtracted spectrum is given in
Fig. 5 (lower panel). We accumulated the source X-
ray photons from the 3′-radius circular region around
AX J1747.0−2828, and the background photons, from an
elliptical region with the major axis parallel to the Galac-
tic Plane, excluding the 3′-radius circular regions around
AX J1747.0−2828 itself and Sgr B2 (see Murakami et al.
(1999)).
The spectrum is found to be characterized with a quite
strong line at between 6 -- 7 keV, which is also implied from
the X-ray image. We tried to fit the spectrum with the
thermal bremsstrahlung and a Gaussian line. Then we
found the equivalent width of the line to be quite large,
∼2 keV, and the center energy of the Gaussian to be
6.63±0.06 keV, being consistent with Kα line from helium-
like iron. Hence, the spectrum is definitely a thin thermal
origin with high temperature of several keV or higher.
We then fitted the spectrum with a thin thermal
plasma model. The model well represents the total spec-
tral shape. The best-fit temperature is kT = 6.0+1.9
−1.5 keV,
the abundance, Z > 2 solar, and the hydrogen column
density, NH= (6.1+1.6
−1.1) 1022 H cm−2. The flux is 1.6 10−12
erg s−1 cm−2 in 0.7 -- 10 keV band.
The large column density suggests this source to be
located at near the Galactic Center. Thus, the absorp-
tion corrected X-ray luminosity is estimated at ∼3.6 1034
erg s−1 under the assumption of the distance of 8.5 kpc.
This high temperature and the overabundance suggest
AX J1747.0−2828 to be a possible new candidate of a
young SNR. The luminosity is also within the range of
that of the typical SNR. Although the supernova rem-
nant is the most probable source, the other possibility, for
example, a cataclysmic variable, still cannot be excluded
(e.g., Terada et al. 1999). In any case, it is a new class
object in the Galactic Center region, which is a good can-
didate for the origin of the Galactic Center plasma.
4. Discussion
SNRs are one of the possible candidates for the origin of
the Galactic Center hot plasma.
With ASCA, we completely surveyed the region of
l < 1◦and b < 0.3◦, where a large portion of SNRs in
this region are expected to exist, and partially surveyed
some areas with larger galactic latitude. The number of
the detected SNRs with ASCA is two or three in 7 cata-
loged SNRs, and two for the new candidates in the Galac-
tic Center region. On the other hand, the required number
to be responsible for the Galactic Center plasma is about
103 in the region of l ≤ 1◦and b ≤ 0.5◦. Therefore, the
number of the detected SNRs is quite insufficient.
From recent
line
radio molecular
observations,
Hasegawa et al. (1998) found over 300 shell-like struc-
tures, possibly SNRs, in the region of l ≤ 0.5◦, which
correspond to about 20 shell-like structures along any line
of sight in the region. Therefore, the number of SNRs may
be possibly too much to resolve with ASCA. Future obser-
vations with higher angular resolution would be required
to solve this problem.
The spectra of four of the detected five SNRs (or can-
didates) are relatively soft (G359.1−0.5, G359.1+0.9, and
G0.0−1.3 (AX J1751−29.6)) or have no line-like feature
(G0.9+0.1). Thus they cannot explain the spectrum of the
Galactic Center plasma. On the other hand, a new SNR
candidate G0.56−0.01 (AX J1747.0−2828) is notable for
the strong iron line similar to that of the Galactic Center
plasma. Such class of objects may be a key to understand
the origin of the plasma. Search for such objects will be
strongly encouraged.
Acknowledgements. The authors express their thanks to all the
members of the ASCA team. We are grateful to Drs. J. P.
Hughes and P. Slane for their valuable comments. MS and
YM acknowledge the support from Research Fellowships of the
Japan Society for the Promotion of Science.
References
Chen, W., Gehrels, N., Diehl, R., 1995, ApJ 440, L57
4
Sakano et al.: SNRs in the Galactic Center region observed with ASCA
250
Egger, R., Sun, X., 1998, In: "The Local Bubble and Be-
yond", Breitschwerdt D., Freyberg M.J., Trumper J. (eds.),
Springer, IAU Colloq. 166, 417
nants", http://www.mrao.cam.ac.uk/surveys/snrs/
Green, D.A., 1998, In: "A Catalog of Galactic Supernova Rem-
200
Hartmann, D.H., 1995, ApJ 447, 646
Hasegawa, T., Oka, T., Sato, F., In: The Physics and Chem-
istry of the Interstellar Medium (Abstract Book), Aachen
V.O.S.-V.
150
Helfand, D.J., Becker, R.H., 1987, ApJ 314, 203
Koyama, K., Awaki, H., Kunieda, H., et al., 1989, Nature 339,
603
Koyama, K., Ikeuchi, S., Tomisaka, K., 1986, PASJ 38, 503
100
Koyama, K., Maeda, Y., Sonobe, T., et al., 1996, PASJ 48, 249
Maeda, Y., 1998, Ph.D thesis, Kyoto University
Mereghetti, S., Sidoli, L., Israel, G.L., 1998, A&A 331, L77
Mukai, K., Shiokawa, K., 1993, ApJ 418, 863
Murakami, H., Koyama, K., Sakano, M., Maeda, Y., 1999, this
50
Sakano, M., Koyama, K., Nishiuchi, M., et al., 1999, Advances
volume
in Space Research, in press
Tanaka, Y., Inoue, H., Holt, S.S., 1994, PASJ 46, L37
Terada, Y., Kaneda, H., Makishima, K., et al., 1999, preprint
Uchida, K.I., Morris, M., Yusef-Zadeh, F., 1992, AJ 104, 1533
Yamauchi, S., ASCA Galactic Plane Survey team, 1999, this
volume
Yamauchi, S., Kawada, M., Koyama, K., et al., 1990, ApJ 365,
532
Yokogawa, J., Sakano, M., Koyama, K., Yamauchi, S., 1999,
Advances in Space Research, in press
Yokoyama, T., Tanuma, S., Kudoh, T., Shibata, K., 1998, In:
The central regions of the Galaxy and galaxies, Sofue Y.
(ed.), Kluwer Academic Publishers, IAU Symp. 184, 355
1.6-2.1keV GIS
A1742-294
+0(cid:176) 20¢
-0(cid:176) 20¢
-0(cid:176) 40¢
1E1740.7-2942
+0(cid:176) 00¢
source region
-1(cid:176) 00¢
50
-0(cid:176) 40¢
100
the Snake
background region
radio shell
150
-0(cid:176) 20¢
200
250
-2
10
-3
10
V
e
k
/
s
/
t
n
u
o
c
-4
10
2
0
-2
1
2
Energy (keV)
5
Fig. 1.
(Upper) GIS contour map with 1.6 -- 2.1 keV
band superposed on a schematic diagram of the radio
structures; the radio shell of G359.1−0.5 and the ra-
dio non-thermal filament, the Snake. Contour level
is
linearly spaced and is saturated for A1742−294 and
1E 1740.7−2942. The accumulated regions for the spectra
are also noted. (Lower) The background-subtracted GIS
spectrum. We also show the best-fit model where we fit the
spectrum with the model of the thermal bremsstrahlung
and two narrow Gaussians with interstellar absorption.
These two figures are adopted from Yokogawa et al.
(1999).
c
Sakano et al.: SNRs in the Galactic Center region observed with ASCA
5
220
200
180
160
140
120
100
80
60
40
+1(cid:176) 20¢
+1(cid:176) 00¢
+0(cid:176) 20¢
G0.9+0.1
250
-0(cid:176) 20¢
PSR B1736-29
200
AX J1740.3-2904
+1(cid:176) 40¢
AX J1738.4-2903
SNR G359.1+0.9
150
-0(cid:176) 40¢
AX J1739.6-2911
100
+1(cid:176) 20¢
AX J1739.3-2924
50
-1(cid:176) 00¢
-0(cid:176) 20¢
+0(cid:176) 40¢
+0(cid:176) 00¢
+0(cid:176) 20¢
+0(cid:176) 40¢
+1(cid:176) 00¢
-1(cid:176) 40¢
40
60
80
100
120
140
160
180
200
220
50
100
150
200
250
Fig. 2.
(Upper) GIS2+GIS3 contour map of G0.9+0.1
with 3 -- 10 keV band. The image is smoothed with a Gaus-
sian filter of σ = 0.75′, and corrected for exposure, vi-
gnetting and the GIS grid structure after subtraction of
non-X-ray background (NXB). Contour level is linearly
spaced. The coordinate is in galactic (lII, bII), and the
north is up. (Lower) The same as Fig. 1 lower panel, but
of G0.9+0.1. The fitting model is an absorbed power-law.
)
s
t
n
u
o
c
(
40
35
30
25
20
15
10
5
0
0
2
4
6
8
10
Radius (arcmin)
of
same
as Fig.
2
the
0.7 -- 3.0
but
The
(Upper) The
of G359.1+0.9 with
positions
upper
Fig. 3.
keV
panel,
detected X-ray
band.
sources
(AX J1738.4−2903, AX J1739.3−2924,
AX J1739.6−2911, AX J1740.3−2904), the radio shell
of G359.1+0.9, and the radio pulsar PSR B1736−29 are
indicated. (Lower) The radial profile with the center at
the peak of AX J1739.6−2911. The profile is fitted with
the point spread function (see text). The best-fit model is
given with the dashed line, whereas the dotted line shows
only the background component in the best-fit model.
6
Sakano et al.: SNRs in the Galactic Center region observed with ASCA
-50
-100
-150
-200
G0.0-1.3
(AXJ1751-29.6)
-250
-300
-350
-400
(b)
-1(cid:176) 00¢
-1(cid:176) 20¢
-1(cid:176) 40¢
-2(cid:176) 00¢
200
180
160
140
120
100
80
60
+0(cid:176) 40¢
(l)
+0(cid:176) 40¢
+0(cid:176) 20¢
+0(cid:176) 00¢
-0(cid:176) 19¢ 59†
-0(cid:176) 40¢
Sgr B2 cloud
+0(cid:176) 20¢
AX J1747.0-2828
+0(cid:176) 20¢
1E1743.1-2843
+0(cid:176) 00¢
-50
0
50
100
150
200
250
300
60
80
100
120
140
160
180
200
Fig. 4.
(Upper) The same as Fig. 2 upper panel, but
of G0.0−1.3 (AX J1751−29.6) with 0.7 -- 3.0 keV band.
(Lower) The same as Fig. 1 lower panel, but of G0.0−1.3.
For the background spectrum, we accumulated the pho-
tons from the elliptical region surrounding the source, ex-
cluding the source region, in the same GIS field of view.
The fitting model is the thin thermal plasma model with
absorption.
Fig. 5.
(Upper) The same as Fig. 2 upper panel, but
of G0.56−0.01 (AX J1747.0−2828) with 6.0 -- 7.0 keV
band, which is dominated by iron Kα line. Note that
the image was corrected only for exposure, NXB was
not subtracted, and the region around a bright source
1E 1743.1−2843 was excluded before the smoothing in
order to reduce the contamination from 1E 1743.1−2843.
The positions of the X-ray reflection nebula Sgr B2 cloud
(e.g., Murakami et al. 1999) and 1E 1743.1−2843 are also
indicated. (Lower) The same as Fig. 1 lower panel, but of
G0.56−0.01. The fitting model is the thin thermal plasma
model and 6.4-keV narrow line, both with interstellar ab-
sorption.
|
0710.4274 | 1 | 0710 | 2007-10-23T14:49:42 | Large-scale Propagation of Very Light Jets in Galaxy Clusters | [
"astro-ph"
] | We performed MHD simulations of very light bipolar jets with density contrasts down to 10^-4 in axisymmetry, which were injected into a medium of constant density and evolved up to 200 kpc (200 r_j) full length. These jets show weak and roundish bow shocks as well as broad cocoons and thermalize their kinetic energy very efficiently. We argue that very light jets are necessary to match low-frequency radio observations of radio lobes as well as the bow shocks seen in X-rays. Due to the slow propagation, the backflows and their turbulent interaction in the midplane are important for a realistic global appearance. | astro-ph | astro-ph |
Extragalactic Jets: Theory and Observation from Radio to Gamma Ray
ASP Conference Series, Vol. ??, 2007
T. A. Rector and D. S. De Young (eds.)
Large-scale Propagation of Very Light Jets in Galaxy
Clusters
V. Gaibler, M. Camenzind
Landessternwarte, ZAH, Konigstuhl 12, 69117 Heidelberg, Germany
M. Krause
Astrophysics Group, Cavendish Laboratory, Madingley Road, Cambridge
CB3 0HE, United Kingdom
Abstract. We performed MHD simulations of very light bipolar jets with
density contrasts down to 10−4 in axisymmetry, which were injected into a
medium of constant density and evolved up to 200 kpc (200 rj) full length.
These jets show weak and roundish bow shocks as well as broad cocoons and
thermalize their kinetic energy very efficiently. We argue that very light jets are
necessary to match low-frequency radio observations of radio lobes as well as
the bow shocks seen in X-rays. Due to the slow propagation, the backflows and
their turbulent interaction in the midplane are important for a realistic global
appearance.
1.
Introduction
During the last years, simulations of extragalactic jets with reasonable resolu-
tion and realistic sizes became computationally feasible, which makes compar-
isons between simulated and observed properties possible (Saxton et al. 2002;
Zanni et al. 2003; Carvalho et al. 2005; Krause 2005; O'Neill et al. 2005). Un-
fortunately, the direct physical variables and the observed properties are rather
hard to link, which leaves simulations with a wide range of parameters. Sim-
ulations are mainly governed by the initial setup of the density ratio between
jet and ambient gas, the Mach number and the magnetic field. If the magnetic
field is not dynamically dominant (though important), the density contrast is
the most dominant parameter, but may be one of the hardest to measure. The
thermal jet pressure has turned out to be of little importance in the very light jet
limit (Krause 2003). As the (kinetic) power of a jet can be estimated from ener-
gies in X-ray bubbles, typical values of velocity, lifetime, jet radius and cluster
gas densities indicate that density contrasts of 10−2 to 10−4 (or even lower) are
necessary to describe real sources. Parameter studies support this further, if the
global jet/cocoon/bow shock properties are compared. Thus, we concentrate on
the very light jets with magnetic fields as another important ingredient.
2. Numerical Method and Setup
We examine the evolution of the jets in axisymmetric simulations using the non-
relativistic MHD code NIRVANA (Ziegler & Yorke 1997) and evolve the mag-
1
2
Gaibler, Camenzind, Krause
netic fields using the constrained transport method, which conserves ∇ · B to
machine roundoff errors.
Simulations of very light hydro and MHD jets were performed with density
contrasts η = ρj/ρa between 10−4 and 10−1 where the jet density is ρj and the
ambient gas has a constant density of ρa. We will focus on the MHD jets, as
their hydro counterparts are only for comparison. The bipolar jet was injected
along the Z axis in cylindrical coordinates with a jet radius of rj = 1 kpc, the
jet speed and the sound speed were fixed at 0.6 c and 0.1 c respectively (which
gives internal Mach number 6). The ambient gas has a density of 0.01 mp/cm3
and a temperature of 5 × 107 K. Fully ionized hydrogen was assumed for both
the jet and the external medium. The MHD simulations have an initial dipolar
field in the whole domain with ∼ 20 µG at the jet boundary and a temporally
constant toroidal field with ∼ 15 µG which is confined to the nozzle. For the
10−3 and especially the 10−4 jet, the magnetic fields thus become dynamically
important and influence the appearance. The simulations were run until they
reach the boundary of the grid which has (4000 × 800) or (4000 × 1600) cells
(depending on the density contrast) and the jet radius is resolved with 20 cells.
3. Morphology
Density and temperature distribution for a η = 10−3 jet is shown in Fig. 1.
The jet backflow blows up a pronounced cocoon, surrounded by a thick shell of
shocked ambient matter. Ambient gas is mixed into the cocoon in finger-like
structures due to Kelvin-Helmholtz instabilities at the contact surface. Near
the jet heads, this instability is suppressed by the magnetic field, which leads to
a smoother appearance there. In purely hydrodynamic simulations, this stabi-
lization is absent. As observations at low radio frequencies show quite smooth
contact discontinuities, this indicates the importance of magnetic fields there.
Figure 1. Density and temperature for a bipolar jet with η = 10−3 after 15
Myrs. The upper panel shows the density in units of 10−28 g/cm3, the lower
one shows the logarithm of temperature in units of 1010 K.
Very Light Jets on Large Scales
3
Figure 2.
(as Fig. 3). The bow shock is located at R ≈ 37 kpc.
Pressure slice through the η = 10−3 jet at Z = 0 after 15 Myr
The cocoon is highly turbulent and vortices hitting the jet beam can easily
destabilize and disrupt it for low jet densities. The Mach numbers quickly
decrease and there is no classical "Mach disk" anymore -- the terminal shock
moves back and forth and isn't well-defined.
Because very light jets only propagate slowly, the backflow is strong and the
turbulence makes the interaction between both jets in the midplane important.
These jets have to be simulated bipolarly to get the lateral expansion and hence
the global appearance right.
If only one jet was simulated, the result would
strongly depend on the boundary condition in the equatorial plane (as shown in
Saxton et al. 2002).
Outside of the contact surface is the shocked ambient gas, which is pushed
outwards by the cocoon pressure. The bow shock for very light jets is different
in its shape and strength from that of heavier jets (see section 5.). It is addition-
ally changed by a density profile in the external medium (Krause 2005), which
increases the aspect ratio with time and shows cylindrical cocoons.
As example, a radial pressure slice at Z = 0 is shown for the 10−3 jet
(Fig. 2). The pressure jump at R ≈ 37 kpc is the bow shock and is pretty weak
compared to bow shocks in heavier jets. Shock speed, pressure and density
jump, consistently with the shock jump conditions, give a Mach number of 1.4.
Observations of bow shocks (e.g. Hercules A in Nulsen et al. 2005), which
are possible with modern X-ray telescopes, show low Mach numbers and low
ellipticity, thus supporting the necessity for very light jet parameters. To find
the right cocoon shapes, for comparison low-frequency radio observations have to
be chosen, because at higher frequencies only a small part of the cocoon is visible
as radio lobes (cooled-down electron population in the backflow is invisible at
these frequencies).
4
Gaibler, Camenzind, Krause
Pressure maps for a η = 10−1 and a 10−3 jet with the same
Figure 3.
lengths. The pressure is shown logarithmically in dyne/cm2. The heavier jet
is still much overpressured with respect to the ambient gas and has a much
more elongated bow shock compared to the elliptically-shaped bow shock for
the lighter jet.
Pressure vs. density histogram for a 10−3 jet after 2 Myr (left
Figure 4.
panel) and 15 Myr (right panel). Cell counts, pressure and density are shown
logarithmically using cgs units.
4. Pressure Evolution
The pressure slice shows many variations in Fig. 2, which is not surprising if
one looks at the turbulent motion and the mixing inside the cocoon in Fig. 1.
Strong pressure waves travel through the cocoon and try to find pressure balance.
This process is much more effective for very light jets due to the much slower
jet head propagation and it leads to a rather spherical expansion of the bow
shock, just like an overpressured bubble. The cocoon of the 10−1 jet in Fig. 3 is
overpressured by a factor of 20 with respect to the ambient gas, while being a
factor of only 1.5 for the 10−3 jet (and 4.9 for this jet at t = 2.5 Myr).
This can also be seen in the pressure -- density diagrams in Fig. 4. The
ambient gas is described by the patch near (−26, −10), the jet nozzle by the
cells around (−29, −10). Adiabatic compression and expansion leads to the
oblique and longish features present at different positions. Top right of the jet
Very Light Jets on Large Scales
5
Figure 5.
Evolution of the forward and sideways bow shock strength for
density contrasts η = 10−1 (dotted) and 10−3 (solid) as a function of the
monotonically increasing axial bow shock radius.
nozzle position are the cocoon grid points, which spread over a large range of
density to the right because of mixing with shocked ambient gas, which is the
elongated feature top right of the ambient gas position. Comparing the two
different simulation snapshots, we find that the pressure distribution is quickly
adjusting towards the external pressure, in agreement to the findings in Krause
(2003).
5. Bow Shock and Cocoon
The quick decrease in cocoon pressure naturally affects the strength of the bow
shock as it is this pressure that drives the shock sideways. Fig. 5 shows the tem-
poral evolution of the bow shock strength, in terms of external Mach numbers,
for the forward (R = 0) direction as well as the sideways (Z = 0) direction for
jets with different density contrasts. For easier comparison with observations,
the axial bow shock radius is used for the abscissa instead of time (but both
increase monotonically).
The bow shocks in forward direction are always stronger than the sideways
shocks due to the direct impact of the jet onto the ambient gas. The lighter jet
has a much weaker bow shock in all directions and the differences between the
two directions shown are much less pronounced. The axial diameter of the bow
shock increases proportionally to t0.68 after a slower growth during the first 2
Myr, the width grows similarly as t0.64. An exponent of 0.6 is expected for the
blast wave expansion with constant jet power (Krause 2003), while the slower
growth rate in the initial phase behaves more like a Sedov blast wave (fixed
initial energy amount, exponent is 0.4).
The cocoons, measured by their full (bipolar) lengths and their full Z-
averaged widths, grow like the bow shock in axial direction (∝ t0.71), but much
slower in width (∝ t0.38). This leads to a continuously increasing bow shock vs.
cocoon width ratio (Fig. 6) with a very thick layer of shocked ambient gas. This
effect is weak for heavier jets, but more proncounced the lighter the jet is.
6
Gaibler, Camenzind, Krause
Figure 6.
Bow shock / cocoon width ratio over time for a 10−3 jet.
Figure 7. Aspect ratios over full jet length. Dotted line: 10−1 jet, solid:
10−3 jet. In each case, the lower lines refer to the bow shock and the upper
ones to the cocoon.
The aspect ratios (length/width) for the 10−1 and 10−3 jets are plotted
in Fig. 7. The bow shock for the lighter jet starts with a roughly spherical
shape and only slightly increases its aspect ratio (length/width) to a constant
value of 1.4 (Fig. 7). The heavier jet behaves similarly but approaches a much
higher aspect ratio of 2.6. Thus the aspect ratio of the bow shock may be a
good property to compare with observations. The aspect ratio of the cocoons in
contrast continues to increase, with the heavier jet being on much higher values
at all times.
6. Thermalization
From the quick adjustment of pressure towards an average value, one might
expect a strong conversion of the (kinetic) jet power to thermal energy. This,
in fact, is measured for our simulations. Already for the heavy 10−1 jet, on
Very Light Jets on Large Scales
7
average 67 % of the total energy input is measured as thermal energy increase
and only 33 % as kinetic energy increase. For the 10−2 jet this is 72 % vs.
27 %, and for the 10−3 jet already 81 % of the jet power appear as thermal
energy increase with 16 % going into kinetic energy. Here, already 2 % go into
an increase in magnetic energy, because with lower density the magnetic fields of
constant values become more and more important. The 10−4 simulation showed
a thermalization efficiency of 94 %, but the fractions for kinetic and magnetic
energy are now governed by the strong magnetic pressure and will be examined
in the future.
The overall trend to very efficient thermalization for low-density jets nicely
suits the increasingly spherical bow shock shape due to (isotropic) cocoon pres-
sure. It also provides the cluster with a huge amount of thermal energy and
high-entropy plasma, which may be relevant for the problem of cluster heating
and cooling flows (eg. Magliocchetti & Bruggen 2007).
Acknowledgments. This work was also supported by the Deutsche For-
schungsgemeinschaft (Sonderforschungsbereich 439).
References
Carvalho, J. C., Daly, R. A., Mory, M. P., & O'Dea, C. P. 2005, ApJ, 620, 126
Krause, M. 2003, A&A, 398, 113
Krause, M. 2005, A&A, 431, 45
Magliocchetti, M., & Bruggen, M. 2007, MNRAS, 528
Nulsen, P. E. J., Hambrick, D. C., McNamara, B. R., Rafferty, D., Birzan, L., Wise, M.
W., & David, L. P. 2005, ApJ, 625, L9
O'Neill, S. M., Tregillis, I. L., Jones, T. W., & Ryu, D. 2005, ApJ, 633, 717
Saxton, C. J., Sutherland, R. S., Bicknell, G. V., Blanchet, G. F., & Wagner, S. J. 2002,
A&A, 393, 765
Zanni, C., Bodo, G., Rossi, P., Massaglia, S., Durbala, A., & Ferrari, A. 2003, A&A,
402, 949
Ziegler, U., & Yorke, H. W. 1997, Computer Physics Communications, 101, 54
|
astro-ph/0501096 | 1 | 0501 | 2005-01-06T15:43:08 | Galactic Positrons From Localized Sources | [
"astro-ph"
] | The anomalous bump in the cosmic ray positron to electron ratio at $10 GeV$ can be explained as being a component from a point source that was originally harder than the primary electron background and degrades due to synchrotron and inverse Compton losses in the Galaxy while propagating to the Earth's vicinity. The fit is better than can be obtained with homogeneous injection and is attributed to a minimum age threshold. Annihilating neutralinos can provide a fair fit to the data if they have a mass just above 1/2 the mass of the $Z^o$ and if they annihilate primarily in distant density concentrations in the Galaxy. A possible observational consequence of this scenario would be intense inverse Comptonization of starlight at the Galactic center, with a sharp energy cutoff in the emergent photons as a possible signature of the neutralino mass. | astro-ph | astro-ph |
Galactic Positrons From Localized Sources
David Eichler1, Irit Maor2
ABSTRACT
The anomalous bump in the cosmic ray positron to electron ratio at 10 GeV
can be explained as being a component from a point source that was originally
harder than the primary electron background and degrades due to synchrotron
and inverse Compton losses in the Galaxy while propagating to the Earth's vicin-
ity. The fit is better than can be obtained with homogeneous injection and is
attributed to a minimum age threshold. Annihilating neutralinos can provide
a fair fit to the data if they have a mass just above 1/2 the mass of the Z o
and if they annihilate primarily in distant density concentrations in the Galaxy.
A possible observational consequence of this scenario would be intense inverse
Comptonization of starlight at the Galactic center, with a sharp energy cutoff in
the emergent photons as a possible signature of the neutralino mass.
Subject headings: cosmology: dark matter -- diffusion -- elementary particles --
galaxy: center
1.
Introduction
The possibility that weakly interacting dark matter particles (WIMP's) could annihilate
into detectable cosmic radiation was suggested by Silk & Srednicki (1984). Tylka & Eichler
(1987) noted a reported positron excess, curiously localized near 10 GeV [Mueller & Tang
(1985, 1987); Barwick et al.
(1997, 1998)] and considered whether it could be due to the
annihilation of photinos (as a simple example of neutralinos) in the tens of GeV mass range.
The difficulty was that this process, given the laboratory constraints on the neutralinos,
seemed to fall short of providing enough positrons, and the results were not published.
Various papers on this excess eventually appeared [Tylka (1989); Eichler (1989); Turner &
Wilczek (1990); Coutu et al. (1999)], and some noted that the potential for positron excess
1Physics Department, Ben-Gurion University, Beer-Sheva, Israel, [email protected]
2Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge CB3 0WA, UK,
[email protected]
-- 2 --
could be bolstered by clumpiness in the annihilating dark matter or by decay of weakly
unstable dark matter particles.
The approach usually found in present literature is to try and fit the overall e+/(e+ +e−)
ratio, without giving special attention to the curious behaviour at 10 GeV , see Hooper &
Silk (2004). Baltz & Edsjo (2001) considered a whole class of minimal standard supersym-
metric models and failed to get any non-monotonicity in the e+/(e+ + e−) ratio. Eichler &
Maor (2004) (henceforth Paper I) considered annihilation of particles through the Z o-channel
(which we shall henceforth refer to as virtual Z o-decay) noted that non-relativistic virtual
Z o decay (i.e. when the rest mass of the annihilating dark matter particle is slightly above
1/2 the Z o mass) provide a remarkably good fit to the observed e+/(e+ + e−) ratio below
10 GeV mainly due to the positrons that emerge from decaying muons. At higher energies,
however, the predicted e+/(e+ +e−) ratio rose above the observed values within conventional
assumptions about the injection and propagation. In particular, it was assumed in Paper I
that the positrons and primary electrons are each injected with the same spatial profile, and
that their propagation in the Galaxy is identical. The reason for this rise is that some Z o's
decay directly into high energy e+e− pairs so that the e+ energy is half the Z o mass, and
this gives rise to a high energy bump in the e+/(e+ + e−) ratio at about 50 GeV . While this
bump can be partially washed out by losses and escape, it was found that the high energy
e+/(e+ + e−) ratio is nevertheless apparently too high to fit the observations to within 1σ
error bars. As discussed in Eichler (1989) this is a generic problem for any positron source
that is significantly harder than the primary electrons above 10 GeV .
However, the dark matter annihilation scenario for explaining the positron excess in
any case requires that clumping of the dark matter, and a likely place for this is near the
Galactic center. This means that the positrons in our neighborhood that are dark matter
annihilation products would have a minimum age, i.e. the time needed to diffuse from the
source to our neighborhood, and the age distribution of the positrons that make it to the
Earth's vicinity contain fewer young positrons than the age distribution that one associates
with the standard leaky box model. In this letter we consider that the positrons are injected
by an effectively point source at a finite distance, and show that it greatly improves the
fit over that obtained in Paper I. The positron bump at ∼ 10 GeV can be attributed to
halo-type age ∼ 3× 107 yr for the positrons, for over such a lifetime, positrons losing energy
by synchrotron and inverse Comptonization would end up at about this energy.
We will find that obtaining a good fit from a single source with a single diffusion coeffi-
cient is difficult. However, it is well known that below a certain intensity, e.g. far enough up
front in a diffusion front, cosmic rays can freely stream. Such behavior is observed upstream
of the Earth's bow shock. Theoretical reasons for such free streaming include the difficulty
-- 3 --
of resonantly scattering cosmic rays through 90 degree pitch angle in the linear wave am-
plitude regime. Also, a larger, counterstreaming , finitely stable component of cosmic rays
would stabilize the smaller free streaming positron component. We therefore allow for the
possibility that a small fraction of the positrons freely stream, and arrive at the Earth's
vicinity much younger than the rest. We find that this improves the fit still further. We
find that the low energy non-monotonicity, which appears in the injection spectrum from
non-relativistic Z o decay Paper I, can be produced also by using a simple power law for
the injected spectrum (using the same propagation model). It can thus be produced by the
combination of a harder (but monotonic) spectrum of injected positrons and propagation
effects. We conclude that it is still too early to unambiguously interpret the low energy
behavior of the spectrum as a signature of self annihilating dark matter.
2. Equations and Results
The steady state diffused equation for the particle number density, n(x, r), is
∂n
∂t
= 0 = Dn − Rn +
1
mZ
∂
∂x (cid:18)mZ
dx
dt
n(cid:19) + I(x)δ (r/L)
(1)
dx
x = E/mZ, D is a diffusion operator, R = Bx0.5 is the escape rate with B ∼ f ew×10−151/s,
dt = Ax2 with A = 8.5 × 10−16 erg/s is the Compton loss rate, corresponding to an
mZ
electromagnetic energy density in the Galaxy of 10−12 erg/cm3, L is a distance scale, and
I(x) is the spectrum injected by a point source.
We assume as in Paper I that the primary electrons and background positrons are
injected homogeneously, Dnb = 0:
nb(x) =
mZ
Ax2 exp(cid:20)−
2mZB
A√x (cid:21)Z ∞
x
Ib(x′)exp(cid:20) 2mZB
A√x′ (cid:21) dx′
Ib,e−(x) = Cx−2
Ib,e+(x) = Dx−2.8
f or background e−
f or background e+
(2)
(3)
For the Z o decay injected spectrum we take the diffusion to be one-dimensional with
a diffusion coefficient D, D = D ∂2
∂rr=L = 0
(conserving the number of particles except for the escape term). With these boundary
conditions, the solution to eq. (1) is
∂x2 , and with boundary conditions such that ∂n
nZ(x, r) =
mZ
Ax2 exp(cid:20)−
2mZB
A√x (cid:21) ×
-- 4 --
Z ∞
x
IZ(x′)exp(cid:20) 2mZB
A√x′ (cid:21) ∞
X−∞
cosh πnr
L i exp(cid:20)−(cid:16)πn
L (cid:17)2 mZD
A (cid:18) 1
x −
1
x′(cid:19)(cid:21) dx′(4)
While D (the diffusion coefficient) and L (the size of the leaky box) are free parameters, we
took r = 8 Kpc, the distance to the galactic centre. K ≡ D/r2 gives the inverse time for
diffusion.
We have chosen a one dimensional diffusion because it gives somewhat better results than
3 dimensional diffusion. This is physically plausible if one considers magnetic fields which
will confine the movement of the charged particles. So the geometry is tube-like, with an
effective cross section such that the total volume is the galactic volume, (20 kpc)3.
IZ(x) is the Z decay products,
IZ(x) = N (cid:16) 0.0344Ie(x) + 0.0344Iµ(x) + 0.0069Iτ (x) + 0.6916Ih(x)(cid:17)
Ie(x) = δ(cid:18)x −
2(cid:19)
1
(5)
Iµ(x) =
Iτ (x) =
+
Ih(x) =
2
2
3 (cid:16) 5 − 36x2 + 32x3 (cid:17)
3 (cid:16) 5 − 36x2 + 32x3 (cid:17)
9 (cid:20)−
9 Z 1
95
3 − 108x2 +
10ak −bk ¯x
d¯x
¯x
14
2
28
9 x
1408
3
x3 −(cid:0)25 + 324x2 + 128x3(cid:1) ln (2x)(cid:21)
a1 = 3, b1 = 10
0 < ¯x < 0.1
a2 = 2, b2 = 4
0.1 < ¯x < 1
Each Ich describes the ch channel of decay, and the pre-factors correspond to the branch-
ing ratios. The calculation was done in zeroth order, assuming 3 massless families and
neglecting the top quark, for details see Paper I. Following the discussion there, we take
N = 1.3 × 10−29 1/(cm3 s) as the annihilation rate per unit volume.
Fig.
(1) shows a fit with a single point source of Z o decay and a single diffusion
coefficient. The good fit to the low energies from Paper I is still present, but at the price
that the excess in energies toward x = 1/2 is now is suppressed by the finite age effect. As
the figure shows, we are now facing a scenario which is opposite to Paper I; the finite age
effect tends to suppress the high energy excess at the price of killing it off altogether.
However, there are several possibilities that avoid this problem: There may be more than
-- 5 --
one source, and there may be more than one route (roughly guided by magnetic field lines)
by which the particles diffuse or freely stream from the source to our vicinity. High energy
particles diffuse much less than low energy ones because they are fewer in number and create
less waves. So their self-generated scattering is less efficient. Thus, the fraction of free
streaming particles should be higher at higher energy. Fig. (2) shows a combination of two
Z o decay components, an older, larger one that arrives via diffusion, and a younger, smaller
component that has managed more free streaming. This figure illustrates that if one takes an
age distribution into account, the flexibility in adjusting the high energy spectrum becomes
much larger, and can be fitted to the data.
For sake of comparison, we also include a power law injected spectrum, fig. (3) shows
various power laws, and fig. (4) shows two components with different ages. We find that as
long as the injected power law is hard enough, one can produce a low energy (5−10 GeV ) dip.
The quality of the fit is almost as good for a power law as for virtual Z o decay. We consider
the low energy dip to have qualitatively more significance than the higher points and have
emphasized those data points accordingly in choosing the best fit. We have deliberately not
quantified this with the standard statistical measures. Trying to get the statistically best
parameters (for either power law or Z o decay as injected spectrums) would wash out the low
energy behavior that we are focusing on.
Although we can reproduce the 7 GeV dip, the peak at E ∼ 15 GeV is still too big for
the HEAT data (though too small for the earlier data). This seems to be a generic feature
of our results, regardless of whether the injection source is virtual Z o decay or a power law.
The problem would be worse if the virtual Z o had an energy well above mZ.
3. Possible Observational Consequences
The hypothesis that the neutralino mass mχ is only slightly more than half the Z o mass
is motivated by several factors: The annihilation cross section can be resonantly enhanced
by a factor of 2 or 3 more than that during annihilation in the early universe, when the ther-
mal broadening of the Z resonance somewhat exceeded its natural width, Greist and Seckel
(1991). Moreover, assuming the smallest allowable mass allows the greatest annihilation rate
since the annihilation reaction rate is fixed by the condition that it allows a given cosmic
dark matter contribution. (Although dark matter clumping can enhance the annihilation
rate, a plausible level of such enhancement is limited by observational constraints on dark
matter clumping that are set by stellar distributions in galactic centers.) Making mχ just
above mZ causes the Z o resonance to be asymmetric, but this would be hard to measure
-- 6 --
experimentally because of the weak coupling of the emerging neutralinos at CM collision
energies above 2mχ. On the other hand, that the annihilation cross the virtual Z o must be
close to its mass shell if it is to provide a decent fit suggests that its loop corrections would
be large and it might be discernable or falsifiable with particle collider data on processes
that depend on such loop corrections.
In an astronomical context, a possible observational consequence of a point source of
positrons at the Galactic center could be inverse Comptonization of starlight, which is far
more intense than at a typical point in the Galaxy. The profile of Galactic starlight near
the Galactic center is given by Kent
(1992). The derived photon energy density is then
U(r) = 4.3 × 10−9(r/pc)−0.85 erg/s. Assuming the positrons are produced within a typical
radius r of the Galactic Center, they produce a minimum of EIC(γ) ≡ R 100pc
γ 2σT U(r)dr
in inverse Compton (IC) scattered starlight before escaping the central 100 pc region, and
the luminosity L(≥ γmec2) above γ 2ǫph, where ǫph is the typical energy of the pre-scattered
starlight photons, is R ∞
γ I(γ ′mec2)EIC(γ ′)dγ ′. The most energetic e+e− pairs alone, that is
those which result directly from the Z o decay (Ie(x) in eq. 6), will produce IC luminosity of
3.8 × 1035 erg/s.
r
Fig.
(5) shows the logarithmic derivative of the IC luminosity due to the positrons
only, −γ 2Lγ2 = −dL/d ln(γ 2), as a function of the square of the positron Lorentz factor,
γ 2. Shown in the figure is the minimum IC luminosity as a function of the frequency scaled
to the frequency of the pre-scattered photon (x axis). The minimum luminosity assumes
that the positrons emerge from the central region in a straight line. If the mean free path
λ is less than 100 pc, then the predicted luminosity goes up by roughly a factor of (100pc/λ).
For dark matter annihilation that yields direct monochromatic e+e− pairs at Lorentz fac-
tor γo, this would translate into a sharp cutoff in the IC gamma rays of γ 2
o eV .
In our particular example mχ ≃ mZ/2, this would lead to a cutoff at 1010ǫph ∼ 30 GeV . This
would in principle be detectable by MAGIC, see for example Cortina (2004), if the location
were suitable for observing the Galactic center. Alternatively, it could be detected by HESS
if the energy threshold could be pushed to below 30 GeV . This scenario would not explain
the T eV photons from the Galactic center recently reported by the HESS collaboration,
Aharonian et al. (2004). If, however, there is annihilation in the Galactic Center of heavier
dark matter particles, then direct e+e− pairs might be detectable via such a cutoff in the
T eV gamma ray spectrum at mχ/2.
o ǫph or about 3γ 2
In conclusion, we find that the cosmic ray positron data can be fit with more than one
-- 7 --
hard source of positrons provided that a) they have a chance to lose energy before escaping
the Galaxy and b) they have a minimum age (e.g. they come from discrete, distant sources),
unlike the background primary electrons. They need not be from dark matter annihilation,
but a best case scenario for this is not confidently ruled out by existing data. Detection of
inverse Compton radiation with good energy resolution can in principle provide information
as to the spectrum of the positrons and their point of origin.
We thank T. Alexander, E. Baltz, and A. Dekel for helpful discussions. DE gratefully
acknowledges the support from the Israel-U.S. Binational Science Foundation, the Israel
Science Foundation and the Arnow Chair of Physics at Ben Gurion University. IM gratefully
acknowledges the support from the Leverhulme Trust.
REFERENCES
Aharonian, F. et al. [The HESS Collaboration], arXiv:astro-ph/0408145.
Baltz, E. & Edsjo, J. astro-ph/0109318 (2001)
Barwick, S. W. et al. (HEAT collaboration) Astroph. J. 482, L191 (1997)
Barwick, S. W. et al. (HEAT collaboration) Astroph. J. 498, 779 (1998)
Cortina, J. [the MAGIC Collaboration], arXiv:astro-ph/0407475.
Coutu. S. et al. (HEAT collaboration) Astropart. Phys. 11, 429 (1999)
Eichler, D. Phys. Rev. Lett. 63, 2440 (1989)
Eichler, D & Maor, I. Astropart. Phys. 21 (2004) 195.
Greist, K. and Seckel, D. Phys. Rev. D 43, 3191
Hooper, D. & Silk, J. arXiv:hep-ph/0409104.
Kent, S. M. Astroph. J. 387, 181 (1992)
Mueller, D. & Tang, J. Proc. of Nineteenth International Cosmic Ray Conference, La Jolla,
California, NASA Conf. Pub. No. 2376 (U.S. GPO, Washington, DC), 2, 378 (1985)
Mueller, D. & Tang, J. Astroph. J, 312, 183 (1987)
Silk, J. & Srednicki, M. Phys. Rev. Lett., 53, 624 (1984)
-- 8 --
Turner, M. S. & Wilczek, F. Phys. Rev. D 42, 1001, (1990)
Tylka, A. J. & Eichler, D. University of Maryland preprint (1987)
Tylka, A. J. Phys. Rev. Lett. 63, 40 (1989)
This preprint was prepared with the AAS LATEX macros v5.2.
-- 9 --
0.1
0.2
E/mZ
0.3
0.4
0.5
0.25
0.2
0.15
0.1
0.05
0
0
)
−
e
+
+
e
(
/
+
e
Fig. 1. -- The e+/(e+ + e−) as a function of x = E/mZ, for a single source Z decay injected
spectrum. A = 8.5×10−16 erg
cm3 s ,
and K = 1.9× 10−16 1
s . Data taken from Barwick et al. (1997) (black) and Mueller & Tang
(1987) (grey).
cm3 s , D = 1.3×10−31
s , B = 7.1×10−15 1
s , C = 4.0×10−29
1
1
)
−
e
+
+
e
(
/
+
e
0.25
0.2
0.15
0.1
0.05
0
0
both
K1
K2
0.1
0.2
E/mZ
0.3
0.4
0.5
Fig. 2. -- The e+/(e+ + e−) as a function of x = E/mZ, for a combination of 2 sources of
Z decay injected spectrum. A = 8.5 × 10−16 erg
s , C = 4.9 × 10−29
cm3 s ,
D = 1.3 × 10−31
s . The ratio between the
two components is 1 : 5. Data taken from Barwick et al.
(1997) (black) and Mueller &
Tang (1987) (grey).
s , B = 7.6 × 10−15 1
s and K2 = 2.8 × 10−16 1
cm3 s , K1 = 2.8 × 10−14 1
1
1
-- 10 --
w=−0.2
w=−0.3
w=−0.4
w=−0.5
0.1
0.2
E/mZ
0.3
0.4
0.5
0.25
0.2
0.15
0.1
0.05
0
0
)
−
e
+
+
e
(
/
+
e
Fig. 3. -- The e+/(e+ + e−) as a function of x = E/mZ, for various power laws, Nxw, as
the injected spectrum. A = 8.5 × 10−16 erg
cm3 s ,
D = 1.1 × 10−31
(1997)
(black) and Mueller & Tang (1987) (grey).
cm3 s , and K = 6.6 × 10−17 1
s . Data taken from Barwick et al.
s , B = 4.4 × 10−15 1
s , C = 1.7 × 10−29
1
1
)
−
e
+
+
e
(
/
+
e
0.25
0.2
0.15
0.1
0.05
0
0
both
K1
K2
0.1
0.2
E/mZ
0.3
0.4
0.5
Fig. 4. -- The e+/(e+ + e−) as a function of x = E/mZ, for a combination of 2 sources
of power law (w = −0.3) injected spectrum. A = 8.5 × 10−16 erg
s , B = 4.4 × 10−15 1
s ,
s and K2 = 1.2 × 10−16 1
C = 1.7 × 10−29
s .
The ratio between the two components is 10 : 1. Data taken from Barwick et al.
(1997)
(black) and Mueller & Tang (1987) (grey).
cm3 s, K1 = 7.2 × 10−17 1
cm3 s, D = 1.1 × 10−31
1
1
-- 11 --
x 1036
3
2.5
2
1.5
1
0.5
0
108
109
γ2
/
)
s
g
r
e
(
)
2
γ
l
(
n
d
L
d
/
Fig. 5. -- Differential IC luminosity due to the positrons only, −dL/d ln(γ 2) × (20 kpc)3/V
as a function of γ 2.
|
astro-ph/9610078 | 2 | 9610 | 1996-10-24T22:28:22 | The Best Theory of Cosmic Structure Formation is Cold + Hot Dark Matter (CHDM) | [
"astro-ph"
] | I have been asked to make the case here that CHDM is the best theory of cosmic structure formation, and indeed I believe that it is the best of all those I have considered if the cosmological matter density is near critical and if the expansion rate is not too large (i.e. $h \equiv H_0/(100 \kmsMpc) \lsim 0.6$). I discuss CHDM together with its chief competitor among CDM variants, low-$\Omega_0$ CDM with a cosmological constant ($\Lambda$CDM). $\Lambda$CDM with $\Omega_0 \sim 0.3$ has the possible virtue of allowing a higher expansion rate H_0 for a given cosmic age t_0, but the defect of predicting too much fluctuation power on small scales. Also, except for the $H_0-t_0$ problem, there is not a shred of evidence in favor of a nonzero cosmological constant, only increasingly stringent observational upper bounds on it. CHDM has less power on small scales, in good agreement with data, although it remains to be seen whether it predicts early enough galaxy formation to be compatible with the latest high-redshift data. I also briefly compare CHDM to other CDM variants such as Warm Dark Matter (WDM) and tilted CDM. CHDM has the advantage among $\Omega=1$ CDM-type models of requiring little or no tilt, which appears to be an advantage in fitting recent small-angle CMB anisotropy data. The presence of a hot component that clusters less than cold dark matter lowers the effective $\Omega_0$ that would be measured on small scales, which appears to be in accord with observations, and it may also avoid the discrepancy between the high central density of dark matter halos from CDM simulations compared to evidence from rotation curves of dwarf spiral galaxies. | astro-ph | astro-ph |
The Best Theory of Cosmic Structure Formation
is Cold + Hot Dark Matter (CHDM) ∗
Joel R. Primack
University of California, Santa Cruz, CA 95064
[email protected]
January 18, 2018
1 Summary
The fact that the simplest modern cosmological theory, standard Cold Dark
Matter (sCDM), almost fits all available data has encouraged the search for
variants of CDM that can do better. I have been asked to make the case here
that CHDM is the best theory of cosmic structure formation, and indeed I
believe that it is the best of all those I have considered if the cosmological
matter density is near critical (i.e., Ω0 ≈ 1) and if the expansion rate is not too
large (i.e. h ≡ H0/(100 km s−1 Mpc−1) <∼ 0.6). But I think it will be helpful to
discuss CHDM together with its chief competitor among CDM variants, low-Ω0
CDM with a cosmological constant (ΛCDM). While the predictions of COBE-
normalized CHDM and ΛCDM both agree reasonably well with the available
data on scales of ∼ 10 to 100 h−1 Mpc, each has potential virtues and defects.
ΛCDM with Ω0 ∼ 0.3 has the possible virtue of allowing a higher expansion rate
H0 for a given cosmic age t0, but the defect of predicting too much fluctuation
power on small scales. CHDM has less power on small scales, so its predictions
appear to be in good agreement with data on the galaxy distribution, although
it remains to be seen whether it predicts early enough galaxy formation to be
compatible with the latest high-redshift data. Also, several sorts of data suggest
that neutrinos have nonzero mass, and the variant of CHDM favored by this data
-- in which the neutrino mass is shared between two species of neutrinos -- also
seems more compatible with the large-scale structure data. Except for the H0 −
t0 problem, there is not a shred of evidence in favor of a nonzero cosmological
constant, only increasingly stringent upper bounds on it from several sorts of
measurements. Two recent observational results particularly favor high cosmic
∗To appear in Critical Dialogues in Cosmology, ed. N. Turok (World Scientific, 1996).
1
density, and thus favor Ω = 1 models such as CHDM over ΛCDM -- (1) the
positive deceleration parameter q0 > 0 measured using high-redshift Type Ia
supernovae, and (2) the low primordial deuterium/hydrogen ratio measured
in two different quasar absorption spectra.
If confirmed, (1) means that the
cosmological constant probably cannot be large enough to help significantly with
the H0 − t0 problem; while (2) suggests that the baryonic cosmological density
is at the upper end of the range allowed by Big Bang Nucleosynthesis, perhaps
high enough to convert the "cluster baryon crisis" for Ω = 1 models into a crisis
for low-Ω0 models. I also briefly compare CHDM to other CDM variants such as
Warm Dark Matter (WDM) and tilted CDM. CHDM has the advantage among
Ω = 1 CDM-type models of requiring little or no tilt, which appears to be an
advantage in fitting recent small-angle cosmic microwave background anisotropy
data. The presence of a hot component that clusters less than cold dark matter
lowers the effective Ω0 that would be measured on small scales, which appears to
be in accord with observations, and it may also avoid the discrepancy between
the high central density of dark matter halos from CDM simulations compared
to evidence from rotation curves of dwarf spiral galaxies.
2
Introduction
"Standard" Ω = 1 Cold Dark Matter (sCDM) with h ≈ 0.5 and a near-
Zel'dovich spectrum of primordial fluctuations [1] until a few years ago seemed
to many theorists to be the most attractive of all modern cosmological mod-
els. But although sCDM normalized to COBE nicely fits the amplitude of the
large-scale flows of galaxies measured with galaxy peculiar velocity data [2], it
does not fit the data on smaller scales:
it predicts far too many clusters [3]
and does not account for their large-scale correlations [4], and the shape of the
power spectrum P (k) is wrong [5, 6]. Here I discuss what are perhaps the two
most popular variants of sCDM that might agree with all the data: CHDM and
ΛCDM. The linear matter power spectra for these two models are compared in
Figure 1 (from Ref.
[7]) with the real-space galaxy power spectrum obtained
from the two-dimensional APM galaxy power spectrum [5]. The ΛCDM and
CHDM models essentially bracket the range of power spectra in currently pop-
ular cosmological models which are variants of CDM.
CHDM cosmological models have Ω = 1 mostly in cold dark matter but
with a small admixture of hot dark matter, light neutrinos contributing Ων =
mν,tot/(92h2eV) ≈ 0.2, corresponding to a total neutrino mass of mν,tot ≈ 5 eV
for h = 0.5. CHDM models are a good fit to much observational data [8, 9]
-- for example, correlations of galaxies and clusters and direct measurements
of the power spectrum P (k), velocities on small and large scales, and other
statistics such as the Void Probability Function (probability P0(r) of finding
no bright galaxy in a randomly placed sphere of radius r). My colleagues and
I had earlier shown that CHDM with Ων = 0.3 predicts a VPF larger than
2
Figure 1: Power spectrum of dark matter for ΛCDM and CHDM models consid-
ered in this paper, both normalized to COBE, compared to the APM galaxy real-
space power spectrum. (ΛCDM: Ω0 = 0.3, ΩΛ = 0.7, h = 0.7, thus t0 = 13.4
Gy; CHDM: Ω = 1, Ων = 0.2 in Nν = 2 ν species, h = 0.5, thus t0 = 13 Gy;
both models fit cluster abundance with no tilt, i.e. np = 1. From Ref. [7].)
observations indicate [10], but new results based on our Ων = 0.2 simulations
in which the neutrino mass is shared equally between two neutrino species [8]
show that the VPF for this model is in excellent agreement with observations.
However, our simulations [12] of COBE-normalized ΛCDM with h = 0.7 and
Ω0 = 0.3 lead to a VPF which is too large to be compatible with the data [11].
Moreover, there is mounting astrophysical and laboratory data suggesting
that neutrinos have non-zero mass [8, 13]. The analysis of the LSND data
through 1995 [14] strengthens the earlier LSND signal for ¯νµ → ¯νe oscillations.
Comparison with exclusion plots from other experiments implies a lower limit
µe ≡ m(νµ)2 − m(νe)2 >∼ 0.2 eV2, implying in turn a lower limit mν >∼
∆m2
0.45 eV, or Ων >∼ 0.02(0.5/h)2. This implies that the contribution of hot dark
matter to the cosmological density is larger than that of all the visible stars
(Ω∗ ≈ 0.004 [15]). More data and analysis are needed from LSND's νµ →
νe channel before the initial hint [16] that ∆m2
µe ≈ 6 eV2 can be confirmed.
Fortunately the KARMEN experiment has just added shielding to decrease its
3
background so that it can probe the same region of ∆m2
µe and mixing angle,
with sensitivity as great as LSND's within about two years. The Kamiokande
data [17] showing that the deficit of E > 1.3 GeV atmospheric muon neutrinos
increases with zenith angle suggests that νµ → ντ oscillations [18] occur with
an oscillation length comparable to the height of the atmosphere, implying that
∆m2
τ µ ∼ 10−2 eV2 [17] -- which in turn implies that if either νµ or ντ have
large enough mass (>∼ 1 eV) to be a hot dark matter particle, then they must
be nearly degenerate in mass, i.e. the hot dark matter mass is shared between
these two neutrino species. The much larger Super-Kamiokande detector is
now operating, and we should know by about the end of 1996 whether the
Kamiokande atmospheric neutrino data that suggested νµ → ντ oscillations will
be confirmed and extended [19]. Starting in 1997 there will be a long-baseline
neutrino oscillation disappearance experiment to look for νµ → ντ with a beam
of νµ from the KEK accelerator directed at the Super-Kamiokande detector,
with more powerful Fermilab-Soudan, KEK-Super-Kamiokande, and possibly
CERN-Gran Sasso long-baseline experiments in subsequent years.
Evidence for non-zero neutrino mass evidently favors CHDM, but it also dis-
favors low-Ω models. Because free streaming of the neutrinos damps small-scale
fluctuations, even a little hot dark matter causes reduced fluctuation power on
small scales and requires substantial cold dark matter to compensate; thus evi-
dence for even 2 eV of neutrino mass favors large Ω and would be incompatible
with a cold dark matter density Ωc as small as 0.3 [8]. Allowing Ων and the tilt
to vary, CHDM can fit observations over a somewhat wider range of values of the
Hubble parameter h than standard or tilted CDM [20]. This is especially true if
the neutrino mass is shared between two or three neutrino species [8, 21, 22, 23],
since then the lower neutrino mass results in a larger free-streaming scale over
which the power is lowered compared to CDM; the result is that the cluster
abundance predicted with Ων ≈ 0.2 and h ≈ 0.5 and COBE normalization (cor-
responding to σ8 ≈ 0.7) is in reasonable agreement with observations without
the need to tilt the model[24] and thereby reduce the small-scale power further.
(In CHDM with a given Ων shared between Nν = 2 or 3 neutrino species, the
linear power spectra are identical on large and small scales to the Nν = 1 case;
the only difference is on the cluster scale, where the power is reduced by ∼ 20%
[21, 8].)
Another consequence of the reduced power on small scales is that structure
formation is more recent in CHDM than in ΛCDM. This may conflict with obser-
vations of damped Lyman α systems in quasar spectra, and other observations
of protogalaxies at high redshift, although the available evidence does not yet
permit a clear decision on this (see below). While the original Ων = 0.3 CHDM
model [25, 26] certainly predicts far less neutral hydrogen in damped Lyman
α systems (identified as protogalaxies with circular velocities Vc ≥ 50 km s−1)
than is observed [27, 28], lowering the hot fraction to Ων ≈ 0.2 dramatically
improves this [28, 29]. Also, the evidence from preliminary data of a fall-off of
4
the amount of neutral hydrogen in damped Lyman α systems for z >∼ 3 [30] is
in accord with predictions of CHDM [28].
However, as for all Ω = 1 models, h >∼ 0.55 implies t0 <∼ 12 Gyr, which con-
flicts with age estimates from globular cluster [31] and white dwarf cooling [32].
The only way to accommodate both large h and large t0 within the standard
FRW framework of General Relativity is to introduce a positive cosmological
constant (Λ > 0) [33, 34].
ΛCDM flat cosmological models with Ω0 = 1 − ΩΛ ≈ 0.3, where ΩΛ ≡
Λ/(3H 2
0 ), were discussed as an alternative to Ω = 1 CDM since the beginning
of CDM [1, 35]. They have been advocated more recently [36] both because
they can solve the H0 − t0 problem and because they predict a larger fraction
of baryons in galaxy clusters than Ω = 1 models. Early galaxy formation
also is often considered to be a desirable feature of these models. But early
galaxy formation implies that fluctuations on scales of a few Mpc spent more
time in the nonlinear regime, as compared with CHDM models. As has been
known for a long time, this results in excessive clustering on small scales. My
colleagues and I have found that a typical ΛCDM model with h = 0.7 and Ω0 =
0.3, normalized to COBE on large scales (this fixes σ8 ≈ 1.1 for this model),
is compatible with the number-density of galaxy clusters[24], but predicts a
power spectrum of galaxy clustering in real space that is much too high for
wavenumbers k = (0.4 − 1)h/Mpc [12]. This conclusion holds if we assume
either that galaxies trace the dark matter, or just that a region with higher
density produces more galaxies than a region with lower density. One can see
immediately from Figure 1 that there will be a problem with this ΛCDM model,
since the APM power spectrum is approximately equal to the linear power
spectrum at wavenumber k ≈ 0.6h Mpc−1, so there is no room for the extra
power that nonlinear evolution certainly produces on this scale (see Figure 1 of
Ref. [12] and further discussion below). The only way to reconcile the model
with the observed power spectrum is to assume that some mechanism causes
strong anti-biasing -- i.e., that regions with high dark matter density produce
fewer galaxies than regions with low density. While theoretically possible, this
seems very unlikely; biasing rather than anti-biasing is expected, especially on
small scales [37]. Numerical hydro+N-body simulations that incorporate effects
of UV radiation, star formation, and supernovae explosions [44] do not show
any antibias of luminous matter relative to the dark matter.
Our motivation to investigate this particular ΛCDM model was to have H0
as large as might possibly be allowed in the ΛCDM class of models, which in
turn forces Ω0 to be rather small in order to have t0 >∼ 13 Gyr. There is little
room to lower the normalization of this ΛCDM model by tilting the primordial
power spectrum Pp(k) = Aknp (i.e., assuming np significantly smaller than the
"Zel'dovich" value np = 1), since then the fit to data on intermediate scales will
be unacceptable -- e.g., the number density of clusters will be too small [12].
Tilted ΛCDM models with higher Ω0, and therefore lower H0 for t0 >∼ 13 Gyr,
appear to have a better hope of fitting the available data, based on comparing
5
quasi-linear calculations to the data [12, 38]. But all cosmological models with
a cosmological constant Λ large enough to help significantly with the H0 − t0
problem are in trouble with new observations providing strong upper limits on
Λ [39]: gravitational lensing [40], HST number counts of ellptical galaxies [41],
and especially the preliminary results from measurements using high-redshift
Type Ia supernovae [42]. The analysis of the data from the first 7 of the Type
Ia supernovae from the LBL group [43] gave Ω0 = 1 − ΩΛ = 0.94+0.34
−0.28, or
equivalently ΩΛ = 0.06+0.28
−0.34 (< 0.51 at the 95% confidence level).
It is instructive to compare the Ω0 = 0.3, h = 0.7 ΛCDM model that we have
been considering with standard CDM and with CHDM. At k = 0.5h Mpc−1,
Figs. 5 and 6 of Ref. [45] show that the Ων = 0.3 CHDM spectrum and that
of a biased CDM model with the same σ8 = 0.67 are both in good agreement
with the values indicated for the power spectrum P (k) by the APM and CfA
data, while the CDM spectrum with σ8 = 1 is higher by about a factor of two.
As Figure 2 shows, CHDM with Ων = 0.2 in two neutrino species [8] also gives
nonlinear P (k) consistent with the APM data.
3 Cluster Baryons
I have recently reviewed the astrophysical data bearing on the values of the
fundamental cosmological parameters, especially Ω0 [39]. One of the arguments
against Ω = 1 that seemed hardest to answer was the "cluster baryon crisis" [46]:
for the Coma cluster the baryon fraction within the Abell radius is
fb ≡
Mb
Mtot
≥ 0.009 + 0.050h−3/2,
(1)
where the first term comes from the galaxies and the second from gas. If clusters
are a fair sample of both baryons and dark matter, as they are expected to
be based on simulations, then this is 2-3 times the amount of baryonic mass
expected on the basis of BBN in an Ω = 1, h ≈ 0.5 universe, though it is just
what one would expect in a universe with Ω0 ≈ 0.3. The fair sample hypothesis
implies that
Ω0 =
Ωb
fb
= 0.33(cid:18) Ωb
0.05(cid:19)(cid:18) 0.15
fb (cid:19) .
(2)
A review of the quantity of X-ray emitting gas in a sample of clusters [47]
finds that the baryon mass fraction within about 1 Mpc lies between 10 and
22% (for h = 0.5; the limits scale as h−3/2), and argues that it is unlikely that
(a) the gas could be clumped enough to lead to significant overestimates of
the total gas mass -- the main escape route considered in [46] (cf. also [48]).
If Ω = 1, the alternatives are then either (b) that clusters have more mass
than virial estimates based on the cluster galaxy velocities or estimates based
on hydrostatic equilibrium [49] of the gas at the measured X-ray temperature
6
CHDM, 2nu, Pcold(k)
Figure 2: Comparison of APM galaxy power spectrum (triangles) with nonlinear
cold particle power spectrum from CHDM model considered in this paper (upper
solid curve). The dotted curves are linear theory; upper curves are for z = 0,
lower curves correspond to the higher redshift z = 9.9. (From Ref. [7].)
(which is surprising since they agree [50]), (c) that the usual BBN estimate
Ωb ≈ 0.05(0.5/h)2 is wrong, or (d) that the fair sample hypothesis is wrong [51].
Regarding (b), it is interesting that there are indications from weak lensing [52]
that at least some clusters may actually have extended halos of dark matter --
something that is expected to a greater extent if the dark matter is a mixture
of cold and hot components, since the hot component clusters less than the
cold [53, 54].
If so, the number density of clusters as a function of mass is
higher than usually estimated, which has interesting cosmological implications
(e.g., σ8 is a little higher than usually estimated). It is of course possible that the
solution is some combination of alternatives (a)-(d). If none of the alternatives
is right, then the only conclusion left is that Ω0 ≈ 0.33. The cluster baryon
problem is clearly an issue that deserves very careful examination.
It has recently been argued [55] that CHDM models are compatible with
7
the X-ray data within observational uncertanties of both the BBN predictions
Indeed, the rather high baryon fraction Ωb ≈ 0.1(0.5/h)2
and X-ray data.
implied by recent measurements of low D/H in two high-redshift Lyman limit
systems [56, 57] helps resolve the cluster baryon crisis for all Ω = 1 models --
it is escape route (c) above. With the higher Ωb implied by the low D/H, there
is now a "baryon cluster crisis" for low-Ω0 models! Even with a baryon fraction
at the high end of observations, fb <∼ 0.2(h/0.5)−3/2, the fair sample hypothesis
with this Ωb implies Ω0 >∼ 0.5(h/0.5)−1/2.
4 Warm Dark Matter vs. CHDM
It will be instructive to look briefly at Warm Dark Matter (WDM), both to see
that some variants of CDM have less success than others in fitting cosmological
observations, and also because there is renewed interest in WDM. Although
CHDM and WDM are similar in the sense that both are intermediate models
between CDM and HDM, CHDM and WDM are quite different in their impli-
cations.
The problems with a pure Hot Dark Matter (HDM) adiabatic cosmology are
well known: free-streaming of the hot dark matter completely destroys small-
scale fluctuations, so that the first structures that can form are on the mass
scale of clusters or superclusters, and galaxies must form by fragmentation of
these larger structures; but observations show that galaxies are much older than
superclusters, which have low overdensity and are still forming. Moreover, with
the COBE upper limit to the normalization of HDF, hardly any structure at all
will form even by the present epoch.
WDM is a simple modification of HDM, obtained by changing the assumed
average number density n of the particles. In the usual HDM, the dark mat-
ter particles are neutrinos, each species of which has nν = 108 cm−3, with a
corresponding mass of m0 = Ωνρ0/nν = Ων92h2 eV, with Ων = 1 − Ωb ≈ 1.
In WDM, there is a new parameter, m/m0, the ratio of the mass of the warm
particle to the above neutrino mass; correspondingly, the number density of
the warm particles is reduced by the inverse of this factor, so that their total
contribution to the cosmological density is unchanged. Pagels and I [58] pro-
posed perhaps the first WDM particle candidate, a light gravitino, which was
the lightest supersymmetric particle (LSP) in the earliest version of supersym-
metric phenomenology (which was subsequently largely abandoned in favor of a
hidden-sector sypersymmetry breaking scheme, but which is now being recon-
sidered: cf. [59]). Olive & Turner [60] proposed left-handed neutrinos as WDM.
In both cases, these particles interact much more weakly than neutrinos, de-
couple earlier from the hot big bang, and thus have diluted number density
compared to neutrinos since they do not share in the entropy released by the
subsequent annihilation of species such as quarks. This is analogous to (but
more extreme than) the neutrinos themselves, which have lower number density
8
today than photons because the neutrinos decouple before e+e− annihilation
(and also because they are fermions).
In order to investigate the cosmological implications of any dark matter can-
didate, it is necessary to work out the gravitational clustering of these particles,
first in linear theory, and then after the amplitude of the fluctuations grows into
the nonlinear regime. Colombi, Dodelson, & Widrow [61] recently did this for
WDM, and Figure 1 in their paper compares the square of the linear transfer
functions for WDM and CHDM.
One often can study large scale structure just on the basis of such linear
calculations, without the need to do computationally expensive simulations of
the non-linear gravitational clustering. Such studies have shown that matching
the observed cluster and galaxy correlations on scales of about 20-30 h−1 Mpc
in CDM-type theories requires that the "Excess Power" EP ≈ 1.3, where
EP ≡
σ(25 h−1 Mpc)/σ(8 h−1 Mpc)
[σ(25 h−1 Mpc)/σ(8 h−1 Mpc)]sCDM
,
(3)
and as usual σ(r) = (δρ/ρ)(r) is the rms fluctuation amplitude in randomly
placed spheres of radius r. The EP parameter was introduced in the COBE-
DMR interpretation paper [62], and in a recent paper [24] we have shown that
EP is related to the spectrum shape parameter Γ [64] by Γ ≈ 0.5(EP )−3.3.
For CHDM and other models, this is a useful generalization since the cluster
correlations do seem to be a function of this generalized Γ, with Γ ≈ 0.23 to
match cluster correlation data [24]. Peacock & Dodds [63] have shown that
Γ ≈ 0.23 also is required to match large scale galaxy clustering data.
Since calculating σ(r) is a simple matter of integrating the power spectrum
times the top-hat window function,
σ(r) = Z ∞
0
P (k)W (kr)k2dk
(4)
the linear calculations immediately allow determination of EP for WDM and
CHDM. Ref. [61] shows that for WDM to give the required EP , the parameter
value m/m0 ≈ 1.5 − 2, while for CHDM the required value of the CHDM
parameter is Ων ≈ 0.3. But for WDM with m/m0 >∼ 2, the spectrum lies
a lot lower than the CDM spectrum at k >∼ 0.3h−1 Mpc (length scales λ <∼
20 h−1 Mpc), which in turn implies that formation of galaxies, corresponding to
the gravitational collapse of material in a region of size ∼ 1 Mpc, will be strongly
suppressed compared to CDM. Thus WDM will not be able to accommodate
simultaneously the distribution of clusters and galaxies. But CHDM will do
much better.
Probably the only way to accommodate WDM in a viable cosmological model
is as part of a mixture with hot dark matter, which might even arise naturally
in a supersymmetric model [65] of the sort in which the gravitino is the LSP
[59]. Cold plus "volatile" dark matter is a related possibility [66].
9
There are many more parameters needed to describe the presently available
data on the distribution of galaxies and clusters and their formation history
than the few parameters needed to specify a CDM-type model. Thus it should
not be surprising that at most a few CDM variant theories can fit all this data.
Once it began to become clear that standard CDM was likely to have problems
accounting for all the data, after the discovery of large-scale flows of galaxies
was announced in early 1986 [67], I advised Jon Holtzman in his dissertation
research to work out the linear theory for a wide variety of CDM variants [21]
so that we could see which ones would best fit the data [22]. The clear winners
were CHDM with Ων ≈ 0.3 if h ≈ 0.5, and ΛCDM with Ω0 ≈ 0.2 if h ≈ 1.
Variants of both these models remain perhaps the best bets still.
5 CHDM: Early Structure Troubles?
Aside from the possibility mentioned at the outset that the Hubble constant
is too large and the universe too old for any Ω = 1 model to be viable, the
main potential problem for CHDM appears to be forming enough structure
at high redshift. Although, as I mentioned above, the prediction of CHDM
that the amount of gas in damped Lyman α systems is starting to decrease at
high redshift z >∼ 3 seems to be in accord with the available data, the large
velocity spread of the associated metal-line systems may indicate that these
systems are more massive than CHDM would predict (see e.g., [68, 69]). Also,
results from a recent CDM hydrodynamic simulation [70] in which the amount
of neutral hydrogen in protogalaxies seemed consistent with that observed in
damped Lyman α systems led the authors to speculate that CHDM models
would produce less than enough; however, since the regions identified as damped
Lyman α systems in the simulations were not actually resolved, this will need
to be addressed by higher resolution simulations for all the models considered.
Finally, Steidel et al. [71] have found objects by their emitted light at red-
shifts z = 3 − 3.5 apparently with relatively high velocity dispersions (indicated
by the equivalent widths of absorption lines), which they tentatively identify
as the progenitors of giant elliptical galaxies. Assuming that the indicated ve-
locity dispersions are indeed gravitational velocities, Mo & Fukugita (MF) [72]
have argued that the abundance of these objects is higher than expected for the
COBE-normalized Ω = 1 CDM-type models that can fit the low-redshift data,
including CHDM, but in accord with predictions of the ΛCDM model consid-
ered here.
(In more detail, the MF analysis disfavors CHDM with h = 0.5
and Ων >∼ 0.2 in a single species of neutrinos. They apparently would argue
that this model is then in difficulty since it overproduces rich clusters -- and
if that problem were solved with a little tilt np ≈ 0.9, the resulting decrease in
fluctuation power on small scales would not lead to formation of enough early
objects. However, if Ων ≈ 0.2 is shared between two species of neutrinos, the
resulting model appears to be at least marginally consistent with both clusters
10
and the Steidel objects even with the assumptions of MF. The ΛCDM model
with h = 0.7 consistent with the most restrictive MF assumptions has Ω0 >∼ 0.5,
hence t0 <∼ 12 Gyr. ΛCDM models having tilt and lower h, and therefore more
consistent with the small-scale power constraint discussed above, may also be
in trouble with the MF analysis.) But in addition to uncertainties about the
actual velocity dispersion and physical size of the Steidel et al. objects, the
conclusions of the MF analysis can also be significantly weakened if the grav-
itational velocities of the observed baryons are systematically higher than the
gravitational velocities in the surrounding dark matter halos, as is perhaps the
case at low redshift for large spiral galaxies [73], and even more so for elliptical
galaxies which are largely self-gravitating stellar systems in their central regions.
Given the irregular morphologies of the high-redshift objects seen in the
Hubble Deep Field [74] and other deep HST images, it seems more likely that
they are relatively low mass objects undergoing starbursts, possibly triggered by
mergers, rather than galactic protospheroids. Since the number density of the
brightest of such objects may be more a function of the probability and duration
of such starbursts rather than the nature of the underlying cosmological model,
it may be more useful to use the star formation or metal injection rates [75]
indicated by the total observed rest-frame ultraviolet light to constrain models
[76]. The available data on the history of star formation [77, 78, 75] suggests that
most of the stars and most of the metals observed formed relatively recently,
after about redshift z ∼ 1; and that the total star formation rate at z ∼ 3
is perhaps a factor of 3 lower than at z ∼ 3, with yet another factor of ∼ 3
falloff to z ∼ 4 (although the rates at z >∼ 3 could be higher if most of the
star formation is in objects too faint to see). This is in accord with indications
from damped Lyman α systems [79] and expectations for Ω = 1 models such as
CHDM, but not with the expectations for low-Ω0 models which have less growth
of fluctuations at recent epochs, and therefore must form structure earlier. But
this must be investigated using more detailed modelling, including gas cooling
and feedback from stars and supernovae [76], before strong conclusions can be
drawn.
There is another sort of constraint from observed numbers of high-redshift
protogalaxies that would appear to disfavor ΛCDM. The upper limit on the
number of z >∼ 4 objects in the Hubble Deep Field (which presumably corre-
spond to smaller-mass galaxies than most of the Steidel objects) is far lower
than the expectations in low-Ω0 models, especially with a positive cosmological
constant, because of the large volume at high redshift in such cosmologies [80].
Thus evidence from high-redshift objects cuts both ways, and it is too early to
tell whether high- or low-Ω0 models will ultimately be favored.
11
6 Advantages of Mixed CHDM Over Pure CDM
Models
There are three basic reasons why a mixture of cold plus hot dark matter works
better than pure CDM without any hot particles: (1) the power spectrum shape
P (k) is a better fit to observations, (2) there are indications from observations
for a more weakly clustering component of dark matter, and (3) a hot compo-
nent may help avoid the too-dense central dark matter density in pure CDM
dark matter halos. I will discuss each in turn.
(1) Spectrum shape. As I explained in discussing WDM vs. CHDM
above, the pure CDM spectrum P (k) does not fall fast enough on the large-k
side of its peak in order to fit indications from galaxy and cluster correlations and
power spectra. The discussion there of "Excess Power" is a way of quantifying
this. This is also related to the overproduction of clusters in pure CDM. The
obvious way to prevent Ω = 1 sCDM normalized to COBE from overproducing
clusters is to tilt it a lot (the precise amount depending on how much of the
COBE fluctuations are attributed to gravity waves, which can be increasingly
important as the tilt is increased). But a constraint on CDM-type models
that is likely to follow both from the high-z data just discussed and from the
preliminary indications on cosmic microwave anisotropies at and beyond the
first acoustic peak from the Saskatoon experiment [81] is that viable models
cannot have much tilt, since that would reduce too much both their small-
scale power and the amount of small-angle CMB anisotropy. As I have already
explained, by reducing the fluctuation power on cluster scales and below, COBE-
normalized CHDM naturally fits both the CMB data and the cluster abundance
without requiring much tilt. The need for tilt is further reduced if a high
baryon fraction Ωb >∼ 0.1 is assumed [82], and this also boosts the predicted
height of the first acoustic peak. No tilt is necessary for Ων = 0.2 shared
between Nν = 2 neutrino species with h = 0.5 and Ωb = 0.1.
Increasing
the Hubble parameter in COBE-normalized models increases the amount of
small-scale power, so that if we raise the Hubble parameter to h = 0.6 keeping
Ων = 0.2 and Ωb = 0.1(0.5/h)2 = 0.069, then fitting the cluster abundance
in this Nν = 2 model requires tilt 1 − np ≈ 0.1 with no gravity waves (i.e.,
T /S = 0; alternatively if T /S = 7(1 − np) is assumed, about half as much tilt
is needed, but the observational consequences are mostly very similar, with a
little more small scale power). The fit to the small-angle CMB data is still good,
and the predicted Ωgas in damped Lyman α systems is a little higher than for
the h = 0.5 case. The only obvious problem with h = 0.6 applies to any Ω = 1
model -- the universe is rather young: t0 = 10.8 Gyr.
(2) Need for a less-clustered component of dark matter. The fact
that group and cluster mass estimates on scales of ∼ 1 h−1 Mpc typically give
values for Ω around 0.1-0.2 [84], while larger-scale estimates give larger values
around 0.3-1 [2] suggests that there is a component of dark matter that does
12
not cluster on small scales as efficiently as cold dark matter is expected to do.
In order to quantify this, my colleagues and I have performed the usual group
M/L measurement of Ω0 on small scales in "observed" Ω = 1 simulations of both
CDM and CHDM [83]. We found that COBE-normalized Ων = 0.3 CHDM gives
ΩM/L = 0.12 − 0.18 compared to ΩM/L = 0.15 for the CfA1 catalog analyzed
exactly the same way, while for CDM ΩM/L = 0.34 − 0.37, with the lower
value corresponding to bias b = 1.5 and the higher value to b = 1 (still below
the COBE normalization). Thus local measurements of the density in Ω = 1
simulations can give low values, but it helps to have a hot component to get
values as low as observations indicate. We found that there are three reasons
why this virial estimate of the mass in groups misses so much of the matter in the
simulations: (1) only the mass within the mean harmonic radius rh is measured
by the virial estimate, but the dark matter halos of groups continue their roughly
isothermal falloff to at least 2rh, increasing the total mass by about a factor of 3
in the CHDM simulations; (2) the velocities of the galaxies are biased by about
70% compared to the dark matter particles, which means that the true mass
is higher by about another factor of 2; and (3) the groups typically lie along
filaments and are significantly elongated, so the spherical virial estimator misses
perhaps 30% of the mass for this reason. Our visualizations of these simulations
[53] show clearly how extended the hot dark matter halos are. An analysis of
clusters in CHDM found similar effects, and suggested that observations of the
velocity distributions of galaxies around clusters might be able to discriminate
between pure cold and mixed cold + hot models [54]. This is an area where
more work needs to be done -- but it will not be easy since it will probably
be necessary to include stellar and supernova feedback in identifying galaxies in
simulations, and to account properly for foreground and background galaxies in
observations.
(3) Preventing too dense centers of dark matter halos. Flores and
I [85] pointed out that dark matter density profiles with ρ(r) ∝ r−1 near the
origin from high-resolution dissipationless CDM simulations [87] are in serious
conflict with data on dwarf spiral galaxies (cf. also Ref.
[86]), and in possible
conflict with data on larger spirals [88] and on clusters (cf. [89, 90]). Navarro,
Frenk, & White [73] agree that rotation curves of small spiral galaxies such as
DDO154 and DDO170 are strongly inconsistent with their universal dark mat-
ter profile ρN F W (r) ∝ 1/[r(r + a)2]. I am at present working with Stephane
Courteau, Sandra Faber, Ricardo Flores, and others to see whether ρN F W is
consistent with data from high- and low-surface-brightness galaxies with mod-
erate to large circular velocities are consistent with this universal profile. The
failure of simulations to form cores as observed in dwarf spiral galaxies either
is a clue to a property of dark matter that we don't understand, or is telling
us the simulations are inadequate. It is important to discover whether this is a
serious problem, and whether inclusion of hot dark matter or of dissipation in
the baryonic component of galaxies can resolve it. It is clear that including hot
dark matter will decrease the central density of dark matter halos, both because
13
the lower fluctuation power on small scales in such models will prevent the early
collapse that produces the highest dark matter densities, and also because the
hot particles cannot reach high densities because of the phase space constraint
[91, 54]. But this may not be enough.
7 Best Bet CDM-Type Models
As I said at the outset, I think CHDM is the best bet if Ω0 turns out to be near
unity and the Hubble parameter is not too large, while ΛCDM is the best bet
if the Hubble parameter is too large to permit the universe to be older than its
stars with Ω = 1.
Both theories do seem less "natural" than sCDM. But although sCDM won
the beauty contest, it doesn't fit the data. CHDM is just sCDM with some light
neutrinos. After all, we know that neutrinos exist, and there is experimental
evidence -- admittedly not yet entirely convincing -- that at least some of these
neutrinos have mass, possibly in the few-eV range necessary for CHDM.
Isn't it an unnatural coincidence to have three different sorts of matter --
cold, hot, and baryonic -- with contributions to the cosmological density that
are within an order of magnitude of each other? Not necessarily. All of these
varieties of matter may have acquired their mass from (super?)symmetry break-
ing associated with the electroweak phase transition, and when we understand
the nature of the physics that determines the masses and charges that are just
adjustable parameters in the Standard Model of particle physics, we may also
understand why Ωc, Ων, and Ωb are so close. In any case, CHDM is certainly
not uglier than ΛCDM.
In the ΛCDM class of models, the problem of too much power on small
scales that I discussed at some length for Ω0 = 0.3 and h = 0.7 ΛCDM implies
either that there must be some physical mechanism that produces strong, scale-
dependent anti-biasing of the galaxies with respect to the dark matter, or else
that higher Ω0 and lower h are preferred, with a significant amount of tilt to get
the cluster abundance right and avoid too much small-scale power [12]. Higher
Ω0 >∼ 0.5 also is more consistent with the evidence summarized above against
large ΩΛ and in favor of larger Ω0, especially in models such as ΛCDM with
Gaussian primordial fluctuations. But then h <∼ 0.63 for t0 >∼ 13 Gyr.
Among CHDM models, having Nν = 2 species share the neutrino mass gives
a better fit to COBE, clusters, and small-scall data than Nν = 1, and moreover
it appears to be favored by the available experimental data [8]. But it remains
to be seen whether CHDM models can fit the data on structure formation at
high redshifts.
Acknowledgments. This work was partially supported by NASA and NSF
grants at UCSC. I thank my collaborators, especially Anatoly Klypin, for many
helpful discussions of the material presented here.
14
References
[1] G.R. Blumenthal, S. Faber, J.R. Primack, & M.J. Rees 1994, Nature, 311,
517.
[2] A. Dekel 1994, Ann. Rev. Astron. Astroph., 32, 371; and Dekel & Burstein,
these proceedings.
[3] S.D.M. White, G. Efstathiou, & C.S. Frenk 1993, MNRAS, 262, 1023.
[4] E.g., S. Olivier, J. Primack, G.R. Blumenthal, & A. Dekel 1993, ApJ 408,
17.
[5] C.M. Baugh & G. Efstathiou 1994, MNRAS, 267, 32; see also Efstathiou's
contribution in this volume.
[6] S. Zaroubi, A. Dekel, Y. Hoffman, T. Kolatt 1996, astro-ph/9603068; cf.
T. Kolatt & A. Dekel 1966, astro-ph/9512132, ApJ submitted, and Dekel's
contribution in this volume.
[7] J.R. Primack & A. Klypin 1996, in Proc. Internat. Conf. on Sources and
Detection of Dark Matter in the Universe, UCLA, February 1996, D. Cline
and D. Sanders, eds, Nucl. Phys. B Proc. Suppl., in press.
[8] J.R. Primack, J. Holtzman, A. Klypin, & D.O. Caldwell, 1995, Phys. Rev.
Lett., 74, 2160.
[9] D. Pogosyan, & A.A. Starobinsky 1995, ApJ, 447, 465; A. Liddle, D.H.
Lyth, R.K. Schaefer, Q. Shafi, & P.T.P. Viana 1996, MNRAS, 281, 531,
and references therein.
[10] S. Ghigna, S. Borgani, S. Bonometto, L. Guzzo, A. Klypin, J.R. Primack,
R. Giovanelli, & M. Haynes 1994, ApJ, 437, L71.
[11] S. Ghigna, S. Borgani, M. Tucci, S. Bonometto A. Klypin, & J.R. Primack
1996, ApJ, submitted.
[12] A. Klypin, J.R. Primack, & J. Holtzman, ApJ, 466, 1.
[13] G.M. Fuller, J.R. Primack, & Y.-Z. Qian 1995, Phys. Rev. D, 52, 1288.
[14] C. Athanassopoulos et al. 1996, nucl-ex/9605001, Phys. Rev. submitted,
and nucl-ex/9605003, Phys. Rev. Lett., submitted (also available at http:
//nu1.lampf.lanl.gov/ lsnd/). Using only the data for Eν = 36 − 60 MeV,
for which the background is lower than the larger neutrino energy range
used in last figure in these preprints, lowers the lower limit on sin2 2θ and
thus increases the range of allowed ∆m2
µe (private communications from D.
Caldwell & S. Yellin, May 1996; D.O. Caldwell, in Neutrino'96, Helsinki,
June 1996).
15
[15] P.J.E. Peebles, Physical Cosmology (Princeton University Press, 1993), eq.
(5.150).
[16] D.O. Caldwell 1995, in Trends in Astroparticle Physics, Stockholm, Sweden
22-25 September 1994, eds. L. Bergstrom, P. Carlson, P.O. Hulth, & N.
Snellman, Nucl. Phys. B, Proc. Suppl., 43, 126.
[17] Y. Fukuda, et al. 1994, Phys. Lett. B, 280, 146.
[18] The Kamiokande data is consistent with atmospheric νµ oscillating to any
other neutrino species with a large mixing angle. But (see discussion and
references in, e.g., [8, 13]) νµ oscillating to νe with a large mixing angle
is probably inconsistent with reactor and other data, and νµ oscillating to
a sterile neutrino νs (i.e., one that does not interact via the usual weak
interactions) with a large mixing angle is inconsistent with the usual Big
Bang Nucleosynthesis constraints.
[19] Y. Suzuki, at Neutrino'96, Helsinki, June 1996, and private communication
June 1996.
[20] D. Pogosyan, & A.A. Starobinsky 1995, ApJ, 447, 465; A. Liddle, et al.,
Ref. [9].
[21] J. Holtzman 1989, ApJS, 71, 1.
[22] J. Holtzman & J.R. Primack 1993, ApJ, 405, 428.
[23] D. Pogosyan, & A.A. Starobinsky 1995, astro-ph/9502019; K.S. Babu, R.K.
Schaefer, & Q. Shafi 1996, Phys. Rev. D, 53, 606.
[24] S. Borgani, L. Moscardini, M. Plionis, K.M. G´orski, J. Holtzman, A.
Klypin, J.R. Primack, C.L. Smith, & R. Stompor 1996, submitted to New
Astronomy.
[25] M. Davis, F. Summers, & D. Schlegel 1992, Nature, 359, 393.
[26] A. Klypin, J. Holtzman, J.R. Primack, & E. Regos 1993, ApJ, 416, 1.
[27] H.J. Mo & J. Miralda-Escude 1994, ApJ, 430, L25; G. Kauffmann & S.
Charlot 1994, ApJ, 430, L97; C.-P. Ma & E. Bertschinger 1994, ApJ, 434,
L5.
[28] A. Klypin, S. Borgani, J. Holtzman, & J.R. Primack 1995, ApJ, 444, 1.
[29] C.-P. Ma 1995, in Dark Matter, AIP Conference Proceedings 336, p. 420.
[30] L.J. Storrie-Lombardi, R.G. McMahon, & M.J.
Irwin 1996, astro-
ph/9608147, MNRAS, in press. Cf. Art Wolfe's contribution in these pro-
ceedings.
16
[31] Chaboyer, B., Demarque, P., Kernan, P.J., & Krauss, L.M. 1996, Science,
271, 957.
[32] T.D. Oswalt, J.A. Smith, & M.A. Wood 1996, Nature, 382, 692.
[33] O. Lahav, P. Lilje, J.R. Primack, & M.J. Rees 1991, MNRAS, 251, 128.
[34] S.M. Carroll, W.H. Press, & E.L. Turner 1992, Ann. Rev. Astron. Astro-
phys, 30, 499.
[35] P.J.E. Peebles 1984, ApJ, 284, 439.
[36] G. Efstathiou, W.J. Sutherland & S.J. Maddox 1990, Nature 348, 705; L.A.
Kofman, N.Y. Gendin, & N.A. Bahcall 1993, ApJ, 413, 1; R. Cen, N.Y.
Gendin & J.P. Ostriker 1993, ApJ, 417, 387; R.A. Croft & G. Efstathiou
1994, MNRAS, 267, 390; J.P. Ostriker & P.J. Steinhardt 1995, Nature, 377,
600; L.M. Krauss, & M.S. Turner 1995, General Relativity & Gravitation,
27, 1137.
[37] G. Kauffmann, A. Nusser, & M. Steinmetz, astro-ph/9512009.
[38] A.R. Liddle, D.H. Lyth, P.T.P. Viana, & M. White, 1996, MNRAS, 282,
281.
[39] J.R. Primack, astro-ph/9604184, to appear in International School of
Physics "Enrico Fermi", Course CXXXII: Dark Matter in the Universe,
Varenna, eds. S. Bonometto, J.R. Primack, & A. Provenzale (IOS Press,
Amsterdam, in press). A popular version is J. Roth & J.R. Primack 1996,
Sky & Telescope, 91(1), 20.
[40] C. Kochanek 1996, ApJ, 466, 638.
[41] S.P. Driver, R.A. Windhorst, S. Phillipps, & P.D. Bristow 1996, ApJ 461,
525.
[42] S. Perlmutter et al 1996., astro-ph/9602122, to appear in Thermonuclear
Supernovae (NATO ASI), eds. R. Canal, P. Ruiz-LaPuente, & J. Isern.
[43] S. Perlmutter et al. 1996, astro-ph/9608192, submitted to ApJ.
[44] G. Yepes, R. Kates, A. Khokhlov & A. Klypin 1996, astro-ph/9605182.
[45] A. Klypin, R. Nolthenius, & J.R. Primack 1995, astro-ph/9502062, ApJ,
in press (Jan 10, 1997).
[46] S.D.M. White, & C.S. Frenk 1991, ApJ, 379, 52; S.D.M. White et al. 1993,
Nature, 366, 429.
[47] D.A. White, & A.C. Fabian 1995, MNRAS, 273, 72.
17
[48] K.F. Gunn & P.A. Thomas 1995, astro-ph/9510082.
[49] C. Balland & A. Blanchard 1995, astro-ph/9510130.
[50] N.A. Bahcall & L.M. Lubin 1994, ApJ, 426, 513; M. Bartelmann & R.
Narayan 1995, in Dark Matter, AIP Conference Proceedings 336, p. 307.
[51] M. Lowenstein & R.F. Mushotzky 1996, astro-ph/9608111, discuss two poor
clusters with similar total mass distributions but baryon fractions differing
by a factor ∼ 2.
[52] G. Squires, N. Kaiser, A. Babul, G. Fahlman, D. Woods, M. Neumann, &
H. Bohringer 1996, ApJ, 461, 572. Cf. G. Luppino, M. Lowenstein, in Proc.
Internat. Conf. on Sources and Detection of Dark Matter in the Universe,
UCLA, February 1996, D. Cline & D. Sanders, eds, Nucl. Phys. B Proc.
Suppl., in press.
[53] D. Brodbeck, D. Hellinger, R. Nolthenius, J.R. Primack, & A. Klypin 1997,
ApJ, in press (with accompanying video). Still and Mpeg excerpts are avail-
able from Primack web page at http: //physics.ucsc.edu
[54] L. Kofman, A. Klypin, D. Pogosyan, & J.P. Henry 1995, astro-ph/9509145.
[55] R.W. Strickland & D.N. Schramm, astro-ph/9511111.
[56] D. Tytler, X.-M. Fan & S. Burles 1996, Nature, 381, 207; S. Burles &
D. Tytler 1996, astro-ph/9603070; D. Tytler & S. Burles 1996, astro-
ph/9606110, in Proc. Origin of Matter and Evolution of Galaxies in the
Universe, in press.
[57] See Hogan's contribution to this volume for a dissenting view. The only
high-redshift deuterium detections that are secure are those by Tytler et
al., but Hogan questions whether Tytler et al. might have overestimated
the hydrogen content in the same systems, and thereby underestimated
D/H; Tytler & Burles answer this objection in detail in the last ref. in [56].
[58] H. Pagels & J.R. Primack 1982, Phys. Rev. Lett., 48, 223.
[59] M. Dine, A. Nelson, Y. Nir, & Y. Shirman 1996, Phys. Rev. D53, 2658; S.
Dimopoulos, M. Dine, S. Raby, & S. Thomas 1996, Phys. Rev. Lett., 76,
3494.
[60] K.A. Olive & M.S. Turner 1982, Phys. Rev. D25, 213.
[61] S. Colombi, S. Dodelson, & L.M. Widrow 1996, ApJ, 458, 1.
[62] E.L. Wright et al. 1992, ApJ, 396, L13.
18
[63] J.A. Peacock & S.J. Dodds 1994, MNRAS, 1020; -- 1996, MNRAS, 280,
L19.
[64] G. Efstathiou, J.R. Bond, & S.D.M. White 1992, MNRAS, 258, 1p. Cf.
J.M. Bardeen, J.R. Bond, N. Kaiser, & A.S. Szalay 1986, ApJ, 304, 15. For
CDM and the ΛCDM family of models, Γ = Ωh.
[65] S. Borgani, A. Masiero, & M. Yamaguchi 1996, hep-ph/9605222.
[66] E. Pierpaoli, P. Coles, S. Bonometto, & S. Borgani 1996, astro-ph/9603150,
ApJ, in press, and references therein.
[67] D. Burstein 1986, in Galaxy Distances and Deviations from Universal Ex-
pansion, eds B.F. Madore & R.B. Tully, Dordrecht.
[68] L. Lu, W.L.W. Sargent, D.S. Womble, & T.A. Barlow 1996, ApJ, 457, L1.
[69] A. Wolfe, these proceedings. Note however that the effects of dust extinction
(cf. Y. Pei & S.M. Fall 1995, ApJ, 454, 69; M. Pettini et al. 1996, astro-
ph/9607093) and gravitational lensing (M. Bartelmann & A. Loeb 1996,
ApJ, 457, 529) should be taken into account in modelling damped Lyman
α systems.
[70] N. Katz, D.H. Weinberg, L. Hernquist, & J. Miralda-Escude 1996, ApJ,
457, L57.
[71] C.A. Steidel et al. 1996, ApJ, 462, L17.
[72] H.J. Mo & M. Fukugita 1996, ApJ, 467, L9.
[73] J.F. Navarro, C.S. Frenk, & S.D.M. White 1996, ApJ, 462, 563. Also, D.
Zaritsky & S.D.M. White 1994, ApJ, 435, 599 (and work in preparation)
find little correlation between the rotation velocity of optical galaxies and
that of their extended dark matter halos. I thank M. Steinmetz for empha-
sizing the relevance of these facts.
[74] S. van den Bergh et al. 1996, AJ, 112, 359; R.G. Abraham et al. 1996,
MNRAS, 279, L47.
[75] P. Madau et al. 1996, astro-ph/9607172.
[76] R. Somerville and I are doing this for all currently popular models, using a
semi-analytic merging hierarchy model of the sort pioneered by G. Kauff-
mann, S.D.M. White, & B. Guiderdoni, 1993, MNRAS, 264, 201. We have
related work underway with G. Larsen and S.M. Fall.
[77] J. Gallego, J. Zamorano, A. Aragon-Salamanca, & M. Rego 1996, ApJ,
455, L1.
19
[78] S.J. Lilly, O. Le Fevre, F. Hammer, & D. Crampton 1996, ApJ, 460, L1.
[79] S.M. Fall, S. Charlot, & Y.C. Pei 1996, ApJ, 464, L43.
[80] K.M. Lanzetta, A. Yahil, & A. Fern´andez-Soto 1996, Nature, 381, 759; A.
Yahil, K.M. Lanzetta, A. Campos, & A. Fern´andez-Soto 1996, in prepara-
tion.
[81] C.B. Netterfield et al. 1996, ApJ, 445, L69. Cf. J. Silk, these proceedings.
[82] M. White, P.T.P. Viana, A.R. Liddle, & D. Scott 1996, astro-ph/9605057.
[83] R. Nolthenius, A. Klypin, J. Primack 1997, ApJ, in press.
[84] N. Bahcall, in these proceedings.
[85] R. Flores & J.R. Primack 1994, ApJ, 427, L1.
[86] B. Moore 1994, Nature, 370, 629.
[87] J. Dubinski & R.G. Carlberg 1991, ApJ, 378, 496; M.S. Warren, P.J. Quinn,
J.K. Salmon, & W.H. Zurek 1992, ApJ, 399, 405; M. Crone, A. Evrard, &
D. Richstone 1994, ApJ, 434, 402.
[88] R. Flores, J.R. Primack, G.R. Blumenthal, & S.M. Faber 1993, ApJ, 412,
443.
[89] J. Miralda-Escud´e 1995, ApJ, 438, 514.
[90] R. Flores & J.R. Primack 1996, ApJ, 457, L5.
[91] S. Tremaine & J.E. Gunn 1979, Phys. Rev. Lett., 42, 407.
20
|
astro-ph/9905193 | 1 | 9905 | 1999-05-15T07:24:30 | The Luminosity Function of M3 | [
"astro-ph"
] | We present a high precision, large sample luminosity function (LF) for the Galactic globular cluster M3. With a combination of ground based and Hubble Space Telescope data we cover the entire radial extent of the cluster. The observed LF is well fit by canonical standard stellar models from the red giant branch (RGB) tip to below the main sequence turnoff point. Specifically, neither the RGB LF-bump nor subgiant branch LF indicate any breakdown in the standard models. On the main sequence we find evidence for a flat initial mass function and for mass segregation due to the dynamical evolution of the cluster. | astro-ph | astro-ph | The Luminosity Function of M31
R.T. Rood2, E. Carretta3, B. Paltrinieri4, F. R. Ferraro5,6
F. Fusi Pecci6,7, B. Dorman8,
A. Chieffi9, O. Straniero10, R. Buonanno11
ABSTRACT
We present a high precision, large sample luminosity function (LF) for the Galactic
globular cluster M3. With a combination of ground based and Hubble Space Telescope
data we cover the entire radial extent of the cluster. The observed LF is well fit by
canonical standard stellar models from the red giant branch (RGB) tip to below the main
sequence turnoff point. Specifically, neither the RGB LF-bump nor subgiant branch LF
indicate any breakdown in the standard models. On the main sequence we find evidence
for a flat initial mass function and for mass segregation due to the dynamical evolution
of the cluster.
Subject headings: globular clusters: individual (M3) -- stars: red giant -- -- stars: evo-
lution
9
9
9
1
y
a
M
5
1
1
v
3
9
1
5
0
9
9
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
1Based in part on observations with the NASA/ESA Hubble Space Telescope, obtained at the Space Telescope Science Institute, which
is operated by AURA, Inc., under NASA contract NAS5-26555
2Astronomy Dept, University of Virginia, P.O.Box 3818, Charlottesville, VA 22903-0818; [email protected]
3Osservatorio Astronomico di Padova, vicolo dell'Osservatorio 5, 35122 Padova, ITALY; [email protected]
4Istituto di Astronomia-Universit´a "La Sapienza", via G.M. Lancisi 29, I-00161 Roma, Italy; [email protected]
5European Southern Observatory, Karl Schwarzschild Strasse 2, D-85748 Garching bei Munchen, Germany
6Osservatorio Astronomico di Bologna, via Ranzani 1, 40126 Bologna, ITALY; ferraro,[email protected]
7Stazione Astronomica di Cagliari, 09012 Capoterra, ITALY
8Laboratory for Astronomy & Solar Physics, Code 681, NASA/GSFC, Greenbelt MD 20771; [email protected]
9Istituto di Astrofisica Spaziale-CNR, 00044 Frascati, ITALY; [email protected]
10Osservatorio Astronomico di Collurania, 64100 Teramo, ITALY; [email protected]
11Osservatorio Astronomico di Roma, 00040 Monte Porzio, ITALY; [email protected]
1
1.
Introduction
One of the classical tests of stellar evolutionary cal-
culations is through the comparison of theoretical and
observed stellar luminosity functions (LFs) of globu-
lar clusters (e.g., Renzini & Fusi Pecci 1988). The
LF for the stars below the main sequence turn-off
(MSTO) can be related to the stellar initial mass func-
tion and be used as a probe of stellar dynamics within
a cluster. The LF of stars after the MSTO depends
primarily on the rate of evolution and provides a di-
rect and straight-forward test of evolutionary calcula-
tions. Basically the post-MSTO LF measures the de-
velopment of the hydrogen burning shell and its pro-
gression through the star. Features in the observed
LF can be related to interior structure. For example,
as first noted by Iben (1968), when the H-burning
shell advances through the composition discontinuity
left by the maximum penetration of convective enve-
lope, the luminosity briefly decreases. This leads to
the so-called LF bump. More subtle observables are
also present: the slope of the LF below and above the
LF bump are different because the H-burning shell is
in the first case passing through a region of varying
H abundance and later through a region of constant
H (Rood & Crocker 1985, Fusi Pecci et al. 1990). A
breakdown in canonical stellar evolution theory (Ren-
zini & Fusi Pecci 1988) can affect the LF in the region
of the subgiant branch (SGB).12 Examples include
extra energy transport by Weakly Interacting Mas-
sive Particles (WIMPs) (Faulkner & Swenson 1993)
or the gravitational settling of helium (Proffit & Van-
denBerg 1991, Castellani et al. 1997, Straniero et al.
1997), both of which affect the cluster age calibration.
To explore short evolutionary phases or to search
for the small changes in the LF produced by reason-
able modifications to the canonical models requires
large samples.
Indeed, to explore phenomena like
those cited above, such large samples are required
that photometry must extend to the crowded in-
ner parts of clusters. To be useful in constructing
12The canonical assumptions include the neglect of phenomena
like rotation, non-convective matter transport, etc.
in stellar
model calculations. Some of these are certainly present in real
stars, but they are very difficult to include in calculations. This
leads to two approaches to dealing with non-canonical phenom-
ena: one is to try to model the phenomenon ordinarily in a
simplistic way requiring the introduction of free parameters.
The other is to make the most stringent possible comparison
between observations and canonical models -- the approach we
adopt here.
a LF, photometric samples must be complete and
accurate -- a requirement which has come into conflict
with the necessity of working near cluster cores. Ad-
equate samples to identify the LF-bump were first
obtained for 47 Tuc by King, DaCosta & Demar-
que (1985) and for 11 other clusters by Fusi Pecci
et al. (1990) (see Cassisi & Salaris 1997 for a review).
At lower luminosities LF studies have produced some
puzzling results. Stetson (1991) in a LF formed from
combined data from the low metallicity clusters M68,
M92, and NGC 6397 found an excess of stars on
the SGB just above the MSTO. Bolte (1994) found
a similar result for M30 which has been confirmed
by Bergbusch (1996), Guhathakurta et al. (1998) and
Sandquist et al. (1999). This excess was especially
intriguing because that was the "signature" expected
if WIMP energy transport was important (Faulkner
& Swenson 1993). However Sandquist et al. (1996) in
the largest sample LF obtained to date did not find a
similar result for the somewhat more metal rich clus-
ter M5.
Clearly larger and better LFs are desirable, and
these are now possible with the Hubble Space Tele-
scope (HST ) which has the capability to obtain pho-
tometry near cluster centers. While very large HST
samples of fainter stars can be obtained, the small
field of view limits the number of evolved stars which
can be observed. However, the combination of ground
based photometry of the outer cluster and HST data
can yield a large sample LF which spans the range of
cluster stellar luminosities. In this paper we present
such an LF obtained for the cluster M3 (NGC 5272).
This is the largest, most precise, and most complete
LF yet obtained for a globular cluster.
2. HST-WFPC2 Data Sample
2.1. Observations and Data Analysis
The observations were obtained April 25, 1995 as
part of the HST program GO5496 (P.I. F. Fusi Pecci),
the first of our programs devoted to testing the the-
oretical models using luminosity functions of a set of
Galactic Globular Clusters with different metallicity.
A set of exposures was secured through UV (F255W
and F336W) and optical filters (F555W and F814W).
Preliminary results on UV data has been already
presented in Ferraro et al. (1997a) and Laget et al.
(1998); here we will report on results obtained using
only the V and I exposures. Total exposure times
were 400 sec in V (F555W) and 560 sec in I (F814W);
2
each exposure is listed in Table 1 by Ferraro et al.
(1997a).
The initial data analysis (including bias, dark and
flat-field corrections) was made through the standard
HST pipeline. A master unsharp mask image for
each filter was made using MIDAS to register all the
long-exposure images, yielding a median frame with
cosmic rays statistically eliminated. The search of the
individual star components was made on the V me-
dian frame, following the usual procedure with RO-
MAFOT (Buonanno et al. 1983). Then we performed
the PSF fitting on each individual frame separately
using a Moffat function plus a numerical map of the
residuals to better account for the contribution of the
stellar PSF wings. The average instrumental magni-
tudes for each star were transformed to the standard
Johnson V and I system using the Holtzmann et al.
(1995) recipes. The magnitudes thus obtained were
eventually shifted to match the independent zero-
point calibration adopted in Ferraro et al.
(1997b,
hereafter M3CCD). The V , V − I CMD for the HST
global sample is shown in Figure 1.
Special care has been devoted to ensure that the
effects of blends (and crowding) do not significantly
alter the derived LF. The effects of blending can be
quite problematic in CMDs of dense regions of glob-
ular clusters. On the main sequence, blended stars
can potentially be mistaken for binary stars, while on
the subgiant branch they be shifted erroneously to
brighter bins. Blends can mimic incompleteness by
shifting the two stars off the main loci such that they
are dropped from the LF. To minimize these effects
suspected blended objects with multiple stellar com-
ponents lying off the main branches of the CMD were
individually examined by eye, exploiting the fully in-
teractive capability of the package ROMAFOT. In
this way we were able to check and, if necessary, elim-
inate spurious objects. The degree to which blends
contaminate the final sample can be estimated from
the artificial stars tests described below. Some of the
artificial stars which were not recovered were lost as
blends. Thus the degree of incompleteness at a given
level serves as an upper limit to the blend-fraction.
2.2. Completeness
A crucial component in the calculation of an ac-
curate LF is the determination of corrections for in-
completeness.
In our data crowding is the primary
source of incompleteness. Moreover, because of the
large gradient in the spatial density of stars present
in our frame, the completeness at any level of magni-
tude must be determined as a function of the position
with respect to the cluster center.
In order to quantify the efficiency in detecting stars
as function of the magnitude and the degree of crowd-
ing we carried out extensive artificial star tests in the
V filter frames, following the procedure described be-
low.
1. Each chip was split into two annular regions cen-
tered on the cluster center:
INNER: r < 50′′ including the entire PC1 field
and the innermost part of the three WFC
fields;
OUTER: r > 50′′ covering the outer part of
the three WFCs.
M3 is a cluster of moderate central density.
With a core radius rc = 30′′ (Djorgovski 1993)
one would expect the stellar density to be ap-
proximately constant across the PC as is appar-
ent from Figure 2. However, the stellar density
has dropped appreciably at the outer parts of
the WFs. We have adopted 50′′ as a bound-
ary between regions of higher and lower pro-
jected stellar density. A map of the WFPC2
field of view is shown Figure 2 with the circle
(at r ∼ 50′′) delineating the two regions.
2. A set of artificial stars was generated using the
Tiny Tim package (Krist 1994), and these were
added to the original frames in random posi-
tions. In order to avoid a spurious crowding en-
hancement, the size of the added sample (about
300 total artificial images or ∼ 80 for each chip)
was ≤ 10% of the total number of stars origi-
nally found in each chip. Then the star search-
ing and fitting procedures were repeated on the
artifically enriched frames.
3. An artificial star was considered to be "recov-
ered" on the basis of the following criteria:
• the differences in position, ∆x, ∆y, be-
tween the input artificial star and the re-
covered one are ∆x, ∆y < 1 pixel;
• the difference in magnitude, ∆mag, is ∆mag <
0.3.
The fraction of artificial stars missed during any
step of the procedure is a good estimate of the
3
actual number of stars missed at each magni-
tude level during the reduction procedure.
and simply counting all the stars in each bin within
±5σV −I of the MRL.
4. The completeness factor (Φ) was finally com-
puted as the numbers of stars recovered (Nrec)
with respect to the number of stars simulated
(Nsim) during each trial. The final value of Φ
has been obtained as an average of 3 -- 6 inde-
pendent trials for each magnitude bin (0.2 mag
width) in each of the two regions (INNER and
OUTER).
The result of these tests is summarized in Fig-
ure 3a,b. The completeness at the MSTO level (V ∼
19) is > 90% in both the INNER and OUTER re-
gions, thus the star counts up to that level have been
corrected by less than 10%. The corrections for com-
pleteness were applied separately in each part (inner
or outer) of each field (PC, WFCs) so that the total
corrected sample properly accounts for the differing
degrees of crowding.
2.3. The Mean Ridge Line of the RGB, SGB,
and Upper MS
If there were no photometric errors all of the RGB,
SGB, and Upper MS stars would lie along a line in
the CMD. We approximate this line with a mean ridge
line (MRL) which has been determined by an iterative
procedure:
1. A first rough selection of stars belonging to
the RGB, SGB, MS was performed by eye, ex-
cluding the horizontal branch (HB), asymptotic
giant branch (AGB), and blue straggler stars
(BSS);
2. This preliminary sample was divided into 0.5
mag width bins;
3. For each bin the median V − I color was com-
puted;
4. All stars lying more than 5σV −I from the median
value was rejected.
Points (3) and (4) of the procedure were iterated
until the solution was stable. The resulting MRL
points are listed in Table 1. Only about 2% of the
sample was rejected in determining the MRL.
Luminosity functions were constructed for each
chip by dividing each individual CMD in 0.2 mag bins,
Figure 4 shows a zoom of luminosity function in
the Turn-Off region for the total HST sample, before
and after correction for completeness. Errors have
been computed accordingly to the following relation
(see also Bolte 1989):
σbin = [
N 1/2
Φ
+
N · σΦ
Φ2
]
(1)
where N is the number of stars observed in each bin,
Φ is the completeness factor, and σΦ is the associ-
ated error which was determined from the rms of the
repeated completeness trials.
3. Matching the Ground-Based Observations:
the Global LF
HST data presented here complement the large
data set we have obtained for M3 during the last
decades. The photographic sample was originally
presented by Buonanno et al. (1994) and the CCD
ground based data have been discussed by Ferraro
et al. (1993, 1997b). The entire photographic sample
extends over an area between 2′ and 7′ from the clus-
ter center. Here we consider a subsample (hereafter
referred to as PHOTO) of stars lying in the annulus
between 3.′5 < r < 6′ from the cluster center (region
AI in Buonanno et al. 1994). In this annulus all stars
down V ∼ 22 have been measured. The CCD data
cover a field of view 7′ × 7′ roughly centered on the
cluster center; stars in the innermost central regions
(r < 20′′) have been not measured. In this sample the
deeper exposures (V and I) only reached to V ∼ 20.
In order to ensure an high level (formally 100%) of
completeness we include only stars with V < 17 lying
in the annulus 100′′ -- 210′′ from the cluster center. We
will refer to this subsample as CCD. The area cov-
ered by the CCD subsample and the the others fields
in shown Figure 1 of Ferraro et al. (1997b).
In the PHOTO and CCD subsets (in analogy to the
HST sample) we applied the procedure described in
Section 2.2, selecting bona fide RGB-SGB-MS mem-
bers taking all stars within ±5σ in color from the
MRL in the V, B−V and V, V −I CMDs respectively.
The CMDs of all stars selected in the three samples
(PHOTO, CCD, HST respectively) are shown in Fig-
ure 5. For each of these samples we independently
derive the LF. Star counts have been corrected using
the respective completeness curve: the PHOTO sam-
4
V
HB
15.57
15.62
15.63
15.64
15.64
15.62
15.63
15.64
15.66
15.71
15.76
15.83
15.91
15.99
16.10
16.21
16.37
16.54
16.65
V − I
0.775
0.731
0.684
0.631
0.585
0.272
0.219
0.184
0.146
0.105
0.079
0.043
0.008
-0.021
-0.050
-0.069
-0.080
-0.088
-0.091
V
RGB
12.44
12.68
12.90
13.10
13.30
13.55
13.79
14.05
14.28
14.56
14.82
15.09
15.36
15.64
15.89
16.21
16.52
16.70
16.94
17.20
17.36
17.63
18.05
Table 1
Mean ridge line for the HST sample.
V − I
V
V − I
1.589
1.473
1.403
1.347
1.294
1.241
1.193
1.150
1.115
1.077
1.045
1.017
0.989
0.964
0.939
0.911
0.890
0.879
0.867
0.856
0.849
0.836
0.815
18.16
18.26
18.33
18.37
18.42
18.47
18.52
18.57
18.62
18.70
18.78
18.86
18.98
19.10
19.18
19.26
19.38
19.78
20.10
20.30
20.50
20.98
21.34
0.803
0.787
0.764
0.746
0.724
0.696
0.668
0.647
0.629
0.616
0.608
0.600
0.593
0.592
0.593
0.595
0.597
0.624
0.654
0.672
0.693
0.756
0.805
5
ple has been corrected using Table 7 by Buonanno et
al. (1994), the CCD sample (for V < 17) is virtually
complete, thus no correction has been applied, and
the HST sample has been corrected using the com-
pleteness curves plotted in Figure 3a,b. In construct-
ing the global LF we considered only the magnitude
range where the completeness was greater than 50%.
Thus the global LF is restricted to V ∼< 21 by the
most incomplete region in the HST sample -- the inner
region of the WF4 and WF2 cameras (Figure 3a,b).
Since the CCD sample spanned a different magni-
tude range than the others, we determined a scaling
factor to produce a proper match. We followed the
procedure explained in Bolte (1994):
let ni,k be the
number of stars, after the proper completeness cor-
rection, in bin i of the sub-sample k, where k = 1, 2, 3
indicates the PHOTO, the global HST , and the CCD
sub-samples, respectively. Thus, for k = 1, 2 the re-
sulting LF is simply given by the sum
N
′
i =
2
Xk=1
ni,k
in the range 12 < V < 21. However, in order to
include the CCD sub-sample (k = 3) we derived a
scaling factor s from the range in magnitude (12 <
V < 17.0) which is common to all sub-samples:
s = P2
k=1 ni,kPi ni,k
P3
k=1 ni,kPi ni,k
so that the number of stars for each bin in the com-
bined LF is
Ni = s × (N
′
i + ni,k=3).
Table 2 lists the final LF. Column 1 lists the mag-
nitude bins (which are 0.2 mag. in width); column 2
gives Ni, the total scaled number of objects per 0.2
mag; column 3 contains the errors in the Ni. Even
though it extends only 2 mag below the turnoff, this
LF, including more than 50000 stars, is the most pop-
ulated LF ever published for a galactic globular clus-
ter.
It is useful to compare our LF to that for M5 ob-
tained by Sandquist et al. (1996, hereafter SBSH), the
best previous globular cluster LF. Doing this, it is ap-
propriate to compare sample size in specific parts of
the CMD. Our "bright" sample which corresponds to
the upper 4.5 mag of the RGB (or V ≤ 17) contains
6
944 stars. For such bright stars SBSH were able to
work near the cluster center, and their sample in the
comparable RGB interval was about 1000 stars. For
fainter stars ground based photometry becomes in-
creasingly affected by problems of photometric blends
(see SBSH). Blends can be especially insidious when
they "move" a star from one part of the LF to an-
other, because this is just the sort of thing one would
expect in a breakdown of canonical stellar modeling.
To avoid blends only the sparsely populated outer
cluster can be used, and it is especially difficult to
get adequate samples of the SGB region. Here our
blend-free HST sample plays a crucial role. We have
6085 stars brighter than MSTO; SBSH have 3300. In
a 0.1 mag bin at turnoff we have 820 stars compared
to SBSH's 280.
4. Results
4.1. Preliminaries
There are certain issues which must be addressed
in comparing LFs either with other observational LFs
or with theoretical LFs. The first is how the samples
are normalized. Beyond this there is some interplay
between assumed distance modulus, age, and chemi-
cal composition.
4.1.1. Normalization
One must determine a "normalization" factor to
adjust for sample size.
In recent LF studies (e.g.
Stetson 1991, Bolte 1994, Sandquist et al. 1996) both
the lower RGB and the main sequence region have
been used for normalization. We adopt a different
approach normalizing to the total number of stars
brighter than the MSTO at V = 19.10. Operationally
this has the advantage that the turnoff is well defined
and because of the large number of stars involved the
normalization is not much affected by Poisson statis-
tics. The small imprecision in determining the loca-
tion of the turnoff (∼< 0.1 mag) produces only a very
small error (∼< 1%) in the normalization.
4.1.2. Chemical composition
For M3 Kraft et al. (1992), found [Fe/H] = −1.47,
based on detailed fine abundance analysis of high res-
olution spectra of red giant stars. However, their
study is not fully self-consistent since while stel-
lar analysis was performed using model atmospheres
from the grid of Bell et al. (1976, hereafter BEGN),
Luminosity function of the global sample (PHOTO+CCD+HST).
Table 2
V
12.70
12.90
13.10
13.30
13.50
13.70
13.90
14.10
14.30
14.50
14.70
14.90
15.10
15.30
15.50
15.70
15.90
16.10
16.30
16.50
16.70
16.90
17.10
17.30
Log(Ni)
σNi
0.70
0.30
0.78
0.85
0.95
0.85
0.90
0.90
1.11
1.08
1.32
1.15
1.34
1.38
1.76
1.60
1.51
1.65
1.70
1.79
1.85
1.95
2.04
2.06
0.45
0.26
0.45
0.41
0.52
0.50
0.50
0.47
0.61
0.55
0.67
0.58
0.77
0.72
0.92
0.87
0.75
0.83
0.93
0.96
1.00
1.06
1.09
1.11
V
17.50
17.70
17.90
18.10
18.30
18.50
18.70
18.90
19.10
19.30
19.50
19.70
19.90
20.10
20.30
20.50
20.70
20.90
21.10
21.30
21.50
21.70
21.90
Log(Ni)
σNi
2.05
2.22
2.26
2.46
2.72
2.97
3.06
3.18
3.25
3.32
3.37
3.42
3.45
3.49
3.53
3.56
3.56
3.59
3.58
3.58
3.55
3.57
3.66
1.10
1.20
1.23
1.35
1.52
1.68
1.75
1.83
1.88
1.94
1.98
2.01
2.04
2.08
2.12
2.14
2.16
2.19
2.21
2.24
2.27
2.35
2.48
7
the reference solar iron abundance was extracted from
the empirical Holweger and Muller (1974, hereafter
HM) solar model. This inconsistency was resolved by
Carretta & Gratton (1997) who use the Kurucz (1993)
grid of model atmospheres for both solar and stellar
analysis, giving [Fe/H] = −1.34. The offset with the
study of Kraft et al. (from which the observational
material was taken) is simply due to the fact that the
HM model is ∼ 150 K warmer than the BEGN models
in the line formation region, explaining the 0.13 dex
difference in the derived [Fe/H].
Oxygen in red giants can by affected by interior nu-
clear processing. Indeed, M3 is one of the several clus-
ters with an observed anti-correlation between O and
Na along the RGB (see for instance Kraft et al. 1992).
This anti-correlation arises because of non-standard
mixing on the upper RGB. Therefore, the best esti-
mate for the [O/Fe] ratio is that obtained considering
only unmixed stars, i.e. giants with [Na/Fe] ≈ 0.0
dex. Using Kraft et al. (1992) we estimate that
[O/Fe] = 0.19 ± 0.04 dex based on 5 stars.
One typically assumes that the other α-elements
(e.g. Kraft et al. 1993) have the same enhancement
as O and uses two metallicity parameters, [Fe/H] and
[α/Fe], in computing models. For many problems one
can get by with just one metallicity parameter using
the short cut of Salaris et al. (1993). This makes use of
the global metallicity ([M /H]) based on the mass frac-
tion of all elements heavier than helium as compared
to solar. They found that scaled solar models with
[Fe/H] = [M/H] mimicked models based on observed
[Fe/H] with α-enhancement. For M3 [M/H] = −1.2.
So while ideally we would compare M3 data to models
with [Fe/H] = −1.34 and [O/Fe] = 0.2, a comparison
to scaled solar models with [Fe/H] = [M/H] = −1.2
should be an adequate surrogate. Given typical er-
rors in metallicity determinations it is appropriate to
explore a region ±0.3 dex from these values.
4.1.3. Distance modulus and age
In principle, fitting cluster main sequences to sub-
dwarfs with precise Hipparcos parallaxes should lead
to relatively high precision distance moduli. However,
M3 was not included in the first round of clusters with
distances determined in this way (Reid 1997, Gratton
et al. 1997). The main reason was that the fiducial
cluster main sequence ridge line (from Ferraro et al.
1997b; CCD97) in the V, B − V plane did not reach
deep enough with the photometric accuracy required
to apply sub-dwarf fitting to determine distance mod-
8
uli. However, using an expanded sample of local sub-
dwarfs from the newly available Hipparcos catalogue
(Carretta et al. 1999a) it is possible to give an esti-
mate for the distance modulus of M3 in the framework
of the longer distance scale defined by Gratton et al.
(1997) and Reid (1997). With E(B −V ) = 0.02±0.01
and [Fe/H] = −1.30 (to be consistent with reddening
and metallicity scales used in Gratton et al. 1997), the
true distance modulus for M3 is (m − M )0 = 15.13
(Gratton 1998, private communication, adopting the
Ferraro et al 1997b photometry). A cautionary flag
to the accuracy of sub-dwarf fitting is set by the fact
that different groups using this method do not get
the same result (Pont et al. 1997; Reid 1997; Gratton
et al. 1997). Small details are important (Carretta
et al. 1999a), and systematic effects could leave resid-
ual errors of 0.2 -- 0.3 mag in (m − M ).
There are other independent estimates of the dis-
tance to M3. Cudworth (1979) gives a kinematic
distance for M3 of (m − M )0 = 14.91. From the
RR Lyrae Sandage and Cacciari (1990) give values
of (m − M )0 ranging from 14.81 to 15.00 depending
on which metallicity/absolute magnitude relation is
adopted for the RR Lyrae. Using the latest metal-
licity/absolute magnitude determined from Baade-
Wesselink distances to field RR Lyrae (Fernley et al.
1998) and our adopted metallicity ([M/H] = −1.2),
we get (m−M )0 = 14.86 (assuming E(B−V ) = 0.01).
In a recent analysis of the RR Lyrae distance scale,
De Santis (1996) gives (m − M )0 = 15.03. Marconi
et al. (1998) find that (m − M )0 = 14.94 gives a good
fit to the LF of the lower main sequence. Ferraro
et al. (1999) have recently obtained (m−M )0 = 15.05
within the framework of a homogeneous re-analysis
of the evolved sequences of the CMD in a sample of
about 60 GGCs. They derived the distance modu-
lus from the comparison of the observed level of the
ZAHB (VZAHB ) and the theoretical models computed
by SCL97. Their use of synthetic HBs to determine
the "observed" VZAHB is a significant improvement
over earlier applications of the same technique.
Given the present state of uncertainty we consider
it reasonable to consider (m − M )V = 14.8 -- 15.2. The
question of age is intertwined with that of distance.
The longer distance scales imply smaller ages. For
example, with the "sub-dwarf distance" given above
and using stellar models from Straniero et al. 1997
(private communication), the corresponding absolute
age for M3 is about 12 Gyr. An decrease in (m − M )
of 0.07 mag leads to an increase in age of 1 Gyr. Hence
ages of 11 -- 15 Gyr for M3 are not beyond question.
4.2. The Red Giant Branch
The differential LF for our global sample is shown
in Figure 6. The dominant determinant of RGB-LF
is simply energy conservation via the so-call energy
consumption theorem (Renzini & Buzzoni 1986). Ba-
sically the number of stars in some interval along the
RGB, Ni, depends inversely on the luminosity at that
point, Ni ∝ 1/L. So log Ni ∝ Mbol, and neglect-
ing bolometric corrections log Ni should drop linearly
with magnitude moving up the RGB just as seen.
Theoretical evolutionary models predict the exis-
tence of a "feature" in the RGB called the LF-bump.
This LF-bump marks the evolutionary stage at which
the H-burning shell passes through the composition
discontinuity left by the maximum penetration of
the convective envelope (Iben 1968, Renzini & Fusi
Pecci 1988). As emphasized first by Rood & Crocker
(1985), the practical detection of such a feature re-
quires very large RGB samples. Because of this, our
global sample gives us the possibility to obtain one of
the most precise determinations of the LF-bump.
Models also predict that in the integrated LF the
slope changes on either side of LF-bump. The slope
change results from the fact that the H-burning shell
is moving through a region of increasing hydrogen
abundance below the bump and constant hydrogen
abundance above the bump. This change in slope
can more reliably locate the bump than observations
of the bump itself (Rood & Crocker 1985; Fusi Pecci
et al. 1990).
In Figure 6 the LF-bump is very well defined in
both the differential and integrated LF, and is located
at V = 15.45 ± 0.05 (in good agreement with the pre-
vious determinations (Buonanno et al. 1994, Ferraro
et al. 1997b). Also note in Figure 6 the nomenclature
we use to refer to the different branches. Our nomen-
clature is that of the theorist in which the branches
are defined in terms of the supposed interior struc-
ture. There is often confusion, in particular, over the
term sub-giant branch. To the theorist, and in this
paper, SGB refers to the transition region between the
MSTO and the lower RGB where the ridge line in the
CMD is more horizontal V = 18.2 -- 18.6. Structurally
this is the branch of thick H-shell burning where the
degenerate core develops. For historical reasons ob-
servers often refer to stars along the lower-RGB as
sub-giants.
We compare our results to two new sets of theo-
retical evolutionary tracks/isochrones/LFs which use
the new OPAL opacities and equation of state (e.g.,
Rogers & Iglesias 1992). This should offer signif-
icant improvement over earlier models particularly
for opacity dependent LF features like the LF-bump.
The first set (the no-diffusion models described in
Straniero et al. 1997, hereafter SCL97) is available
for scaled-solar abundances and may be used for any
given value of α enhancement with the Salaris et al.
(1993) prescription. The second (Bergbush & Van-
denBerg 1999, herafter V97) has [α/Fe] = 0.3. The
V97 models also include improved color-temperature
relations. On the RGB the V97 models utilize a
non-Langrangian numerical technique which is not
optimal for investigations of the RGB bump. This
technique might be expected to smear and perhaps
slightly lower the LF-bump.
In Figure 7 the global differential LF is compared
to SCL97 and V97 models for (m − M )V = 15.03,
age = 12 Gyr. For SCL97 models we took [Fe/H] =
[M/H] = −1.2. The V97 models have [Fe/H] = −1.41
and [α/Fe] = 0.3 which is equivalent to [M/H] =
−1.2. The integrated LF compared to SCL97 is shown
in Figure 8. The fits are really quite good with the
region of poorest fit at V ∼ 18 just above the SGB.
That the theoretical and observed LFs agree so well
for stars brighter than V ∼ 18 shows that the hydro-
gen profile in the region M (r) ∼ 0.2 -- 0.4 M⊙ is given
fairly accurately by canonical models.
The values of (m − M ), age, and metallicity which
lead to the best LF fit have the virtues of also agree-
ing well with the longer Hipparcos distance scale with
their correspondingly shorter ages and with the abun-
dances obtained by Carretta & Gratton (1998). Fur-
thermore, these parameters fit in well with those ob-
tained from a new re-analysis of the location of the
LF-bump in a sample of more than 40 GGCs by Fer-
raro et al. (1999). They find that the previous discrep-
ancy in the bump location between the observations
and models is completely removed using new models
and considering the newer global metallicity scale.
The good fit to the LF-bump comes at the expense
of some mismatch at the SGB/lower-RGB transition.
To explore how varying the parameters affect the fit
we must quantify the quality of the fits. Formally we
can perform a Chi-squared test computing
χ2 =
N
Xi=1
(cid:18) log(Nobserved)i − log(Nmodel)i
log(Nobserved)i
(cid:19)2
9
For the best fit to the entire data set (13 < V < 20.5),
shown in Figure 7, there is a slightly better agreement
with the SCL97 models (χ2
V97 =
0.13).
SCL97 = 0.11 and χ2
One could conceivably prefer parameters which do
not give the best overall fit. The factors which de-
termine the luminosity of the "theoretical" LF-bump
have been discussed in Fusi Pecci et al. (1990), Chi-
effi and Gratton (1986) and more recently and exten-
sively in Cassisi & Salaris (1997). The most impor-
tant factors are the opacity in the 1 -- 2 × 106 K range
and the degree that convective elements undershoot
the neutral buoyancy point of the convective enve-
lope. Elsewhere the post-TO LF depends mostly on
the rate of fuel consumption (but see §4.3 below), and
one might expect it to be a considerably more robust
prediction of the models than the location of the LF-
bump. Given this one might tolerate some residual
error in the LF-bump location to achieve a better fit
elsewhere and, thus, exclude the bump region when
measuring the quality of the fit. If we do so for the
fits shown in Figure 7 we find χ2
SCL97 xbump = 0.01
and χ2
V97 xbump = 0.02.
In Figure 9 we explore the effect of varying the dis-
tance modulus, age and chemical composition. Since
the quality of the fit is essentially the same for SCL97
and V97, we show results only for the SCL97 mod-
els. We have found it impossible to fix parameters so
that the model fits the SGB region and the LF-bump
equally well. The best fit in the SGB region results
from a higher age of 14 Gyr. However the good fit
in the SGB region brings with it a worse fit for the
LF-bump. The fit to the LF-bump can be restored
at an age of 14 Gyr by adopting a higher metallic-
ity and (m − M ) smaller than we consider plausible
(Figure 9b). The upper end of the allowable (m − M )
range gives good fit to the LF-bump (Figure 9c) at the
expense of a slightly worse fit to the SGB region and
a small age (10 Gyr). The Chi-square values both in-
cluding and excluding the LF-bump region are given
in Table 3.
From the χ-values listed in Table 3 we conclude
that considering both the two main observables of
the LF (the LF-bump and the SGB-rise) we can find
a good agreement (with both SCL97 and V97 mod-
els) when we take age = 12, (m − M )V = 15.03 and
[M/H] ∼ −1.2 (see Figure 7). On the other hand,
if we consider that some residual error still might be
present in the theoretical prediction of the LF-bump
location, we get the best agreement between theory
10
and observations if age = 14, (m − M )V = 14.80 and
[M/H] ∼ −1.2 (see Figure 9a).
A full discussion of the age should also bring in
the isochrone fits. These are complicated both by the
color calibration and the color-temperature relations.
Probably because of this our isochrone fits using the
ages and distances determined from the LF give no
further indication of whether the age of M3 is closer
to 12 or 14 Gyr.
4.3. The Lower Red Giant Branch and Sub-
Giant Branch
The lower-RGB & SGB LF can be especially in-
teresting. This is the region in which there has been
some indication of a breakdown in the canonical mod-
els for very low metallicity clusters (Stetson 1991,
Bolte 1994) and recently confirmed for M30 (Berg-
bush 1996, Guhathakurta et al. 1998, Sandquist et al.
1999).
It is the region affected by non-canonical
assumptions like WIMP energy transport (Faulkner
& Swenson 1993, Degl'Innocenti, Weiss, & Leone
(1997)), and helium settling (Proffit & VandenBerg
1991, Straniero et al. (1997)), both of which could
reduce cluster age estimates. Rapid stellar rotation
(VandenBerg, Larsen, & de Propris 1998) and [α/Fe]
could also affect the SGB region of the LF. We illus-
trate all of these in Figure 11 (see also Figs. 2 & 3 of
Vandenberg et al. 1998). The predicted changes are
all small, and age differences can mimic these changes.
As shown in Figs. 7 & 8, the the theoretical and
observed LFs agree well in the SGB region and lower-
RGB. Indeed the fits are pretty good under a wide
range of assumptions (Figure 9). We should note that
for the excess of lower-RGB stars in M30 cited above,
the observed LF lies above the lower-RGB at essen-
tial every bin. For M3 our "bad" fits correspond to
only one or two mismatched bins. The fits can be es-
sentially perfect (Figure 9a) if we accept some error
in the bump location. Basically, there is no indica-
tion for a breakdown in the canonical models. SBSH
found a similar result for M5.
On the other hand the observational consequences
of plausible non-canonical effects are small so even
our good fit may not place very tight constraints on
their magnitude. We do not attempt to place such
constraints now because we feel that a traditional LF
analysis may not be the appropriate way to analyze
the SGB. By projecting the star count information
onto the "magnitude" axis one loses the information
Chi-square results between observations and SCL models
Table 3
13 < V < 17
17 < V < 20.5
13 < V < 20.5
[M/H] = −1.2
(m − M )V = 15.03
age = 12 Gyr
[M/H] = −1.2
(m − M )V = 14.80
age = 14 Gyr
[M/H] = −0.82
(m − M )V = 14.65
age = 14 Gyr
[M/H] = −1.3
(m − M )V = 15.20
age = 10 Gyr
0.101
0.194
0.103
0.112
0.014
0.008
0.026
0.018
0.114
0.201
0.130
0.131
contained in the color distribution of the stars. In a
future paper we will report on our efforts to study
the distribution "along" the SGB (see for example,
Bergbush & VandenBerg 1997), and provide quanti-
tative estimates as to the degree that non-canonical
assumptions can be ruled out.
4.4. Below the Main Sequence Turnoff
Below the MSTO the LF primarily probes the stel-
lar initial mass function (IMF) and cluster dynamics.
Having been acquired primarily to study post-MSTO
evolution, the data we describe in this paper reaches
only ∼ 2 mag below the turnoff which corresponds to
a mass significantly above the lower mass cutoff for
main sequence stars. Deeper LFs (based on almost
25,000 stars) exploring this region are presented in
Marconi et al. (1998) and Carretta et al. (1999c).
It is widely recognized that stars in dense systems
(like GGCs) are subject to energy exchange through
stellar encounters and rapidly evolve toward a state
of equipartition of energy (so that stars of lower mass
will have higher velocities). As a direct consequence
low mass stars should have larger average distance
from the center and their distribution will be less cen-
trally concentrated than that of higher mass stars.
This mass segregation effect is expected to produce
an observable effect on the LF if a sufficiently large
radial region of the cluster is sampled.
Such mass segregation has already been found in
GGCs surveyed by HST: NGC 6397 (King et al.
1995), 47 Tuc (Paresce et al. 1995 ), M15 (De Marchi
& Paresce 1996), NGC 6752 (Ferraro et al. 1997c). Pi-
otto et al. (1997) give an extended discussion of this
topic. Mass segregation has been found in M30 using
ground based data (Bolte 1989).
Even with our limited MS coverage, we can see ev-
idence for mass segregation in M3. In Figure 12 we
show a comparison of the LFs in three separate radial
regions: r < 20′′; 50′′ < r < 100′′; 210′′ < r < 360′′.
As above the LFs have been normalized to the total
number of stars brighter than the MSTO. The inner
regions show a significant drop of star counts over
the magnitude range V ∼ 20 − 21.5. We interpret
the observed depletion as due to the mass segrega-
tion effect. The depletion is obvious even in the small
interval of masses sampled. The effect shown in Fig-
ure 12 cannot due to residual incompleteness of the
samples. In order to completely remove the effect we
11
would have to assume a level of completeness which
is far smaller than the estimated value. For exam-
ple in the innermost sample r < 20′′ (which is almost
completely contained in the PC) the measured com-
pleteness must be almost a factor 2 in error (dropping
from 0.8 down to 0.45) in order to remove the effect.
Is the observed decrease in the star counts con-
sistent with the theoretical expectation? In order to
derive the relative distribution of stars of different
mass at different distances from the cluster center we
have used the formula given by King et al. (1995) fol-
lowing the procedure as explained in DeMarchi and
Paresce (1996). The faint end of the LF should be
depleted in the core region with respect to the outer
regions by the factor f = (ρ0/ρext)m2/m1−1 where ρ0
and ρext are the stellar densities in the core and in the
external region respectively, and m2 and m1 are the
stellar masses at the faint extreme and TO level of
the LF, respectively. We use the tables of SCL97 and
assume (m − M )V = 15.0 to relate observed magni-
tudes to mass -- the MSTO at V = 19.1 corresponds
to MV ∼ 4 and a mass of ∼ 0.84 M⊙, V = 20.5 cor-
responds to MV = 5.5 and a mass of ∼ 0.73 M⊙,
and V = 21.9 corresponds MV = 6.95 and a mass of
∼ 0.61 M⊙, respectively.
The spatial densities can be approximated by the
In the case of M3
Eqn. 5 in King et al. (1995).
the external region lies at rext ∼ 290′′, and taking
rc ∼ 30′′ (Djorgovski 1993), we estimate that the ra-
tio (ρ0/ρext) is ∼ 900. Adopting the mass values ob-
tained above, the density of low-mass stars should be
a factor 9000.13 ∼ 2.4 less than that in the central.
This value is compatible with the observed depletion
which is 975/548 ∼ 1.8.
We can also explore the stellar initial mass function
(IMF) using our data. The IMF is normally approx-
imated by a power-law mass spectrum of the form
φ(M )dM = M −(1+x)dM where φ(M )dM is the num-
ber of stars in the mass range M to M + dM . The
observed slope of the IMF measured in terms of x
should vary with radius in a way consistent with a
underlying global slope. To verify this we use the dy-
namical models of Pryor, Smith, & McClure (1986).
We compute x using Eq. 3 of Bolte (1989). The two
magnitude intervals used are 18.55 < V < 20.55
and 20.55 < V < 21.95. Figure 13 shows the ob-
served x plotted as a function of radius. We see
that x(observed) varies r in a way consistent with the
x(global) ∼ 0. This is close to the value we inferred
from our total sample.
5. Summary
We have obtained a V -band luminousity function
for the globular cluster M3 using a combination of
ground based photographic and CCD observations
and HST observations of the cluster center. The sam-
ple is the largest, most complete LF every obtained
for a globular cluster. It is free from the problems of
photometric blends which plague ground-based LFs.
In the crucial MSTO/SGB region our sample is more
than three times larger than previous LFs.
Following a traditional LF function analysis we find
no surprises -- canonical stellar models seem to fit the
data well. In particular,
• The location of the RGB LF-bump agrees well
new theoretical models incorporating the OPAL
opacities and enhancements of the α-elements.
There is no indication of significant "under-
shooting" by the convective envelope. The best
overall fit is achieved with age = 12, (m −
M )V = 15.03 and global metallicity [M/H] ∼
−1.2 (equivalent to [Fe/H] = −1.34 and [O/Fe] =
0.2). On the other hand, if we consider that the
physics determining the bump location might
have more uncertainty than that determing SGB
evolution, we might tolerate a worse fit for the
bump to achieve a better fit for the SGB. In
that case we find age = 14, (m − M )V = 14.80
and [M/H] ∼ −1.2
• The slope of the integrated LF changes across
the LF bump as predicted by the models. This
implies that the H-profile (and thus fuel con-
sumed in the earlier stages) is given reasonably
accurately by standard models.
• The LF is well fit in the SGB region. There is
no indication of a problem with the canonical
models as had been earlier found for low metal-
licity clusters.
• Below the TO we find evidence for a "flat" IMF
(x ≈ 0) and evidence for mass segregation which
is well fit by King models.
The two best observed GGC LFs, that of M3 pre-
sented here and M5 by SBSH, seem to be a vindi-
cation of the quality of theoretical models.
In con-
trast, the anomalous LF of M30 first found by Bolte
(1994) has inconveniently persisted in later studies
(Bergbush 1996, Guhathakurta et al. 1998, Sandquist
12
et al. 1999). Why should the theory work so well in
some cases and fail in others? One can explore possi-
ble deficiencies in the models as in VandenBerg et al.
(1998). However, we feel that the ultimate resolution
of the problem lies in the aquisition of more high qual-
ity LFs to determine how common anomalous LFs are
and what cluster parameters they correlate with.
We thank Don VandenBerg for providing informa-
tion prior to publication and for very helpful email
discussions. RTR & BD are supported in part by
NASA Long Term Space Astrophysics Grant NAG 5-
6403 and STScI/NASA Grants GO-5969, GO-6804.
This research was partially supported by the Agenzia
Spaziale Italiana (ASI) and by the MURST as part of
the project Stellar Evolution. FRF acknowledges the
ESO Visiting Program for its hospitality.
REFERENCES
Bell, R. A., Gustafsson, B., Nordh, H. L., & Olofsson,
S. G. 1976, A&A, 46, 391
Bergbush, P. A. 1996, AJ, 112, 1061
Bergbush, P. A., & VandenBerg, D. A. 1992, ApJS,
81, 163
Bergbush, P. A., & VandenBerg, D. A. 1997, AJ, 114,
2604
Bergbush, P. A., & VandenBerg, D. A. 1999, in prepa-
ration
Bolte, M. 1989, ApJ, 341, 168
Bolte, M. 1994, ApJ, 431, 223
Carretta, E., & Gratton, R. G. 1997, A&AS, 121, 95
Cassisi, S., & Salaris, M. 1997, MNRAS, 285, 993
Castellani, V., Ciacio, F., Degl'Innocenti, S., &
Fiorentini, G. 1997, A&A, 322, 801
Chieffi, A. & Gratton, R. G. 1986, Mem. Soc. Astron.
It., 57, 395
Cudworth, K. M. 1979, AJ, 84, 1312
Degl'Innocenti, S., Weiss, A., & Leone, L. 1997, A&A,
319, 487
De Marchi G., & Paresce, F. 1996, ApJ, 467, 658
De Santis, R. 1996, A&A, 306, 755
Djorgovski, S. G. 1993 in ASP Conf. Ser. 50, Struc-
ture and Dynamics of Globular Clusters, ed. S. G.
Djorgovski & G. Meylan (Sanfrancisco: ASP), 373
Faulkner, J., & Swenson, F. J. 1993, ApJ, 411, 200
Fernley, J., Carney, B. W., Skillen, I., Cacciari, C., &
Janes, K. 1998, MNRAS, 293, L61
Ferraro, F. R., Fusi Pecci, F., Cacciari, C., Corsi,
C.E., Buonanno, R., Fahlman, G. G., & Richer,
H.B., 1993, AJ, 106, 2324
Ferraro, F. R., Paltrinieri, B., Fusi Pecci, F., Cacciari,
C., Dorman, B., Rood, R.T., Buonanno, R., Corsi,
C., Burgarella, D., & Laget, M. 1997a, A&A, 324,
915
Ferraro, F. R., Carretta, E., Corsi, C., Fusi Pecci,
F., Cacciari, C., Buonanno, R., Paltrinieri, B., &
Hamilton, D., 1997b, A&A, 320, 757
Buonanno, R., Buscema, G., Corsi, C. E., Ferraro, I.,
& Iannicola, G. 1983, A&A, 126, 278
Ferraro, F. R., Carretta, E., Bragaglia, A., Renzini,
A., & Ortolani, S. 1997c, MNRAS, 286, 1012
Buonanno, R., Corsi, C. E., Buzzoni, A., Cacciari, C.,
Ferraro, F. R., & Fusi Pecci, F. 1994, A&A, 290,
69
Ferraro, F. R., Messineo, M., Fusi Pecci, F., De Palo,
A., Straniero, O., Chieffi, A., & Limongi, M., 1999,
AJ, submitted
Carretta, E., Gratton, R. G., Clementini, G., & Fusi
Pecci, F. 1999a, astro-ph 9902086
Fusi Pecci, F., Ferraro, F. R., Crocker, D. A., Rood,
R. T., & Buonanno, R. 1990 A&A, 238, 95
Carretta, E., Gratton, R. G., Sneden, C., & Bra-
gaglia, A., 1999b, in Galaxy Evolution: Connect-
ing the Distant Universe with the Local Fossil
Record, Observatoire de Paris-Meudon, September
1998, in press (astro-ph/9812095)
Carretta, E., et al, 1999c, in preparation
Gratton, R. G., Fusi Pecci, F., Carretta, E., Clemen-
tini, G., Corsi, C., & Lattanzi, M. 1997, ApJ, 491,
749
Guhathakurta, P., Webster, Z. T., Yanney, B.,
Schneider, D. P., & Bahcall, J. N. 1998, AJ, 116,
1757
13
Holweger, H., & Muller, E. A. 1974, Solar Physics,
39, 19
Holtzmann, J. A., Burrows, C. J., Casertano, S.,
Hester, J. J., Trauger, J. T., Watson, A. M., &
Worthey, G., 1995, PASP, 107, 1065
Iben, I. 1968, Nature, 220, 143
Laget, M., Ferraro, F. R., Paltrinieri, B., & Fusi Pecci,
F. 1998, A&A, 332, 93
King, I. R., Da Costa, G. S. & Demarque, P. 1985,
ApJ, 299, 674
King, I. R., Sosin, C., & Cool, A. 1995, ApJL, 452,
13
Kraft, R. P., Sneden, C., Langer, G. E., & Prosser,
C. F. 1992, AJ, 104, 645
Rood, R. T., & Crocker, D. A. 1985, in The Pro-
duction and Distribution of C, N, O Elements, ed.
J. Danziger, F. Matteucci, & K. Ajar (Garching:
ESO), 61
Sandquist, E. L., Bolte, M., Stetson, P. B., & Hesser,
J. E. 1996, ApJ, 470, 910
Sandquist, E. L., Bolte, M., Langer, G. E., Hesser, J.
E., & Mendes de Oliveira, C. 1999, ApJ, in press,
astro-ph/9810259
Salaris, M., Chieffi, A., & Straniero, O. 1993, ApJ,
414, 580
Stetson, P. B. 1991, in ASP Conf. Ser. 13, The Forma-
tion and Evolution of Star Clusters, ed. K. Janes
(Sanfrancisco: ASP), 88
Straniero, O., Chieffi, A., & Limongi, M. 1997, ApJ,
Kraft, R. P., Sneden, C., Langer, G. E., & Shetrone,
490, 425
M. D. 1993, AJ, 106, 1190
Krist, J. 1994, Tiny Tim User's Manual Version 4.0
Kurucz, R. L. 1993, CD-ROM 13 and CD-ROM 18
Marconi, G., Buonanno, R., Carretta, E., Ferraro, F.
R., Fusi Pecci, F., Montegriffo, P., De Marchi, G.,
Paresce, F., & Laget, M. 1998, MNRAS, 293, 479
Piotto, G., Cool, A., & King, I. R. 1997, AJ, 113,
1345
Paresce, F., De Marchi, G., & Jedrzejewski, R. 1995,
ApJL, 442, 57
Pont, F., Mayor, M., Turon, C., & VandenBerg, D.
A. 1998, A&A, 329, 87
Proffit, C. R., & VandenBerg, D. A. 1991, ApJS, 74,
473
Pryor, C., Smith, G. M., & McClure, R. D. 1986, AJ,
92, 1358
Reid, I. N. 1997, AJ, 114, 161
Reid, I. N. 1998, AJ, 115, 204
Renzini, A., & Buzzoni, A. 1986, in Spectral Evolu-
tion of Galaxies, ed. C. Chiosi, A. Renzini (Dor-
drect: Reidel), 135
Renzini, A., & Fusi Pecci, F., 1988, ARA&A, 26, 199
Rogers, F. J., & Iglesias, C. A. 1992, ApJ, 401, 361
Trager, S., Djorgovski, S., & King, I. 1993, in Struc-
ture and Dynamics of Globular Clusters: ASPCS
vol. 50, eds. S Djorgowski & G. Meylan (San Fran-
cisco: ASP), 347
Vandenberg, D. A., Larson, A. M., & De Propris, R.
1998, PASP, 110, 98
This 2-column preprint was prepared with the AAS LATEX
macros v4.0.
14
6. FIGURE CAPTIONS
360′′.
Fig. 1. -- The V , V − I CMD for the HST global
sample. The sample contains more than 37,000 stars.
Fig. 13. -- Comparison between the X value (com-
puted as in Bolte 1989) and the dynamical models
from Pryor et al. 1986.
Fig. 2. -- Map of the WFPC2 field of view The cir-
cle (at r ∼ 50′′) delineates the INNER and OUTER
regions.
Fig. 3. -- The smoothed interpolating curves of com-
pleteness as a function of V magnitude.
Fig. 4. -- The zoomed LF with and without com-
pleteness corrections. The Main-Sequence-Turn-Off
point is labeled.
Fig. 5. -- CMDs of stars selected for the LF in each
of the three adopted samples: PHOTO, CCD, HST,
respectively.
Fig. 6. -- Differential Luminosity Function of the
global sample (PHOTO+CCD+HST). The names we
use to refer to various parts of the LF are indicated.
7. -- Differential
Fig.
luminosity function of the
global sample compared with theoretical models from
SCL97 (panel a) and V97 (panel b). The chosen val-
ues for distance modulus, age and chemical composi-
tion are indicated.
Fig. 8. -- Integrated luminosity function of the global
sample compared with theoretical models by SCL97.
Fig. 9. -- The differential LF of the global sample
is compared with SCL97 theoretical models varying
metallicity, distance modulus and age.
Fig. 10. -- The LF-bump region of the integrated
(panel a) and differential (panel b) LF: The position
of the LF-bump is indicated. The solid line is the ob-
served LF. The dashed line is based on SCL97 models.
The short dashed lines in (panel a) are straight line
fits to the LF above and below the LF-bump and are
drawn to aid the eye in seeing the change in slope.
11. -- The dependence of the SGB LF on
Fig.
WIMPs, He diffusion, [α/F e].
Fig. 12. -- A comparison of the LFs in three separate
radial regions: r < 20′′; 50′′ < r < 100′′; 210′′ < r <
15
V
HB
15.57
15.62
15.63
15.64
15.64
15.62
15.63
15.64
15.66
15.71
15.76
15.83
15.91
15.99
16.10
16.21
16.37
16.54
16.65
V − I
0.775
0.731
0.684
0.631
0.585
0.272
0.219
0.184
0.146
0.105
0.079
0.043
0.008
-0.021
-0.050
-0.069
-0.080
-0.088
-0.091
V
RGB
12.44
12.68
12.90
13.10
13.30
13.55
13.79
14.05
14.28
14.56
14.82
15.09
15.36
15.64
15.89
16.21
16.52
16.70
16.94
17.20
17.36
17.63
18.05
Table 4
Mean ridge line for the HST sample.
V − I
V
V − I
1.589
1.473
1.403
1.347
1.294
1.241
1.193
1.150
1.115
1.077
1.045
1.017
0.989
0.964
0.939
0.911
0.890
0.879
0.867
0.856
0.849
0.836
0.815
18.16
18.26
18.33
18.37
18.42
18.47
18.52
18.57
18.62
18.70
18.78
18.86
18.98
19.10
19.18
19.26
19.38
19.78
20.10
20.30
20.50
20.98
21.34
0.803
0.787
0.764
0.746
0.724
0.696
0.668
0.647
0.629
0.616
0.608
0.600
0.593
0.592
0.593
0.595
0.597
0.624
0.654
0.672
0.693
0.756
0.805
16
Luminosity function of the global sample (PHOTO+CCD+HST).
Table 5
V
12.70
12.90
13.10
13.30
13.50
13.70
13.90
14.10
14.30
14.50
14.70
14.90
15.10
15.30
15.50
15.70
15.90
16.10
16.30
16.50
16.70
16.90
17.10
17.30
Log(Ni)
σNi
0.70
0.30
0.78
0.85
0.95
0.85
0.90
0.90
1.11
1.08
1.32
1.15
1.34
1.38
1.76
1.60
1.51
1.65
1.70
1.79
1.85
1.95
2.04
2.06
0.45
0.26
0.45
0.41
0.52
0.50
0.50
0.47
0.61
0.55
0.67
0.58
0.77
0.72
0.92
0.87
0.75
0.83
0.93
0.96
1.00
1.06
1.09
1.11
V
17.50
17.70
17.90
18.10
18.30
18.50
18.70
18.90
19.10
19.30
19.50
19.70
19.90
20.10
20.30
20.50
20.70
20.90
21.10
21.30
21.50
21.70
21.90
Log(Ni)
σNi
2.05
2.22
2.26
2.46
2.72
2.97
3.06
3.18
3.25
3.32
3.37
3.42
3.45
3.49
3.53
3.56
3.56
3.59
3.58
3.58
3.55
3.57
3.66
1.10
1.20
1.23
1.35
1.52
1.68
1.75
1.83
1.88
1.94
1.98
2.01
2.04
2.08
2.12
2.14
2.16
2.19
2.21
2.24
2.27
2.35
2.48
17
Chi-square results between observations and SCL models
Table 6
13 < V < 17
17 < V < 20.5
13 < V < 20.5
[M/H] = −1.2
(m − M )V = 15.03
age = 12 Gyr
[M/H] = −1.2
(m − M )V = 14.80
age = 14 Gyr
[M/H] = −0.82
(m − M )V = 14.65
age = 14 Gyr
[M/H] = −1.3
(m − M )V = 15.20
age = 10 Gyr
0.101
0.194
0.103
0.112
0.014
0.008
0.026
0.018
0.114
0.201
0.130
0.131
18
This figure "f5.gif" is available in "gif"(cid:10) format from:
http://arxiv.org/ps/astro-ph/9905193v1
X=2.0
X=1.35
X=0.67
X=0.0
|
astro-ph/9906229 | 1 | 9906 | 1999-06-14T09:48:56 | The Galactic contribution to high latitude diffuse gamma-ray emission | [
"astro-ph"
] | Recent evidence for a large Galactic halo, based on cosmic-ray radioactive nuclei, implies a significant contribution from inverse Compton emission at high Galactic latitudes. We present predictions for the expected intensity distribution, and show that the EGRET gamma-ray latitude distribution is well reproduced from the plane to the poles. We show that the Galactic component at high latitudes may be comparable to the extragalactic emission in some energy ranges. | astro-ph | astro-ph |
Proc. 26th ICRC (Salt Lake City, 1999), OG2.4.03
The Galactic contribution to high latitude diffuse γ-ray emission
Andrew W. Strong1, Igor V. Moskalenko1,2, and Olaf Reimer1
1MPI fur extraterrestrische Physik, D -- 85740 Garching, Germany
2Institute for Nuclear Physics, Moscow State University, 119 899 Moscow, Russia
Abstract
Recent evidence for a large Galactic halo, based on cosmic-ray radioactive nuclei, implies a significant con-
tribution from inverse Compton emission at high Galactic latitudes. We present predictions for the expected
intensity distribution, and show that the EGRET gamma-ray latitude distribution is well reproduced from the
plane to the poles. We show that the Galactic component at high latitudes may be comparable to the extra-
galactic emission in some energy ranges.
1 Introduction:
The origin of the truly extragalactic γ-ray background is still unknown. The models discussed range from
the primordial black hole evaporation (Page & Hawking 1976) and annihilation of exotic particles in the
early Universe (Cline & Gao 1992) to the contribution of unresolved discrete sources such as active galaxies
(Sreekumar et al. 1998), while the spectrum of the extragalactic emission itself is uncertain. The latter can
be addressed only by the accurate study of the Galactic diffuse emission at high Galactic latitudes. Moreover
there is growing evidence for a large halo contribution to the γ-ray background. An indication for a γ-ray halo
was also found by Dixon et al. (1998) from analysis of EGRET data. Studies of 10Be (Strong & Moskalenko
1998) gave the range zh = 4 − 12 kpc for nucleons, Webber & Soutoul (1998) find zh = 2 − 4 kpc from 10Be
and 26Al data, Ptuskin & Soutoul (1998) find zh = 4.9+4
−2 kpc.
Gamma rays provide a tracer of the electron halo via inverse Compton (IC) emission. A study of the
Galactic emission requires a systematic approach: computation of a realistic interstellar radiation field and
self-consistent calculation of the electron spectrum in 3D. We use our GALPROP model1, which has been
shown to be consistent with many kinds of data related to cosmic ray origin and propagation, to calculate the
Galactic contribution to the high latitude diffuse γ-ray emission (Strong, Moskalenko, & Reimer 1999). The
models have been described in full detail in Strong & Moskalenko (1998), for a recent review of our results
see Strong & Moskalenko (1999).
2 Description of the models:
The models are three dimensional with cylindrical symmetry in the Galaxy. For a given halo size the
diffusion coefficient (as a function of momentum) and the reacceleration parameters are determined by the
Boron-to-Carbon ratio; the momentum-space diffusion coefficient is related to the spatial coefficient (Berezin-
skii et al. 1990). The injection spectrum of particles is assumed to be a power law in momentum, if necessary
with a break. The magnetic field is adjusted to match the 408 MHz synchrotron longitude and latitude distri-
butions. The interstellar hydrogen and He distribution uses HI and CO surveys and information on the ionized
component. Energy losses for particles by ionization, Coulomb interactions, bremsstrahlung, inverse Comp-
ton, and synchrotron are included. The distribution of cosmic-ray sources is adjusted (Strong & Moskalenko
1998) to match the cosmic-ray distribution obtained from EGRET γ-ray data (Strong & Mattox 1996). The
interstellar radiation field is based on stellar population models and COBE results, plus the cosmic microwave
background (Strong, Moskalenko, & Reimer 1999). Inverse Compton scattering is treated as in Moskalenko
& Strong (1999) including the effect of the anisotropy of the ISRF, and gas related γ-ray intensities (π0-decay
and bremsstrahlung) are computed using the column densities of HI and H2 based on 21-cm and CO surveys.
1 For
interested users our model and data sets are available in the public domain on the World Wide Web,
http://www.gamma.mpe-garching.mpg.de/∼aws/aws.html
0.5<l< 30.0 , 330.0<l<359.0
-5.0<b< 5.0
70-100MeV
1.0<l<180.0/181.0<l<359.0
Figure 1: Gamma-ray energy spectrum of the in-
ner Galaxy (300◦ ≤ l ≤ 30◦, b ≤ 5◦) compared
with our model calculations (electron injection in-
dex −1.8, and modified nucleon spectrum). Curves
show the contribution of IC, bremsstrahlung, and π0-
decay, and the total. Data: EGRET (Strong & Mat-
tox 1996), COMPTEL (Strong et al. 1999), OSSE
(l = 0, 25◦: Kinzer et al. 1999).
Figure 2: High latitude distribution (enlarged) of
70 -- 100 MeV γ-rays from the EGRET compared to
our model calculation. Separate components:
IC
(dashes), bremsstrahlung (thin histogram), π0-decay
(thick histogram), horizonal line:
isotropic back-
ground. EGRET data (point sources removed): dot-
ted line.
3 Inner Galaxy:
Recent results from both COMPTEL and EGRET indicate that IC scattering is a more important contributor
to the diffuse emission that previously believed. The puzzling excess in the EGRET data > 1 GeV relative
to that expected for π0-decay has been suggested to originate in IC scattering from a hard interstellar electron
spectrum (e.g., Pohl & Esposito 1998). Our combined approach allows us to test this hypothesis (Strong,
Moskalenko, & Reimer 1999).
Our first concern was to reproduce the γ-ray spectrum of the inner Galaxy. This is possible by invoking
a hard electron spectrum with injection spectral index -- 1.8, which after propagation (with reacceleration)
provides consistency with radio synchrotron data. Following Pohl & Esposito (1998), for this model we do
not require consistency with the locally measured electron spectrum above 10 GeV because the rapid energy
losses cause a clumpy distribution so that this is not necessarily representative of the interstellar average. For
this case, the interstellar electron spectrum deviates strongly from that locally measured.
Further improvement can be obtained by allowing some freedom in the nucleon spectrum at low energies.
Some freedom is allowed since solar modulation affects direct measurements of nucleons below 20 GeV,
and the locally measured nucleon spectrum may not necessarily be representative of the average on Galactic
scales either in spectrum or intensity due to details of Galactic structure (e.g. spiral arms). By introducing
some flattening of the nucleon spectrum below 20 GeV, a small steepening above 20 GeV, and a suitable
normalization, an improved match to the inner Galaxy EGRET spectrum is indeed possible (Fig. 1).
The modified nucleon spectrum must be checked against the stringent constraints on the interstellar spec-
trum provided by antiprotons and positrons. (Such tests sample the Galactic-scale properties of CR p and He
0.5<l<179.0 , 180.5<l<359.0
70.0<b< 89.0
0.5<l<179.0 , 180.5<l<359.0
70.0<b< 89.0
Figure 3: Energy spectrum of γ-rays from high Galactic latitudes (b ≥ 70◦, all longitudes) for zh = 4
kpc (left) and zh = 10 kpc (right). Shaded areas: EGRET total intensity from Cycle 1 -- 4 data. COMPTEL
data: high-latitude total intensity (open boxes: Bloemen et al. 1999, diamonds: Kappadath 1998, crosses:
Weidenspointner et al. 1999).
rather than just the local region, independent of fluctuations due to local primary CR sources.) As expected
the ¯p and e+ predictions are higher than for the conventional model (with nucleon spectrum matching the local
measurements) but still within the observational limits (Strong, Moskalenko, & Reimer 1999).
This is our best model so far. Further tests against the γ-ray longitude and latitude profiles at all the EGRET
energies show a good overall agreement (an example is shown in Fig. 2).
In order to reproduce the low-energy (< 30 MeV) γ-ray emission via diffuse processes it is necessary
to invoke an upturn of the electron spectrum below about 200 MeV to compensate the increasing ionization
losses. (A steep slope continuing to higher energies would violate the synchrotron constraints on the spectral
index.) However, the adoption of such a steep low-energy electron spectrum has problems associated with
the very large power input to the interstellar medium (Skibo et al. 1997), and is ad hoc with no independent
supporting evidence. Moreover the OSSE-GINGA γ-ray spectrum is steeper than E−2 below 500 keV (Kinzer
et al. 1999) which would require an even steeper electron injection spectrum than adopted here. It is more
natural to consider that the COMPTEL excess is just a continuation of the same component producing the
OSSE-GINGA spectrum. Most probably therefore the excess emission at low energies is produced by a
population of sources such as supernova remnants, as has been proposed for the diffuse hard X-ray emission
from the plane observed by RXTE (Valinia et al. 1998), or X-ray transients in their low state as suggested for
the OSSE diffuse hard X-rays (Lebrun et al. 1999).
4 High latitude γ-rays and the size of the electron halo:
We use our model for calculation of the Galactic contribution to the high latitude diffuse γ-ray emission.
The high latitude γ-ray intensity increases with halo size due to IC emission (though much less than linearly
due to electron energy losses), which allows us to put an upper limit on the halo size. Fig. 3 shows the γ-ray
spectrum towards the Galactic poles for zh = 4 kpc, and 10 kpc. zh = 10 kpc is possible although the latitude
profiles around 100 MeV are then very broad and at the limit of consistency with EGRET data. Further the
isotropic component would have to approach zero above 300 MeV, so that this halo size can be considered an
upper limit.
If the halo size is 4 -- 10 kpc as we argue, the contribution of Galactic emission to the total at high latitudes
is larger than previously considered likely and has consequences for the derivation of the diffuse extragalac-
tic emission (e.g., Sreekumar et al. 1998). An evaluation of the impact of our models on estimates of the
extragalactic spectrum is beyond the scope of the present work.
5 Conclusions:
The large electron/IC halo suggested here reproduces well the latitude variation of γ-ray emission from
the plane to the poles, which can be taken as support for the halo size deduced from independent studies of
cosmic-ray composition. Halo sizes in the range zh = 4 − 10 kpc are favoured by both analyses.
References
Berezinskii, V.S., et al. 1990, Astrophysics of Cosmic Rays (Amsterdam: North Holland)
Bloemen, H., et al. 1999, Astroph. Lett. Comm. (3rd INTEGRAL Workshop), in press
Cline, D.B., & Gao, Y.-T. 1992, A&A 256, 351
Dixon, D.D., et al. 1998, New Astronomy 3, 539
Kappadath, S.C. 1998, PhD Thesis, University of New Hampshire, USA
Kinzer, R.L., Purcell, W.R., & Kurfess, J.D. 1999, ApJ 515, 215
Lebrun, F., et al. 1999, Astroph. Lett. Comm. (3rd INTEGRAL Workshop), in press
Moskalenko, I.V., & Strong, A.W. 1999, submitted (astro -- ph/9811296)
Page, D.N., & Hawking, S.W. 1976, ApJ 206, 1
Pohl, M., & Esposito, J.A. 1998, ApJ 507, 327
Ptuskin, V.S., & Soutoul, A. 1998, A&A 337, 859
Skibo, J.G., et al. 1997, ApJ 483, L95
Sreekumar, P., et al. 1998, ApJ 494, 523
Strong, A.W., & Mattox, J.R. 1996, A&A 308, L21
Strong, A.W., & Moskalenko, I.V. 1998, ApJ 509, 212
Strong, A.W., & Moskalenko, I.V. 1999, in ASP Conf. Ser. 171, 154
Strong, A.W., Moskalenko, I.V., & Reimer, O. 1999, submitted (astro -- ph/9811284)
Strong, A.W., et al. 1998, Astroph. Lett. Comm. (3rd INTEGRAL Workshop), in press
Valinia, A., & Marshall, F.E. 1998, ApJ 505, 134
Webber, W.R., & Soutoul, A. 1998, ApJ 506, 335
Weidenspointner, G., et al. 1999, Astroph. Lett. Comm. (3rd INTEGRAL Workshop), in press
|
astro-ph/0312452 | 1 | 0312 | 2003-12-17T15:54:02 | Strangeness in Neutron Stars | [
"astro-ph",
"nucl-th"
] | We discuss the role of strangeness on the internal constitution and structural properties of neutron stars. In particular, we report on recent calculations of hyperon star properties derived from microscopic equations of state for hyperonic matter. Next, we discuss the possibility of having a strange quark matter core in a neutron star, or the possible existence of strange quark matter stars, the so-called strange stars. | astro-ph | astro-ph |
Strangeness in Neutron Stars
Ignazio Bombacia
aDipartimento di Fisica "Enrico Fermi", Universit`a degli Studi di Pisa,
and INFN sezione di Pisa, via Buonarroti, 2, I-56127, Pisa, Italy
We discuss the role of strangeness on the internal constitution and structural properties
of neutron stars. In particular, we report on recent calculations of hyperon star properties
derived from microscopic equations of state for hyperonic matter. Next, we discuss the
possibility of having a strange quark matter core in a neutron star, or the possible existence
of strange quark matter stars, the so-called strange stars.
1. INTRODUCTION
The true nature and the internal constitutions of the ultra-dense compact stars known
as neutron stars is one of the most fascinating enigma in modern astrophysics [ 1, 2, 3].
Different models for the equation of state (EOS) of dense matter predict the neutron star
maximum mass Mmax to be in the range 1.4 -- 2.2 M⊙ (M⊙ ≃ 2 × 1033g is the mass of
the Sun), and a corresponding central density in the range of 4 -- 8 times the normal
saturation density (ρ0 ∼ 2.8 × 1014g/cm3) of nuclear matter. Thus neutron stars are the
most likely sites in the universe in which strangeness-bearing matter with a strangeness to
baryon ratio fS = −S/B ∼ 1 may exist. Strangeness in neutron stars is expected both in
a confined form (hyperons and or kaons), or in a deconfined form (strange quark matter).
Accordingly different types of "neutron stars" are expected theoretically, as schematically
summariezed in Fig. 1.
2. NEUTRON STARS OR HYPERON STARS?
In a conservative and oversimplified picture the core of a neutron star is modeled as
a uniform fluid of neutron rich nuclear matter in equilibrium with respect to the weak
interactions (β -- stable nuclear matter). The presence of hyperons in neutron stars was
first proposed in 1960 by Ambartsumyan and Saakyan [ 4], and since then it has been
investigated by many authors. The reason why hyperons are expected in the high dense
core of a neutron star is very simple, and it is mainly a consequence of the fermionic
nature of nucleons, which makes the nucleon chemical potentials a very rapidly increasing
function of density. As soon as the chemical potential of neutrons becomes sufficiently
large (see Fig. 2), the most energetic neutrons (i.e. those on the Fermi surface) can decay
via the weak interactions into Λ hyperons and form a Fermi sea of this new hadronic
species with µΛ = µn. The Σ− can be produced via the weak process e− + n → Σ− + νe
2
quark−hybrid
star
hyperon
star
o
c
r
e
p
su
N+e
N+e+n
n,p,e, µ
n d c t
u
ni
n,p,e,µ
g
−π
−
K
u,d,s
quarks
H
Σ,Λ,Ξ,∆
absolutely stable
strange quark
matter
u d s
µ
m
s
traditional neutron star
n s
u
p
e
rfluid
p
r
o
t
o
n
s
crust
neutron star with
pion condensate
Fe
6
3
g/cm
11
14
3
3
g/cm
g/cm
10
10
10
strange star
nucleon star
R ~ 10 km
M ~ 1.4 M
Figure 1. Schematic cross section of a neutron star according to different possibilities for
the stellar constituents (adapted from ref. [ 2]).
when the Σ− chemical potential fulfill the condition1 µΣ− = µn + µe. As we can see
from the results depicted in Fig. 2, hyperons appear at a relatively moderate density of
about 2 times the normal saturation density (n0 = 0.16 fm−3) of nuclear matter. Notice
that the Σ− hyperon appears at a lower density than the Λ, even though the Σ− is more
massive than the Λ. This is due to the contribution of the electron chemical potential
µe to the threshold condition for the Σ− (i.e. MΣ− = µn + µe, for free hyperons) and
to the fact that µe in dense matter is large and can compensate for the mass difference
MΣ− − MΛ = 81.76 MeV.
In Fig. 3, we show the profile of a such an hyperon star [ 6]. As we see the hyperonic
matter inner core of the star extend for about 8 km. This radius has to be compared with
the total stellar radius R ∼ 11 km, and with the thickness of the nuclear matter layer
(outer core) which is about 2 km. Thus neutron stars are "giant hypernuclei" [ 7] under
the influence of gravity and strong interactions.
The influence of hyperons on neutron stars properties has been investigated using dif-
ferent approaches to determine the EOS of hyperonic matter. One of the most popular
approaches, to solve this problem, is the relativistic mean field model [ 7, 8]. Some of the
1except from the very initial stage soon after neutron star birth, neutrinos freely escape the star and thus
the neutrino chemical potentials have not to be considered in the chemical equilibrium equations.
3
1300
Interacting hyperons
1250
MΣ−
MΛ
µ
n+ µ
e
µ
n
Free hyperons
µΣ−
µΛ
µ
n+ µ
e
µ
n
1200
1150
1100
1050
1000
950
0.4
0.2
1.2
Baryon number density [fm-3]
0.6
0.8
1
0
0.2
0.4
0.6
0.8
Baryon number density [fm-3]
1
1.2
1300
1250
1200
1150
1100
1050
1000
]
V
e
M
[
s
l
a
i
t
n
e
t
o
p
l
a
c
i
m
e
h
C
950
0
Figure 2. Chemical potentials in β-stable hyperonic matter. Left panel: free hyperons.
Right panel: interacting hyperons (NSC97e interaction). Adaped from ref. [ 5].
parametrizations of the lagrangian of the theory have tried to reconcile measurd values of
neutron star masses with the binding energy of the Λ particle in hypernuclei [ 9, 10]. A dif-
ferent approach is based on the use of local effective potentials to describe the in-medium
baryon-baryon (BB) interaction [ 11]. This method mimic and generalize to the case of
hyperonic matter the one based on the Skyrme nuclear interaction in the case of nuclear
matter. Here we will report on some recent results based on a third approach, which starts
from the basic BB interaction and solve the many-body problen to get the EOS for hyper-
onic matter. This method is based on an extension of the Brueckener-Bethe-Goldstone
(BBG) theory to include hyperonic degrees of freedom [ 12, 13, 14, 15, 6]. In particular,
the study of ref. [ 6] focus on the properties of a newborn neutron star, and explore the
consequences of neutrino trapping in dense matter on the structural properties and on
the early evolution of neutron stars [ 16].
In Fig. 4, we show the EOS for β-stable dense matter (i.e. in equilibrium with respect
[ 15] using the Nijmegen soft-core
to the weak interactions) obtained by Vidana et al.
[ 17] to describe the hyperon-nucleon (YN) and hyperon-
potential (NSC97e) of ref.
hyperon (YY) interaction within the Brueckner-Hartree-Fock (BHF) approximation of
the extended BBG theory. In ref. [ 15], the pure nucleonic contribution to the EOS has
been included using a parametrization of the Akmal-Pandharipande EOS [ 18], where a
semi-phenomenological three-nucleon (NNN) interaction of the Urbana type is added to
the nuclear hamiltonian to reproduce the empirical saturation point of nuclear matter.
As expected, the presence of hyperons makes the EOS much softer with respect to
the pure nucleonic case. The softening of the EOS caused by the presence of hyperons
has important consequences on many macroscopic properties of the star: the maximum
stellar mass is reduced by ∆Mmax ∼ 0.5 -- 0.8 M⊙, and the corresponding central density
is increased. Also, hyperon stars are more compact (i.e. they have a smaller radius) with
4
100
10-1
s
n
o
i
t
c
a
r
f
e
l
c
i
t
r
a
P
10-2
10-3
RY
RH R
n
p
Σ−
e-
µ−
Λ
10-4
0
1
2
4
3
10
Radial coordinate r [km]
8
9
5
6
7
11 12
13
Figure 3. The internal composition of a neutron star with hyperonic matter core. The
stellar baryonic mass is MB = 1.34 M⊙. RY is the radius of the hyperonic core. The
nuclear matter layer extend between RY and RH and has a thickness of about 2 km. The
stellar crust extend between RH and R and is about 1 km thick. (Adapted from Vidana
et al. [ 6]).
respect to pure nucleonic neutron stars. This is illustrated in Fig. 5, where we show the
mass radius-relation for traditional neutron stars and for hyperon stars obtained with the
microscopic EOS of ref. [ 15] (left panel) and with the relativistic mean field EOS (GM3
model) given in ref. [ 9]. The results depicted in Fig. 5 clearly demonstrate that to neglect
hyperons leads to an overstimate of the maximum mass of neutron stars.
It is important to notice the "low" value of the stellar maximum mass, predicted within
the approach of ref. [ 15], which is in contrast with some precise determination of neutron
star masses [ 19]. For example in the case of the neutron star associated to the pulsar
PSR1913+16, the measured stellar mass is [ 20]
MP SR1913+16 = (1.4411 ± 0.0007) M⊙ .
(1)
The prediction of a value for Mmax below the measured neutron star masses is a common
feature of all the present microscopic EOS of hyperonic matter based on G-matrix BHF
calculations [ 13, 15, 6]. For example, the authors of ref.
[ 13], in case of the Argonne
v18 NN interaction, found Mmax = 2.00 M⊙, a corresponding radius of R = 10.54 km
and a central density ρc = 1.11 fm−3 for neutron stars with a pure nucleonic core. When
hyperons are considered as possible stellar constituents, they found [ 13] Mmax = 1.22 M⊙,
a corresponding radius of R = 10.46 km and a central density ρc = 1.25 fm−3. Therefore
the current equations of state for hyperonic matter, deduced from microscopic G-matrix
BHF calculations, are "too soft" to explain observed neutron star masses.
Clearly, one should try to trace the origin of this problem back to the underlying YN
and YY two body interactions or to the possible repulsive three-body baryonic forces
2.5
2
1.5
1
0.5
pn-matter
NN and free hyperons
NN and YN
NN, YN and YY
]
3
m
c
/
g
5
1
0
1
x
[
ρ
y
t
i
s
n
e
d
s
s
a
m
0.2 0.4 0.6 0.8
0
0
1.2
Baryon number density [fm-3]
1
0
0.5
1
1.5
2
ρ [ x 1015 g/cm3]
5
10
8
6
4
2
]
2
m
c
/
e
n
y
d
5
3
0
1
x
[
P
e
r
u
s
s
e
r
P
0
2.5
Figure 4. Equation of state of dense hadronic matter with and without hyperons [ 15].
involving one or more hyperons, not included in the work of ref. [ 12, 13, 14, 15, 6].
Presently this is a subject of very active research by people working in this field. Therefore,
the use of microscopic EOSs of hyperonic matter in the contest of neutron star physics
is of fundamental importance for our understanding of the strong interactions involving
hyperons, and to learn how these interactions behave in dense many-body systems.
3. KAON CONDENSATION IN NEUTRON STARS
The inner core of neutron stars could also contain a Bose-Einstein condensate of negative
kaons [ 21, 22, 23, 24, 25]. As the density of stellar matter is increased, the K − energy
is lowered by the attractive vector mean field originating from dense nucleonic matter.
When the K − energy becomes smaller than the electron chemical potential µe (which is
an increasing function of density) the strangeness changing process e− → K − +ν becomes
possible. The critical density for this process has been calculated to be in the range 2.5
-- 5.0 n0 [ 23, 24].
Due to the lack of space, we do no have the possibilty to discuss the many relevant
implications that kaon condensation has for the structure and the evolution of neutron
stars. We refer the reader to the original literature on the subject (see e.g. [ 21, 22, 23,
24, 25] and references therein quoted).
4. HYBRID STARS
The core of the more massive neutron stars is one of the best candidates in the Universe
where a phase transition from hadronic matter to a deconfined quark phase should occur.
The quark-deconfinement phase transition proceeds through a mixed phase over a finite
range of pressures and densities [ 26, 1]. At the onset of the mixed phase, quark matter
6
2.5
2
1.5
1
]
s
t
i
n
u
s
s
a
m
r
a
l
o
s
[
G
M
0.5
pn-matter
NN and YN
NN, YN and YY
With hyperons
PSR 1913+16
2.5
pn-matter
2
1.5
1
0.5
0
8
9
11 12 13 14
10
Radius [km]
8
9
0
11 12 13 14
10
Radius [km]
Figure 5. Mass-Radius relation for "traditional" neutron stars and hyperon stars calcu-
lated [ 15] within the BHF approach with the NSC9e interaction (left panel) and with
the relativistic mean field EOS GM3 of ref. [ 9] (rigth panel). The dotted horizontal line
indicates the measured mass for the "neutron star" in the radio pulsar PSR1913+16 .
droplets form a Coulomb lattice embedded in a sea of hadrons and in a roughly uniform
sea of electrons and muons. As the pressure increases various geometrical shapes (rods,
plates) of the less abundant phase immersed in the dominant one are expected. Finally the
system turns into uniform quark matter at the highest pressure of the mixed phase [ 27].
Compact stars which possess a quark matter core, either as a mixed phase of deconfined
quarks and hadrons, or as a pure quark matter phase, are called Hybrid Stars [ 1, 2, 3].
Many possible astrophysical signals for the appearence of a quark core in neutron stars
have been proposed in the last few years (see [ 1, 2, 3] and references therein quoted).
Particularly, pulse timing properties of pulsars have attracted much attention since they
are a manifestation of the rotational properties of the associated neutron star. The onset
of quark-deconfinement in the core of the star, will cause a change in the stellar moment
of inertia [ 28]. This change will produce a peculiar evolution of the stellar rotational
period (P = 2π/Ω) which will cause large deviations of the so called pulsar braking
index n(Ω) = (Ω Ω/ Ω2) from the canonical value n = 3, derived within the magnetic
dipole model for pulsars and assuming a constant moment of inertia for the star. The
possible measurement of a value of the braking index very different from the canonical
value (i.e.
n >> 3) has been proposed [ 28] as a signature for the occurrence of the
quark-deconfinement phase transition in a neutron star. However, it must be stressed
that a large value of the braking index could also results from the pulsar magnetic field
decay and/or alignment of the magnetic axis with the rotation axis [ 29].
7
5. STRANGE STARS
The possible existence of a new class of compact stars completely made of deconfined
u,d,s quark matter (strange quark matter (SQM)) is one of the consequences of an hypoth-
esis [ 30] formulated by A.R. Bodmer in 1971 and revived by E. Witten in 1984. These
stars are usually called Strange Stars. According to the Bodmer-Witten hypothesis SQM
could be the true ground state of matter. In other words, at zero temperature and pres-
sure, the energy per baryon of SQM could be less than the energy per baryon of 56Fe,
which is the most tightly bound nucleus in nature. The strange matter hypothesis does
not conflict with the existence of atomic nuclei as conglomerates of nucleons, or with the
stability of "ordinary" matter [ 31, 32, 33]. Thus strange stars may exist in the universe.
One of the most likely strange star candidate is the compact object in the transient
X-ray burst source SAX J1808.4-3658 (ref. [ 34]). This X-ray source was discovered in
1996 by the BeppoSAX satellite. Two bright type-I X-ray bursts were detected, each
lasting less than 30 seconds. Analysis of the bursts in SAX J1808.4-3658 indicates that it
has a peak X-ray luminosity of 6 × 1036 erg/s in its bright state, and a X-ray luminosity
lower than 1035 erg/s in quiescence. SAX J1808.4-3658 is a X-ray millisecond pulsar with
a pulsation period of 2.49 ms, also it is a member of a binary stellar system with orbital
period of two hours. Using the observational data collected by the Rossi X-ray Timing
Explorer during the the 1998 April-May outburst, Li et al. [ 34] have obtained an upper
limit for the compact star radiud as a function of the unknown stellar mass. Comparing
this observational mass-radius (M-R) relation of SAX J1808.4-3658 with the theoretical
M-R realtions for traditional neutron stars, hyperon stars, stars with kaon condensation,
and strange stars Li et al. [ 34] (see their Fig. 1) argue that a strange star model is more
consistent with SAX J1808.4-3658, and suggest that it could be a strange star.
SAX J1808.4-3658 is not the only LMXBs which could harbour a strange star. Recent
studies have shown that the compact stars associated with the X-ray burster 4U 1820-30
(ref.[ 35]), the bursting X-ray pulsar GRO J1744-28 (ref.[ 36]), the X-ray pulsar Her X-1
(ref.[ 37]), the kHz QPOs source 4U 1728-34 (ref.[ 38]), are likely strange star candidates.
Recently, it has been suggested that the isolated compact star RX J1856.5-3754 (ref. [
39]) could be a strange star.
6. QUARK-DECONFINEMENT PHASE TRANSITION IN NEUTRON
STARS AND GAMMA-RAY BURSTS
Gamma Ray Bursts (GRBs) are one of the most violent and mysterious phenomena
in the universe (see e.g.
ref.[ 40] for a general introduction on this subject). During
the last ten years two satellites, the Compton Gamma Ray Observatory (CGRO) and
BeppoSAX, have revolutionized our understanding of GRBs. The Burst And Transient
Source Experiment (BATSE) on board of the CGRO has demonstrated that GRBs origi-
nate at cosmological distances. The BeppoSAX discovered the X-ray afterglow. This has
permitted to determine the position of some GRBs, to identify the host galaxy and, in
a number of cases, to measure the red-shift. If the energy is emitted isotropically, the
measured fluence of the bursts implies an energy of the order of 1053 erg. However, there
is now compelling evidence that the γ-ray emission is not isotropic, but displays a jet-like
geometry. In this case the GRB energy is of the order of 1051 erg [ 41].
8
Many cosmological models for the energy source of GRBs have been proposed. Presently
one of the most popular is the so-called "collapsar", or "hypernova" model. Alternative
models are the merging of two neutron stars (or a neutron star and a black hole) in a
binary system, or the accretion of matter into a black hole. The present report is not the
appropriate place to discuss the various merits and drawbacks of the many theoretical
models for GRBs. In the following, we will mention some recent research which try to
make a connection between GRBs and quark-deconfinement phase transition.
A possible central engine for GRBs is the conversion of a pure hadronic compact star
to a strange star. The stellar conversion is triggered by the formation of a SQM drop in
the center of the hadronic star. This idea was proposed long time ago by Alcook et al. [
42]. Recently detailed calculations, based on different realistic models for the equation of
state of neutron star matter and SQM, have been performed by the authors of ref.[ 43].
They showed that the total amount of energy liberated in the conversion is in the range
(1 -- 4)×1053erg. This energy will be mainly taken away by the neutrinos produced during
the quark-deconfinement phase transition. If the efficiency of the conversion of neutrinos
to γ is of the order of a few percent [ 44], then the birth of a strange star from a neutron
star could be the energy source for GRBs.
A mounting number of observational data suggest a clear connection between supernova
(SN) explosions and GRBs [ 45, 46, 47, 48, 49]. Particularly, in the case of the gamma ray
burst of July 5, 1999 (GRB990705) and in the case of GRB011211, it has been possible to
estimate the time delay between the two events. For GRB990705 the supernova explosion
is evaluated to have occurred a few years before the GRB [ 45, 50], while for GRB011211
about four days before the burst [ 46].
The scenario which emerges from these findings is the following two-stage scenario: (i)
the first event is the supernova explosion which forms a compact stellar remnant, i.e. a
neutron star (NS); (ii) the second catastrophic event is associated with the NS and it is
the energy source for the observed GRB. These new observational data, and the scenario
outlined above, poses severe problems for most of the current theoretical models for the
central energy source of GRBs. The main difficulty of all these models is to give an answer
to the following questions: what is the origin of the second "explosion"? How to explain
the long time delay between the two events?
In the so-called supranova model [ 51] for GRBs the second catastrophic event is the
collapse to a black hole of a supramassive neutron star, i.e. a fast rotating NS with a
baryonic mass MB above the maximum baryonic mass MB,max for non-rotating configu-
rations. In this model, the time delay between the SN explosion and the GRB is equal
to the time needed by the fast rotating newly formed neutron star to get rid of angular
momentum and to reach the limit for instability against quasi-radial modes where the
collapse to a black hole occurs [ 52]. The supranova model needs a fine tuning in the
initial spin period Pin and baryonic stellar mass MB,in to produce a supramassive neutron
star that can be stabilized by rotation up to a few years. For example, if Pin ≥ 1.5 ms,
then the newborn supramassive neutron star must be formed within ∼ 0.03M⊙ above
MB,max [ 52].
In a very recent paper, Berezhiani et al. [ 53] (see also ref. [ 54]) have proposed a new
model to explain the SN -- GRB association and in particular the long time delay inferred
for GRB990705 and GRB011211. In the model of ref.[ 53], the second explosion is related
9
to the conversion from a metastable purely Hadronic Star (neutron star or hyperon star)
into a more compact star in which deconfined quark matter is present (i.e. a hybrid
star or a strange star). The new and crucial idea in the work of ref.
[ 53] with respect
to previous work [ 43], is the metastability of the hadronic star due to the existence of
a non-vanishing surface tension at the interface separating hadronic matter from quark
matter. The mean-life time of the metastable hadronic star can then be connected to the
delay between the SN explosion and the GRB. The nucleation time (i.e. the time to form
a critical-size drop of quark matter) can be extremely long if the mass of the star is small.
Via mass accretion the nucleation time can be dramatically reduced and the star is finally
converted from the metastable into the stable configuration [ 53, 54, 55]. A huge amount
of energy, of the order of 1052 -- 1053 erg, is released during the conversion process and can
produce a powerful gamma ray burst. Within the model proposed by Berezhiani et al. [
53] is is possible to have different time delays between the two events since the mean-life
time of the metastable hadronic star depends on the value of the stellar central pressure.
Thus the model of ref. [ 53] is able to interpret a time delay of a few years (as observed
in GRB990705 [ 45, 50]), of a few days (as in the case of GRB011211 [ 46]), or the nearly
simultaneity of the two events (as in the case of SN2003dh and GRB030329 [ 56]).
Acknowledgments
I am greatly indebted to I. Vidana for providing me some unpublished results from his
Ph.D. thesis and for the great help in preparing most of the figures of the present work.
REFERENCES
1. N.K. Glendenning, Compact Stars: Nuclear Physics, Particle Physics, and General
Relativity, Springer Verlag, 1996.
2. F. Weber, Pulsars as Astrophysical Laboratories for Nuclear and Particle Physics, IoP
Publishing, 1999.
3. D. Blaschke, N. K. Glendenning and A. Sedrakian (eds.), Physics of Neutron Star
Interiors, Lecture Notes in Physics, vol. 57, Springer Verlag, 2001.
I. Vidana, Ph.D. Thesis, University of Barcelona, 2001.
I. Vidana, I. Bombaci, A. Polls, and A. Ramos, Astron. & Astrophys. 399 (2003) 687.
4. V.A. Ambartsumyan and G.S. Saakyan, Sov. Astron. AJ. 4 (1960) 187.
5.
6.
7. N.K. Glendenning, Astrophys. Jour. 293 (1985) 470.
8. J. Ellis, J.I. Kapusta and K.A Olive, Nucl. Phys. B348 (1991) 345.
9. N. K. Glendenning and S.A. Moszkowski, Phys. Rev. Lett. 67 (1991) 2414.
10. H. Huber, M.K. Weigel and F. Weber, Z. Naturforsch. 54A (1999) 77.
11. S. Balberg and A. Gal, Nucl. Phys. A625, (1997) 435.
12. M. Baldo, G.F. Burgio, and H.J. Schulze, Phys. Rev. C 58 (1998) 3688.
13. M. Baldo, G.F. Burgio, and H.J. Schulze, Phys. Rev. C 61 (2000) 055801.
14. I. Vidana, A. Polls, A. Ramos, M. Hjorth-Jensen, and V.G.J. Stoks, Phys. Rev. C 61,
(2000) 025802.
15. I. Vidana, A. Polls, A. Ramos, L. Engvik, and M. Hjorth-Jensen, Phys. Rev. C 62
(2000) 035801.
10
16. M. Prakash, I. Bombaci, M. Prakash, P.J. Ellis, R. Knorren and J.M. Lattimer, Phys.
Rep. 280 (1997) 1.
17. V. G. J. Stock and Th. A. Rijken, Phys. Rev. C 59 (1999) 3009.
18. A. Akmal, V. R. Pandharipande, Phys. Rev. C56 (1997) 2261.
19. M. H. van Kerkwijk, J. van Paradijs, and E.J. Zuiderwijk, Astron. & Astrophys. 303
(1995) 497.
20. J. H. Taylor, and J.M. Weisberg, Astrophys. Jour. 345, 434 (1989).
21. D.B. Kaplan and A. E. Nelson, Phys. Lett. B 175 (1986) 57; 179 (1986) 409(E).
22. G.E. Brown, C.-H. Lee, M. Rho and V. Thorsson, Nucl. Phys. 567 (1994) 937.
23. V. Thorsson, M. Prakash and J.M. Lattimer, Nucl. Phys. A572 (1994) 693.
24. C.-H. Lee, Phys. Rep. 275 (1996) 255.
25. N.K. Gelndenning and J. Schaffner-Bielich, Phys. Rev. Lett. 81 (1998) 4564.
26. N.K. Glendenning, Phys. Rev. D 46 (1992) 1274
27. H. Heiselberg, C.J. Pethick and E.F. Staubo, Phys. Rev. Lett., 70 (1993) 1355
28. N.K. Glendenning, S. Pei and F. Weber, Phys. Rev. Lett. 79 (1997) 1603.
29. T.M. Tauris and S. Konar, Astron. & Astrophys. 376 (2001) 543.
30. A. R. Bodmer, Phys. Rev. D 4 (1971) 1601; E. Witten, Phys. Rev. D 30 (1984) 272.
31. E. Farhi and R. L. Jaffe, Phys. Rev. D 30 (1984) 2379.
32. J. Madsen, Lecture Notes in Physics, Springer Verlag, 516 (1999).
33. I. Bombaci, in ref. [ 3], pag. 253.
34. X.-D. Li, I. Bombaci, M. Dey, J. Dey, and E.P.J. van den Heuvel, Phys. Rev. Lett. 83
(1999) 3776.
35. I. Bombaci, Phys. Rev. C 55 (1997) 1587.
36. K.S. Cheng, Z.G. Dai, D.M. Wai, and T. Lu, Science 280 (1998) 407.
37. M. Dey, I. Bombaci, J. Dey, S. Ray, and B.C. Samanta, Phys. Lett. B 438, 123 (1998);
erratum, Phys. Lett. B 467, 303 (1999).
38. X.-D. Li, S. Ray, J. Dey, M. Dey, and I. Bombaci, Astrophys. Jour. 527 (1999) L51.
39. J. Drake et al., Astrophys. Jour. 572 (2002) 996.
40. T. Piran, Phys. Rep. 314 (1999) 575; Phys. Rep. 333 (2000) 529.
41. D. A. Frayl et al., Astrophys. Jour. 562 (2001) L55 (2001.
42. C. Alcock, E. Farhi, and A. Olinto, Astrophys. Jour. 310 (1986) 261.
43. I. Bombaci, and B. Datta, Astrophys. Jour. 530 (2000) L72.
44. J.D. Salmonson, and J.R. Wilson, Astrophys. Jour. 517 (1999) 859.
45. L. Amati et al., Science 290 (2000) 953.
46. J.N. Reeves et al., Nature 414 (2002) 512.
47. J.S. Bloom et al., Nature 401 (1999) 453.
48. L. Piro et al., Science 290 (2000) 955.
49. L.A. Antonelli et al., Astrophys. Jour. 545 (2000) L39.
50. Lazzati et al., Astrophys. Jour. 556 (2001) 471.
51. M. Vietri and L. Stella, Astrophys. Jour. 507 (1998) L45.
52. B. Datta, A.V. Thampan and I. Bombaci, Astron.& Astrophys. 334 (1998) 943.
53. Z. Berezhiani, I. Bombaci, A. Drago, F. Frontera, and A. Lavagno, Astrophys. Jour.
586 (2003) 1250.
54. I. Bombaci, I. Parenti and I. Vidana, Astrophys. Jour. (2003) submited.
55. I. Vidana, I. Bombaci, I. Parenti, these proceedings.
56. J. Hjorth et al., Nature 423 (2003) 847.
11
|
astro-ph/0505416 | 1 | 0505 | 2005-05-19T19:15:30 | FUSE Measurements of Far Ultraviolet Extinction. II. Magellanic Cloud Sight Lines | [
"astro-ph"
] | We present an extinction analysis of 9 paths through the LMC and SMC based on FUSE observations. To date, just two LMC sight lines have probed dust grain composition and size distributions in the Clouds using spectra including wavelengths as short as 950 A. We supplement these with results from 4 regions distinguished by their IR through UV extinction curves and grouped as LMCAvg, LMC2, SMC bar and SMC wing. Despite the distinct characters of Milky Way and Magellanic Cloud extinction, our results are generally analogous to those found for Galactic curves in that the FUSE portions of each extinction curve are described reasonably well by FM curves fitted only to longer wavelength data and lack any dramatic new extinction features, and any deviations from the CCM formalism continue into FUV wavelengths. An MEM analysis of these curves suggests that LMCAvg and SMC wing sight lines require more silicon and/or carbon in dust than current abundance measurements would allow, while the requirements for LMC2 and SMC bar sight lines do not fully tax the available reservoirs. An intermediate product of this analysis is the measurement of new H_2 abundances in the Magellanic Clouds. Collectively considering Cloud sight lines that possess significant H_2 column densities, E(B-V)/N(HI) ratios are reduced by significant factors relative to the Galactic mean, whereas the corresponding E(B-V)/N(H_2) values more closely resemble their Galactic counterpart. These trends reflect the fact that among these sight lines f(H_2)-values are lower than those common in the Milky Way for paths with similar degrees of reddening. | astro-ph | astro-ph |
FUSE Measurements of Far Ultraviolet Extinction. II. Magellanic
Cloud Sight Lines1
Stefan I. B. Cartledge1, Geoffrey C. Clayton1, Karl D. Gordon2, Brian L. Rachford3, B. T.
Draine4, P. G. Martin5, John S. Mathis6, K. A. Misselt2, Ulysses J. Sofia7, D. C. B.
Whittet8, and Michael J. Wolff9
ABSTRACT
We present an extinction analysis of nine reddened/comparison star pairs in
the Large and Small Magellanic Clouds based on Far-Ultraviolet Spectroscopic
Explorer (FUSE) FUV observations. To date, just two LMC sight lines have
probed dust grain composition and size distributions in the Magellanic Clouds
using spectral data for wavelengths as short as 950 A. We supplement these two
with data from 4 regions distinguished by their IR through UV extinction curves
and grouped as LMCAvg, LMC2, SMC bar and SMC wing. Despite the distinct
characters of extinction in the Clouds and Milky Way, our results are generally
analogous to those found for Galactic curves -- namely, that the FUSE portions
of each extinction curve are described reasonably well by Fitzpatrick & Massa
curves fitted only to longer wavelength data and lack any dramatic new extinc-
tion features, and any deviations from the Cardelli, Clayton, & Mathis (CCM)
1Department of Physics and Astronomy, Louisiana State University, Baton Rouge, LA 70803; scar-
[email protected], [email protected]
2Steward Observatory, University of Arizona, Tucson, AZ 85721; [email protected], mis-
[email protected]
3Center for Astrophysics and Space Astronomy, Department of Astrophysical and Planetary Sciences,
University of Colorado at Boulder, Campus Box 389, Boulder, CO 80309-0389; [email protected]
4Princeton University Observatory, Peyton Hall, Princeton, NJ 08544; [email protected]
5Canadian Institute for Theoretical Astrophysics, University of Toronto, Toronto, Ontario M5S 3H8
Canada; [email protected]
6University of Wisconsin-Madison, 475 North Charter Street, Madison, WI 53706; [email protected]
7Department of Astronomy, Whitman College, Walla Walla, WA 99362; [email protected]
8Department of Physics and Astronomy, Rensselaer Polytechnic Institute, Troy, NY 12180-3590;
[email protected]
9Space Science Institute, 3100 Marine Street, Suite A353, Boulder, CO 80303-1058; [email protected]
-- 2 --
formalism continue into FUV wavelengths. A Maximum Entropy Method analy-
sis of all of these curves suggests that LMCAvg and SMC wing sight lines, whose
extinction parameters more closely resemble those for Galactic paths, require
more silicon and/or carbon in dust than current abundance measurements would
indicate are available. The requirements for LMC2 and SMC bar sight lines do
not fully tax the available reservoirs, in part because large grains contribute less
to the extinction in these directions. An intermediate product of this extinction
analysis is the measurement of new H2 abundances in the Magellanic Clouds.
Collectively considering Cloud sight lines that possess significant H2 column den-
sities, E(B −V )/N(H I) ratios are reduced by significant factors relative to the
Galactic mean, whereas the corresponding E(B −V )/N(H2) values more closely
resemble their Galactic counterpart. These trends reflect the fact that among
these sight lines f (H2)-values are lower than those common in the Milky Way for
paths with similar degrees of reddening.
Subject headings: ISM: abundances -- ultraviolet: ISM
1.
Introduction
An important step toward a complete understanding of the formation, structure, and
composition of interstellar dust is the study and discrimination of factors that produce
changes in the observed wavelength-dependent extinction of stars due to grains. The Large
and Small Magellanic Clouds (LMC and SMC, respectively) are metal-poor relative to the
Galaxy; recent estimates imply that the LMC and SMC have overall metallicities at levels
0.5 and 0.2 -- 0.25, respectively, of the Galactic ISM (Welty et al. 2001). Dust-to-gas ratios
determined for these galaxies, as represented by E(B −V )/N(H I), are also reduced (by
factors of about 4 and 10; Bouchet et al. 1985; Fitzpatrick 1986), indicating that LMC and
SMC dust components and their formation mechanisms are not inconsistent with Galactic
components and mechanisms. Nevertheless, there are differences in extinction among these
three galaxies.
Early investigations of the extinction properties of Magellanic Cloud dust identified
several sight lines that exhibited marked differences from curves produced by grains in the
Milky Way (e.g., Nandy et al. 1980; Koornneef & Code 1981; Rocca-Volmerange et al. 1981).
1Based on observations with the NASA-CNES-CSA Far-Ultraviolet Spectroscopic Explorer, which is op-
erated for NASA by the Johns Hopkins University under NASA contract NAS-32985.
-- 3 --
It was noted in comparing these curves that the 2175 A bump strength was reduced in the
LMC relative to the Galaxy, and somewhat further diminished for curves characteristic of
the SMC. Similarly, the FUV rise evident in LMC and SMC extinction curves is stronger
than that common in the Galaxy. Subsequent studies have identified regional variation in
Magellanic Cloud curves, associating a distinctive LMC wavelength dependence with the
supershell LMC2 (Misselt, Clayton, & Gordon 1999), a steeper UV curve with the star-
forming bar region of the SMC, and extinction more closely resembling Galactic curves with
other portions of the Clouds. Recently, Gordon et al. (2003) (hereafter G03) completed a
comprehensive comparison of Galactic, LMC, and SMC extinction curves from near-infrared
to ultraviolet (UV) wavelengths (λ ∼> 1150 A or 1/λ ∼< 8.7µm−1); G03 confirmed that
curves characteristic of the LMC2 grouping and the SMC bar do not conform to the Cardelli,
Clayton, & Mathis (1989) (CCM) parameterization of Galactic extinction based on RV [≡
A(V )/E(B −V )].
The strength of their UV extinction is a key feature distinguishing LMC, SMC, and
Galactic curves; in particular, the steeper short-wavelength slopes associated with Magel-
lanic Cloud extinction curves can be diagnostic of differences in dust compositions and grain
size distributions, given appropriate measures of elemental abundances in their respective
interstellar media (Draine 2003). However, the extension of Magellanic Cloud extinction
curves into the FUV (> 8.7 µm−1) is complicated by the requirement that interstellar H2
absorption be removed from the spectra before an extinction curve can be produced us-
ing the pair method. Previous efforts in this area, even for Milky Way sight lines, have
been hampered by a lack of high quality data. In particular, instrumental issues of scattered
light, time-variable sensitivity, and a limited sample (both target and comparison objects) re-
stricted the utility of the Copernicus dataset (cf. Jenkins et al. 1986; Snow, Allen, & Polidan
1990), the Voyager UVS data suffer from low resolution which prevents explicit identification
and removal of the molecular hydrogen contribution, and the remaining available data [e.g.,
rocket - Green et al. 1992; Lewis, Cook, & Chakrabarti 2005; Hopkins Ultraviolet Telescope
(HUT) - Buss et al. 1994; Orbiting Retrievable Far and Extreme Ultraviolet Spectrometer
(ORFEUS) - Sasseen et al. 2002] include only a few Galactic sight lines. Far Ultraviolet
Spectroscopic Explorer (FUSE), however, has revolutionized our ability to study interstellar
molecular hydrogen. Its combination of slightly better spectral resolution and much greater
detector sensitivity than Copernicus allows H2 to be detected throughout much larger and
more diverse portions of the Galaxy and even in the Magellanic Clouds (Shull et al. 2000;
Rachford et al. 2002; Tumlinson et al. 2002). Thus, FUSE data are uniquely suited to inves-
tigating the FUV extinction properties of Milky Way, LMC, and SMC sight lines and what
any differences reveal about the dust populations in each galaxy.
Recently, Clayton et al. (2003) used a modified maximum entropy method (MEM) to
-- 4 --
fit extinction components to IR through UV curves for stars both in the Galaxy and in the
Clouds; they demonstrated that the average SMC bar extinction curve was best fit using a
grain distribution in which small silicate and amorphous carbon grains play a much larger
role than in the mean Galactic extinction. However, this result is at odds with Welty et
al. (2001), whose data suggested negligible silicon depletion in several gas cloud components
along a sight line probing the SMC ISM. In order to address issues such as dust composition,
particularly for small grains enhancing FUV extinction in the Clouds, it would be helpful
to explore LMC and SMC extinction curves shortward of the UV range that Clayton et al.
(2003) were confined to by their International Ultraviolet Explorer (IUE) data; unfortunately,
FUV extinction curves have been published to date for only very few reddened Magellanic
Cloud stars (Clayton et al. 1996; Hutchings & Giasson 2001).
In this paper, we construct new extinction curves for nine pairs of reddened and unred-
dened LMC and SMC stars by supplementing the IR through UV curves published by G03
with recent FUSE observations of 16 Magellanic Cloud targets. The FUSE data allow us to
extend the Clayton et al. (2003) analysis of dust composition and grain size distribution to
FUV wavelengths, where extinction is dominated by small grains and the differences between
Galactic and Magellanic Cloud curves are particularly prominent. This is the second in a
series of three papers exploring extinction in the FUV. Paper I (Sofia et al. 2005) dealt with
a small set of Galactic extinction curves characterized by a broad range of RV -values; Paper
III (in preparation) will examine a much larger sample of Galactic sight lines in an effort to
search rigorously for any trends among observed extinction parameters when FUV data are
fully considered.
2. Observations and Data Extraction
The present sample of 16 LMC and SMC stars includes all stars in the FUSE archive for
which the spectral match between the reddened and unreddened stars has been rigorously
evaluated and there exist spectral data from IR to UV wavelengths. Each reddened star
has previously been studied by G03; we use the same photometry sources and procedures to
construct the IR to UV portions of the extinction curves for the pairs adopted in this study,
with the exception that early 2MASS photometry has been superseded in our analysis by
data from the recent All Sky Release. In particular, it should be noted that the RV values
listed in Table 1, which summarizes the properties of each extinction pair, differ slightly from
those published by G03 due to this update of the IR data. Consequently, the values of AV
and N(H I)/AV in the table are also affected. The FUV portions of the extinction curves
(shortward of the H I Lyα line) were derived from the new and archival FUSE observations
-- 5 --
listed in Table 2. All of the raw data were processed using the latest release of CALFUSE
(v3.0). The calibrated spectra for each channel were cross-correlated, shifted to a common
wavelength scale, and combined based on exposure time in cases where several observations
of a star existed. Also, in cases where the different observations or different FUSE channels
produced conflicting flux levels for a given wavelength (e.g., due to channel misalignment),
the data for these observations or channels were linearly adjusted to match the maximum
observed flux across the regions of overlap. General details on FUSE observations have
been published by Sahnow et al. (2000a). We point out here, however, that each observation
produces eight spectra nearly 100 A in length with individual dispersion characteristics. The
LiF channel spectra (LiF1A, LiF1B, LiF2A, LiF2B) cover most wavelengths from 990 -- 1190
A twice over; the spectral region from 910 -- 1110 A are similarly sampled by the SiC channels
(SiC1A, SiC1B, SiC2A, SiC2B). Sample spectra of both poor and fine quality appear in
Figure 1, using the data from the reddened star AzV 456 and its comparison star AzV 70,
respectively.
2.1. Molecular Hydrogen Modelling and Removal
Strong FUV H2 absorption bands can significantly alter the appearance of stellar spectra,
depending upon the column density of molecular gas along the line of sight. Consequently,
a molecular hydrogen model was constructed for each sight line using procedures patterned
generally after those used by Rachford et al. (2001, 2002). The underlying assumption we
made in measuring the interstellar H2 for each sight line was that the model includes only
two velocity components, one corresponding to Milky Way gas and one associated with H2 in
the appropriate Magellanic Cloud. The equivalent width measurements for distinct profiles
of each component were entered into separate tables, so that molecular hydrogen measure-
ments could be made individually for each galaxy using a curve-of-growth (CoG) analysis.
Unfortunately, the velocity separation of Galactic and Magellanic Cloud components often
resulted in blended absorption profiles at several points in a given spectrum. In general, the
construction of a reasonable CoG for most of the sight lines required more equivalent width
measurements than were available from the unblended and isolated H2 absorption lines. To
remedy this problem, the profile-fitting code FITS6P2 and IRAF3 plotting routines were
2The FITS6P code models absorption profiles based on varying the column densities, b-values, and veloc-
ities of input interstellar components. More details on the algorithm can be found in Welty, Hobbs, & York
(1991).
3IRAF is distributed by the National Optical Astronomy Observatories, which are operated by the As-
sociation of Universities for Research in Astronomy, Inc., under cooperative agreement with the National
-- 6 --
used in conjunction with our IDL4 code (Rachford et al. 2001, 2002) to deconvolve blended
profiles and fill out each CoG with as many measurements as could reliably be made. The
CoG analysis, however, was generally limited to J ≥ 3 because the J ≤ 2 lines were sat-
urated to the extent that they possessed broad damping wings that the equivalent width
measurement algorithm could not reliably distinguish from the stellar continuum. Thus, an
iterative curve-fitting procedure was an additional requirement so that simultaneous deter-
minations of Galactic and Magellanic Cloud column densities for these lower H2 rotational
excitation levels could be made. The complete molecular hydrogen models, including col-
umn densities for the available J levels, the associated b-value, and the derived kinetic and
excitation temperatures (these last properties are discussed in section 3.1), are summarized
in Tables 3 and 4 for all of our LMC and SMC sight lines; Table 3 characterizes the Galactic
component for each sight line where one was detected, while Table 4 details the properties
of each Magellanic Cloud component.
Once molecular hydrogen absorption models had been constructed, each spectrum was
corrected for this absorption. All of the FUSE data for each star were combined into a single
spectrum by shifting the calibrated channel spectra to a common wavelength reference and
averaging the fluxes, weighted by the flux uncertainty at each point in the overlapping
regions. Although there are significant differences in the data dispersion properties for each
channel, the merged spectrum for each star is of sufficient quality for the construction of a
reliable extinction curve. The "worm" problem in FUSE data (Sahnow et al. 2000b) was
strong only in LiF1B. Its appearance in the data for this detector segment was identified
by a comparison with the LiF2A flux as the data for all detector segments were merged
into a single spectrum for each sight line; the portions of the LiF1B spectrum contaminated
by the worm were eliminated from the merging process. A running cross-correlation of
the normalized H2 absorption profile corrected for any wavelength mismatches, and the
full final model was divided into the merged FUSE spectrum to complete the "removal" of
the molecular hydrogen features. Atomic hydrogen was not measured for each sight line
individually, but its signature, specifically the Lyα and Lyβ lines, was removed after the
ratio of the reddened and comparison spectra was calculated as a function of wavelength.
Although FUSE spectra cover wavelengths from 905 through 1187 A, data shortward of 1044
A (our Lyβ cutoff) were not included in this analysis due to limitations in data quality.
In the interest of consistency in instrument response, FUSE flux levels were then man-
Science Foundation.
4IDL is an acronym for Interactive Data Language, a common programming tool developed by Research
Systems, Inc.
-- 7 --
ually rescaled to match IUE fluxes over the wavelength range common to both instruments
(1150 − 1185 A) when the full IR to FUV spectra were constructed; given recent improve-
ments to the calibration of IUE data (Massa & Fitzpatrick 2000), flux errors associated with
this instrument have not yet been surpassed by the latest FUSE calibration (Sahnow et al.
2000a). Generally, mean fluxes for the data from each instrument agreed to within 20%,
although there were a few targets for which the discrepancies exceeded this range. The scal-
ings applied to each FUSE spectrum (the ratio of IUE to FUSE flux in the overlap region)
are given in Table 5.
2.2. The Extinction Curves
We have constructed extinction curves using the pair method (e.g., Massa, Savage, &
Fitzpatrick 1983), normalized to A(V ) through RV ; the pairs of reddened and unreddened
stars in the LMC and SMC were selected (G03) and are listed in Table 1.
In an effort
to minimize uncertainties in each extinction curve, particularly from spectral mismatch, a
detailed comparison of the FUSE spectrum for each star with similarly-typed unreddened
candidates was performed, and the extinction pairs matched by G03 using IUE were also
found to be well-matched in the FUV. Assembly of the extinction curve corresponding to
each pair was accomplished using the same procedures and near-infrared through UV data
outlined by G03. We have also constructed average extinction curves for the subsets of the
current sample associated with the LMCAvg and LMC2 regions considered previously in the
UV (Misselt, Clayton, & Gordon 1999; G03). Each of the individual LMC curves and the
average curves have been fit to the Fitzpatrick & Massa (1990) parameterization; the results
are presented in Table 6. The two sight lines in the SMC, one representing each of the bar
and wing regions, are considered individually.
A quick examination of the individual Magellanic Cloud extinction curves presented
in Figure 2 reveals that the FUSE portion of each curve is generally consistent with an
extension of the data from longer wavelengths. When the calibrated and H2-adjusted fluxes
were initially compared, however, slight offsets were apparent between the portions of the
curve on either side of the Lyα line. Comparing the size of the gaps in the Magellanic Cloud
data with the Galactic curves studied in Paper I led to the conclusion that the quality of the
data for our reddened stars might have compromised the accuracy of the extinction curve
construction. In particular, the S/N of the FUSE data seemed to be inversely proportional
to the size of the offset and there were noticeable flux level differences for some spectra
between the IUE and the merged FUSE data in the spectral overlap region. The Galactic
sight lines had much better S/N values, negligible offsets, and good agreement between IUE
-- 8 --
and FUSE fluxes, whereas the Magellanic Cloud data suffered generally poor S/N, noticeable
offsets, and sometimes poor flux agreement in the overlap. In order to reduce the offsets,
the FUSE data for the Cloud sight lines were linearly adjusted (a wavelength-independent
multiplicative factor was applied to the flux) over the entire FUV bandpass so that their
fluxes matched IUE in the instruments' overlap region. An example of one of the more
egregious cases is plotted in Figure 3, including a comparison of the FUSE flux levels both
before and after adjustment to the IUE flux level. The magnitude of each shift was estimated
in increments of 0.05 × f luxIUE; the systematic uncertainty associated with this adjustment
has been propagated through all subsequent error calculations. Typically, the shifts improved
the appearance of the curves, although the offsets were not eliminated in all cases. Given the
general smoothness of the transition between UV and FUV portions of the extinction curve
in most cases, it might seem reasonable to introduce further shifts to align these segments;
nevertheless, because the observed offsets fall generally within the error bounds, because we
are interested in the detailed shape of the curve at UV and FUV wavelengths, and since we
have no independent objective basis for such shifts, no further processing of the individual
curves has been performed. It should also be noted that any residual offsets appear to be
biased in the direction of decreased A(λ)/A(V ); thus, abundance requirements derived from
the FUV data may be slight underestimates.
3. Discussion
3.1. H2 in the Magellanic Clouds
As already mentioned, a notable difficulty in deriving extinction curves that extend into
the FUV is the existence of strong molecular hydrogen absorption bands longward of the
Lyα line. Nevertheless, stellar observations of sufficient quality to allow the H2 absorption
to be modelled and removed not only permit the construction of an extinction curve but
also provide more direct information on the intervening ISM. Recently, Tumlinson et al.
(2002) completed a survey of molecular hydrogen in the LMC and SMC using several FUSE
observations of each Cloud; our sample introduces 15 new sight lines into the mix.
In considering these new molecular hydrogen measurements, however, it should be rec-
ognized that the current FUSE data generally have much lower signal-to-noise ratios than
Galactic observations. In fact, they are also somewhat lower than the values for previously-
observed Magellanic Cloud targets. Tumlinson et al. (2002) reported typical 4σ equivalent
width limits of 30 -- 40 mA, a level similar to values of 20 -- 50 mA for our comparison stars
but much lower than the 50 -- 120 mA characteristic of the reddened targets. Because of these
data limitations, and because the goal of generating a reliable FUV continuum was given
-- 9 --
a somewhat higher priority than making a rigorous assessment of molecular hydrogen, any
column densities smaller than about 1017 cm−2, and a few that are somewhat larger, are
subject to sizable uncertainties. Nevertheless, we consider the models listed in Tables 3 and
4 to be reasonable tallies of the H2 along each sight line.
In order to more fully understand how the new sight lines mesh with the Tumlinson et al.
(2002) Magellanic Cloud H2 analysis, we have adopted the atomic hydrogen measurements
previously published by Fitzpatrick (1985a,b) for the individual sight lines (see Table 7);
for the pair listings in the table, we use the values derived directly from the extinction
curves by G03. It should be noted here that the process of determining atomic hydrogen
column densities for the Magellanic Clouds is complicated by the large quantities detected
along each sight line. The breadth of the Lyα absorption profile produced by typical inter-
stellar hydrogen distributions with line-of-sight lengths exceeding even a few kiloparsecs is
sufficient to mask distinctions between separate velocity components; in particular, the pro-
files for atomic hydrogen gas in the LMC or SMC are often inextricably blended with their
Galactic counterparts. Tumlinson et al. (2002) dealt with this problem by calibrating 21 cm
emission measurements to Lyα from the sight lines in their sample for which the Galactic
and Magellanic Cloud 1215 A profiles were not severely blended. Atomic hydrogen column
densities for the remaining paths in the data set were then estimated using this calibration
and the observed 21 cm emission. G03 assessed Magellanic Cloud H I column densities by
measuring the Lyα profile in the reddened-to-comparison ratio spectrum for each extinction
pair, under the assumption that the Milky Way components would cancel each other. Each
of these methods is subject to a significant level of uncertainty; if vertical error bars were
to be plotted in Figure 4, among the Magellanic Cloud sight lines they might typically be
about 0.3 dex and occasionally larger than 1.0 dex. Nevertheless, differences between the
Fitzpatrick (1985a,b) H I values for the individual sight lines in each extinction pair generally
are well matched by the corresponding column densities derived by G03.
Considered individually, our LMC and SMC sight lines and those studied by Tumlinson
et al. (2002) exhibit similar ranges in characteristics such as the molecular hydrogen fraction
f (H2) (≡ 2N(H2)/[N(H I) + 2N(H2)]) and the Magellanic Cloud portion of the color excess
E ′(B −V );5 values for the two LMC sight lines studied by Gunderson, Clayton, & Green
(1998) also match these samples. Since the current dataset was selected for its extinction
properties, we cannot speak to the overall frequency of molecular hydrogen detection in
the Clouds. However, as shown in Figure 4, the f (H2)-values for paths along which we
detect a Cloud-based H2 component agree with the levels set by previously observed LMC
5Following the procedure of Tumlinson et al. (2002), we adopt the definition E ′(B −V ) = E(B −V ) − x,
where x is 0.075 and 0.037 for LMC and SMC paths, respectively.
-- 10 --
and Galactic sight lines of large E(B −V ). Notably, the sight lines toward AzV 456 and
Sk −67◦2 possess larger molecular hydrogen fractions than any examined by Tumlinson et
al. (2002); meanwhile, among the paths directed toward our comparison stars we detect no
Magellanic H2 components.
Based on the previously-studied sight lines toward Sk -66◦19 and Sk -69◦270, it was
found that while the gas-to-dust ratio N(H I)/E ′(B −V ) is much larger in the Clouds than
typically seen in the Milky Way, the quantity N(H2)/E ′(B −V ) has a similar value in each
galaxy (Clayton et al. 1996; Gunderson et al. 1998). This suggests a close relationship
between dust and H2. Similar behavior is evident for our new sight lines, since although
the ratios N(H I)/E ′(B −V ) for these paths are roughly 4 -- 10 times larger than the Galactic
mean, comparing H2 column densities with the same color excesses leads to a range of values
around the Galactic average. In this way our new sight lines complement the paths studied
by Gunderson et al. (1998) and the more heavily reddened paths of Tumlinson et al. (2002),
implying generally lower values for f (H2) in the Clouds; nevertheless, the current dataset
includes sight lines with unique properties. Plotting the LMC N(H I)/E ′(B −V ) ratio as a
function of E(B − V ) (see Figure 5), there is a clear distinction between Sk -67◦2 and other
reddened sight lines: this one LMC path is characterized by an "atomic" ratio appropriate
to Galactic curves, but because of its large molecular hydrogen column density the total
gas-to-dust ratio approaches the values for other LMC sight lines. Coupled with its Milky
Way-like N(H I) ratio, the agreement exhibited between the gas-to-dust ratios for Sk -67◦2,
the sight line with the largest f (H2)-value in our sample, and sight lines dominated by atomic
gas reinforces the concept that the mechanisms governing dust formation in the Galaxy
and LMC are similar and that any differences in extinction arise out of how interstellar
environmental conditions are manifested in dust grain population characteristics. The SMC
sight line AzV 456 also possesses a "Galactic" value for N(H I)/E ′(B −V ) that gives a
N(HTotal)/E ′(B −V ) ratio more closely resembling the values determined by Tumlinson et
al. (2002) for other SMC paths.
Further details of the interstellar conditions, specifically the temperatures T01 and Tex,
in the Clouds can be derived from the relative rotational level populations of the individual
sight lines (Table 4). When the cloud density is sufficiently high and H2 column densities
are large enough for the gas to become substantially molecular, ortho-para conversion will
bring T01 (≡ E1−E0
ln(9N (0)/N (1)) ) close to the kinetic temperature.
Except for the AzV 462 sight line, its value is fairly uniform among the LMC and SMC
components of our H2 models and we derive a weighted mean temperature of 50±3 K. The
Magellanic Cloud component of H2 toward AzV 462 has the lowest column density among
our extragalactic values, and the assumption that this value is consistent with the kinetic
temperature breaks down. The mean temperature we have derived for the Magellanic Cloud
log(N (J=0)/g0)−log(N (1)/g1) =
k
log(e)
170K
-- 11 --
components neglects AzV 462 and is somewhat lower than the overall Galactic mean of
77±17 K (Savage et al. 1977), but is consistent with both the value 55±8 K derived from
the Galactic subsample for which N(H2) > 20.4 (Rachford et al. 2002) and temperatures
derived by Tumlinson et al. (2002) for Magellanic Cloud gas along sight lines with some of
the largest molecular column densities. It should be noted that the errors ascribed to each
temperature listed in Tables 3 and 4 were derived from column density uncertainties, which
were often large. Thus, kinetic temperatures with uncertainties larger than about 60 K should
be regarded as less reliable. In particular, the formal uncertainty is listed in each case, even
where it exceeds the calculated temperature, in order to indicate the relative reliability of
each determination. We have also determined Tex for each sight line from the slope in a plot
of N(J = 2, 3, 4, 5) versus the excitation potential.6 This excitation temperature reflects the
influence of fluorescent UV pumping and H2 formation in excited states on the molecular
hydrogen rotational level population (Jura 1975; Black & Dalgarno 1976). Like T01, it is
fairly uniform among the LMC and SMC sight lines and we calculate a weighted mean of
244±33 K, similar to previously-determined values for Magellanic Cloud sight lines (Shull et
al. 2000). Comparing the column density ratios N(4)/N(2) and N(5)/N(3) for our sample
with Figure 11 from Tumlinson et al. (2002), we find that these sight lines are reasonably
typical of their high N(H2) paths; consequently, our measurements support their conclusions
that the H2 formation rate in the Clouds is low (10 -- 40% of Galactic rates) and that at least
some regions are illuminated by relatively intense radiation fields (10 -- 100 times the Galactic
mean). We note, however, that the proportion of outliers in our sample that do not fit the
range of Galactic models calculated by Tumlinson et al. (2002) is much smaller than theirs.
Values of T01 and Tex were also derived for the Galactic components of our H2 models. A
mean of T01 = 67±4 K was derived for the sight lines with larger column densities and more
robust measurements, although higher, and likely to be in part non-thermal, temperatures
are implied for several of the more diffuse paths. Excitation temperatures for the Galactic
components (239±38 K) are similar to those we have determined for the Magellanic Clouds.
3.2. FUV Extinction in the Magellanic Clouds
Before FUSE, only two reddened sight lines outside the Galaxy had been studied in the
FUV (Clayton et al. 1996). Both regional extinction groups in the LMC are represented
6More generally, Tex is associated with values of TJ −2,J ≡ EJ −EJ−2
log[N (J −2)/gJ−2]−log[N (J)/gJ ] =
(for J ≥ 2), the temperature derived from comparing the populations of successive even or
170K×(2J −1)
N (J−2)
ln[ 2J+1
2J−3
odd J levels.
N (J)
]
k
log(e)
-- 12 --
by these two stars; specifically, Sk -66◦19 is associated with LMCAvg and Sk -69◦270 with
LMC2 (Misselt et al. 1999). The UV extinction curves for these stars are typically non-
CCM, but they share the trait of most Galactic curves in that their UV extinction slope
blends smoothly into the FUV. These two sight lines have been included for comparison in
Table 6, which defines the LMC2 and LMCAvg subsamples, although they were left out
of the average extinction curves for those regions. Hutchings & Giasson (2001) published
FUV extinction curves for 3 sight lines in each of the LMC and the SMC using FUSE data.
Unfortunately, the values of ∆(B-V) for the reddened/unreddened star pairs chosen for that
study are too small (∼0.02-0.07 mag) for detailed extinction studies. More specifically,
these reddening values are comparable in size to the variation in the foreground E(B −V )
and similar to the magnitude of the photometric uncertainties (Schwering & Israel 1991;
Oestreicher, Gochermann, & Schmidt-Kaler 1995; Massey 2002). So, within the uncertainties
there is no significant reddening difference for any of the star pairs in Hutchings & Giasson
(2001), and those sight lines are not included in this study.
The new LMC and SMC FUV extinction curves presented here seem to follow the trend
of Sk -69◦270 and Sk -66◦19. Despite any small remaining offsets between the IUE and FUSE
extinction curves (see § 2.1), the new FUV curves closely follow an extrapolation of their
FM fits from the UV. Out of our sample, only two curves exhibit significant deviations from
this overall trend, AzV 456 and Sk -69◦228. AzV 456 is our lone reddened star situated in
the SMC wing, and the FUSE portion of its extinction curve more closely resembles its FUV
CCM curve than an extrapolation of its FM fit in the UV. Coincidentally, the disagreement
with the FM fit begins near the onset of molecular hydrogen absorption features; however,
attempting to reconcile the two portions of the FUSE curve by adjusting the H2 column
densities indicates that a single component for the Clouds cannot reproduce the observed
continuum level shift, nor does it appear that a simple combination of overlapping compo-
nents would accomplish this result. The effect does not appear to be related to a difference in
the sensitivity of the various FUSE channels either, since the continuum across this spectral
region is contained in both the LiF1B and LiF2A detector bands. Similar kinks are apparent
to a much smaller degree in the Galactic curves for HD 62542, HD 73882, and HD 210121
(Paper I); nevertheless, those curves match the FM fits within the formal uncertainties [ap-
proximately 14% in A(λ)/A(V) across the FUSE band] and we feel that it would not be
appropriate to speculate further about the origin of these features until a larger data sample
is compiled.
The FUV extinction curve for Sk -69◦228 is very noisy, which would usually indicate a
poor match with its unreddened comparison star, but this same star is a good match through
the UV. The problem seems to arise with a complex of photospheric Fe III lines evident
in the comparison star spectrum, Sk -65◦15. An examination of intermediate continuum
-- 13 --
segments demonstrates that the reddened and comparison stellar spectra have very similar
shapes aside from these lines, and that connecting the lower wavenumber portion of this
star's FUV extinction curve with the two highest wavenumber points would be a reasonable
approximation. Notably, this procedure results in an FUV curve that is consistent with an
extension of the UV FM fit, despite the impression that a small offset is still present after
the FUSE flux adjustment. The curve is still included in further analysis for the sake of
completeness and because our comparison sample is limited, but we note that the mean
curve derived for the LMC2 group, which includes this path, is consistent with the other
curves in the group unaffected by any taint of mismatch. On the whole, the ease with which
the UV and FUV extinction curves line up for our Magellanic Cloud sight lines complements
the results for Milky Way sight lines; namely, that most Galactic sight lines also seem to
show FUV extinction that is a good extrapolation of the FM fits made to the extinction
curves in the UV. Agreement between the CCM relation and FUV extinction, however, has
proved less ubiquitous. For instance, Buss et al. (1994) found that the FUV extinction for
two Galactic sight lines with large (ρ Oph) and small (HD 25443) values of RV follow CCM
closely, but that there are exceptions; one notable example is the bright-nebula sight line
toward HD 37903 which shows a steeper extinction in the FUV, based on Copernicus data,
than would be predicted by CCM. Paper I, which included paths such as HD 210121 and
HD 62542 that do not follow CCM in the UV, came to similar conclusions. Although the
UV extinction curves for each sight line they studied could be smoothly extrapolated into
the FUV using an FM fit, only three paths that followed CCM through UV wavelengths
were also in accord across the FUV. The CCM relations for four others from Paper I either
over or underestimated their FUV extinction, as small discrepancies in the UV became more
pronounced at shorter wavelengths, and the two non-CCM curves diverged more strongly
from their RV -based curves with increasing wavenumber. Because the LMC and SMC curves
do not have a CCM-like wavelength dependence and because the S/N ratios of the individual
extinction curves are relatively low, subtle variations in the FUV slope such as those seen
by Buss et al. (1994) cannot be ruled out.
3.3. FUV MEM Modelling: Grain Properties
To increase the S/N for MEM modelling purposes, we have constructed average extinc-
tion curves for the LMC average and LMC2 regions outlined by Misselt et al. 1999 (see
Table 6); the curves are plotted in Figure 6. The two sight lines in the SMC, which rep-
resent the bar and wing regions, are considered individually. Using the two average LMC
curves and the two SMC curves, we have extended the dust-grain population analysis of
Clayton et al. (2003) into the FUV. We employ a (slightly) modified version of the MEM
-- 14 --
extinction-fitting algorithm (Kim et al. 1994; Kim & Martin 1996).
Instead of using the
number of grains of a given size to describe the dust population, the algorithm employs the
mass distribution in which m(a)da is the mass of dust grains per H atom in the size interval
from a to a+da. Thus, the traditional MRN-type model (Mathis, Rumpl, & Nordsieck 1977)
becomes m(a) ∝ a−0.5. We use a power law with exponential decay (PED) as the template
function for each component. The data are examined at 34 wavelengths, and the grain cross
sections are computed over the range 0.0025 -- 2.7 µm with 50 logarithmically spaced bins.
The shape of the mass distribution is strongly constrained only for data over the interval
0.02 -- 1 µm. Below 0.02 µm, the Rayleigh scattering behavior constrains only total mass;
above 1 µm, the "grey" nature of the dust opacity also forces the MEM algorithm to simply
adjust the total mass, using the shape of the template function to specify the size dependence
of the distribution. The total mass of dust is constrained using both the gas-to-dust ratio
and "cosmic" abundances (i.e., we try not to use more carbon or silicon than is available).
The elemental abundance standards used in this analysis are those adopted in Paper I for
the Galaxy (358 and 35 atoms per million H for C and Si, respectively), and by Clayton
et al. (2003) for the LMC and SMC (110 and 65 for the LMC and 54 and 11 atoms per
million H for the SMC); Table 7 reports the values of the gas-to-dust ratios used here for
the LMC and SMC. We consider only three-component models of homogeneous, spherical
grains: modified "astronomical silicate" (Weingartner & Draine 2001), amorphous carbon
(Zubko et al. 1996), and graphite (Laor & Draine 1993). While it must be acknowledged
that the three grain component system we have used to model these extinction curves is
simpler than might be expected of actual interstellar dust, the results can be quite useful for
identifying grain population properties that distinguish sight lines from one another. The
same component-specific PED constraints (e.g., the onset of the exponential cutoff) adopted
in Paper I were utilized for the current modelling.
The MEM fits to FM parameterizations of the average Magellanic Cloud extinction
curves listed in Table 6 are presented in Figures 7 and 8. The first plot shows the amount of
extinction provided by the three distinct grain components, as well as the total extinction
of the model, compared to the FM fit associated with each extinction curve. The error bars
plotted on these fits are indicative of the mean gap between the FM fit and the underlying
extinction curve in each wavelength bin. The fractions of the adopted elemental abundances
available for each grain component that are required by the best MEM fit are listed in
Table 8. Figure 8 depicts the corresponding mass distributions for different sizes of grains
belonging to each model component relative to the mass of hydrogen.
The proportions of the available silicon and carbon used in the MEM fits to the various
FM parameterizations cover a very wide range, as was seen in fits to IR through UV data
alone by Clayton et al. (2003). Three general factors determine the fraction of silicon and
-- 15 --
carbon that any individual sight line will use. First, the higher the gas-to-dust ratio is, the
more metals are available in the gas phase. Second, the higher the abundances of metals
are, the more material is available. Finally, high values of RV imply a greater than average
mass fraction in larger grains which are not as efficient per unit mass as smaller grains. For
instance, it can be seen from the MEM fits that the SMC wing (AzV 456), which has a
low gas-to-dust ratio and low elemental abundances relative to the Galaxy, uses more than
100% of the available silicon. However, the LMC2 and the SMC bar (AzV 18) regions, which
also have both low elemental abundances but are characterized by higher gas-to-dust ratios,
use less than half the amount of available silicon that the LMCAvg and SMC wing regions
require and also significantly less carbon.
A recent observation has suggested that silicon is relatively undepleted in the SMC bar
(Welty et al. 2001). However, model fits to the average SMC bar extinction indicate that its
curve cannot be fitted with carbon grains alone (Weingartner & Draine 2001; Clayton et al.
2003). Figures 7 and 8 clearly show that this conclusion is born out in the current fits. SMC
extinction for both bar and wing is very steep in the FUV, necessitating the presence of a
large population of small grains. Yet because the 2175 A bump is absent in the bar region,
small (a < 0.02µm) carbon grains do not meet the SMC FUV extinction requirements; thus,
silicates play a very important role in the models. Even our SMC wing sight line, whose
extinction curve follows CCM reasonably well, places a much higher demand on silicon
reserves than carbon: a typical Galactic curve conforming to CCM allows graphite to fulfill
a significant part in FUV extinction due to the strength of its UV bump, but the SMC wing
bump is somewhat weaker than the RV =2.19 CCM curve would imply. The next steepest
sight lines, in the LMC, do not show much difference in demand from a typical Galactic
curve (HD 14250 was chosen from Paper I for illustration purposes), once the gas-to-dust
ratio differences are taken into account. Typically, carbon grains are responsible for most of
the visible extinction and silicon grains for most of the UV extinction; therefore, in general,
both species of grains are needed along any sight line to get a good fit to the extinction
curve. Apart from deficits of large silicate and carbon grains in the SMC, the MEM analysis
of the Magellanic Cloud FUV extinction produces results similar to those derived from an
analysis of several Galactic sight lines (Paper I). The Galactic sight lines, like the SMC wing
or LMCAvg, are more likely to use more than 100% of the available silicon and/or carbon.
Among these sight lines, the strength of the 2175 A bump in relation to the FUV rise is
the factor that distinguishes whether the stiffer abundance demand is placed on silicon or
carbon. Likewise, the element most demanded by the SMC bar sight line and the mean
LMC2 curve is determined by this factor, although the minimal presence of the bump in
these curves allows the modelling to make more efficient use (especially in small grains) of
the available elements.
-- 16 --
The general structure in the size vs m(a) curves for extinction out to 8 µm−1 for the
Magellanic Cloud sight lines does not change much when the FUSE data are included.
Figures 7 and 8 showing the MEM fits and size vs m(a) curves cannot be directly compared
to Figure 3 of Clayton et al. (2003) because there are differences in the input parameters
such as the gas-to-dust ratios. But the MEM models run with the same input parameters
for extinction curve that are cutoff at 8 µm−1 and those shown here, which extend to 10
µm−1, are quite similar; there is at most a few percent difference in the amount of silicon
and carbon required by the two sets of models for the four SMC and LMC curves. The
structure evident in the mass distributions of Figure 8 particularly identify grain sizes for
which each component in our simple model can account for the detailed shape of each
extinction curve. For instance, the two peaks in the amorphous carbon distribution for HD
14250 correspond roughly to maxima appearing in the silicate distribution of Kim et al.
(1994) in their silicate-graphite model. These authors ascribe the peaks to the requirements
of fitting the optical and UV portions of the extinction curve using the spectral wavelength
dependence appropriate to each of the two grain components in their model. In the case
of AzV 456 (the SMC wing), however, efforts to reproduce the extinction curve using this
three-component model appear to require dramatic peaks and dips in the mass distribution
as a function of grain size. Aside from this curve, however, the scale of structure in the
distribution is consistent with levels noted by previous studies (Kim et al. 1994; Clayton
et al. 2003). For AzV 456, the solution requires that the graphite distribution be strongly
peaked in order to account for the 2175 A bump and provide as much extinction as possible
at FUV wavelengths. Since the FUV portion of this curve extends, or even amplifies, the
steep rise at the short-wavelength end of the previous IR through UV result (Clayton et al.
2003), there is no relaxation of the demand on silicon reserves. Specifically, when comparing
MEM solutions for the SMC wing curve with and without the FUSE data, nearly identical
proportions of the available silicon and carbon are utilized. The only notable distinction
between the two solutions is that a few percent of the amorphous carbon demand is shifted
to graphite when the FUV data are included. The extinction curves for our other groups
also smoothly extend from the UV into FUV wavelengths, and the models derived from IR-
to-UV and IR-to-FUV data require almost precisely the same amounts of silicon and carbon
be present in dust grains. The similarity in the MEM results for these two sets of models
may be due to the efficiency of the small grains in FUV extinction, in that large numbers
of such grains are not needed, or perhaps the observed amounts of FUV extinction can be
provided by the larger grains which are also important for extinction at longer wavelengths.
This issue will be investigated further as we study the distinction between global and sight-
line-specific FUV extinction characteristics of using a much larger number of sight lines, in
a future paper.
-- 17 --
4. Conclusion
This paper is the second in a series investigating the FUV characteristics of extinction
due to interstellar dust. The first paper dealt with a small sample of Galactic sight lines
with a variety of RV values, including paths whose extinction does not conform to the CCM
parameterization. This paper examined a similarly-sized set of sight lines probing the Small
and Large Magellanic Clouds, with the goal of determining how the addition of FUV data
to their corresponding extinction curves affected models of Magellanic Cloud dust grain
properties.
Using FUSE observations of four stars in the SMC and 12 stars in the LMC, we were able
to construct two extinction curves for the SMC and seven representing the LMC. Despite the
poor quality of the data for some of the reddened stars, molecular hydrogen was measured for
each sight line and their corresponding absorption features were removed from consideration
in the extinction analysis. Flux mismatches between IUE and FUSE data, likely due to
poor S/N values in the affected observations, were alleviated by rescaling the entire FUSE
spectrum. Comparison of the resulting FUV extinction with the curve constructed only
through wavelengths observed by IUE demonstrated that the FUV portions were generally
consistent with a smooth extrapolation of the IR-to-UV curve. Nevertheless, we note that
a "kink" appeared in the AzV 456 curve, similar to much smaller features present in curves
analysed in Paper I. We plan to address the existence and nature of this kink in a subsequent
paper examining the variation in properties of FUV extinction.
It was anticipated that including FUV data in the MEM analysis of the Magellanic Cloud
dust grain population might give rise to significantly different properties than Galactic dust
since the FUV rise in the Clouds is generally much steeper than it is in the Milky Way,
and the UV bump is smaller. However, aside from some complex structure likely driven by
the simplicity of our grain component model, the sight lines through the Clouds exhibitting
Galactic-like extinction are similar to Galactic paths in that they demand 100% or more
of the available carbon and/or silicon. The only real differences between the populations
for these galaxies are that the SMC wing sight line has a much smaller grain size cutoff
for astronomical silicates and amorphous carbon and a generally smaller dust mass. MEM
solutions for the sight lines through the Clouds that exhibited more distinctly Magellanic-
Cloud-type extinction were able to satisfy their curves using smaller proportions of the
elements available to them; the SMC bar solution, like that for the SMC wing, is characterized
by a smaller grain size cutoff and lower dust mass than Galactic or LMC solutions. The
addition of the FUSE data to the analysis does not dramatically alter the properties of each
dust population, except to shift small amounts of the carbon demand between amorphous
carbon and graphite.
-- 18 --
The molecular hydrogen abundances determined in the process of constructing the full
IR-to-FUV extinction curves supplement those already appearing in the literature. Although
the emphasis has been on eliminating H2 absorption features from the spectra rather than
deriving robust column density measurements, our results are consistent with the larger
recent survey of molecular hydrogen in the Magellanic Clouds by Tumlinson et al. (2002)
and we complement their paths by probing several sight lines with larger column densities.
Among these sight lines, E(B −V )/N(H I) ratios are generally reduced relative to Galactic
values while E(B −V )/N(H2) are roughly the same, implying somewhat lower f (H2)-values
in the Clouds than are typical in the Milky Way for similar degrees of reddening.
This study was supported by NASA grant NAG5-108185.
Black, J. H., & Dalgarno, A. 1976, ApJ, 203, 132
REFERENCES
Bouchet, P., Lequeux, J., Maurice, E., Pr´evot, L., & Pr´evot-Burnichon, M. L. 1985, A&A,
149, 330
Browning, M. K., Tumlinson, J., & Shull, J. M. 2003, ApJ, 582, 810
Buss, R. H., Jr., Allen, M., McCandliss, S., Kruk, J., Liu, J., & Brown, T. 1994, ApJ, 430,
630
Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245 (CCM)
Clayton, G. C., Green, J., Wolff, M. J., Zellner, N. E. B., Code, A. D., Davidsen, A. F.,
WUPPE Science Team, & HUT Science Team 1996, ApJ, 460, 313
Clayton, G. C., Wolff, M. J., Sofia, U. J., Gordon, K. D., & Misselt, K. A. 2003, ApJ, 588,
871
Diplas, A., & Savage, B. D. 1994, ApJ, 427, 274
Draine, B. T. 2003, ARA&A, 41, 241
Fitzpatrick, E. L. 1985a, ApJ, 299, 219
Fitzpatrick, E. L. 1985b, ApJS, 59, 77
Fitzpatrick, E. L. 1986, AJ, 92, 1068
-- 19 --
Fitzpatrick, E. L., & Massa, D. M. 1990, ApJS, 72, 163
Gordon, K. D., & Clayton, G. C. 1998, ApJ, 500, 816
Gordon, K. D., Clayton, G. C., Misselt, K. A., Landolt, A. U., & Wolff, M. J. 2003, ApJ,
594, 279 (G03)
Green, J. C., Snow, T. P., Jr., Cook, T. A., Cash, W. C., & Poplawski, O. 1992, ApJ, 395,
289
Gunderson, K.S., Clayton, G.C., & Green, J.C. 1998, PASP, 110, 60
Hutchings, J. B., & Giasson, J. 2001, PASP, 113, 1205
Jenkins, E. B., Savage, B. D., & Spitzer, L. 1986, ApJ, 301, 355
Jura, M. 1974, ApJ, 191, 375
Jura, M. 1975, ApJ, 197, 581
Kim, S.-H., Martin, P. G., & Hendry, P. D. 1994, ApJ, 422, 164
Kim, S.-H., & Martin, P. G. 1996, ApJ, 462, 296
Koornneef, J., & Code, A. D. 1981, ApJ, 247, 860
Koornneef, J. 1982, A&A, 107, 247
Laor, A., & Draine, B. T. 1993, ApJ, 402, 441
Lewis, N. K., Cook, T. A., & Chakrabarti, S. 2005, ApJ, 619, 367
Longo, R., Stalio, R., Polidan, R. S., & Rossi, L. 1989, ApJ, 566, 267
Massa, D., & Fitzpatrick, E. L. 2000, ApJS, 126, 517
Massa, D., Savage, B.D., & Fitzpatrick, E. L. 1983, ApJ, 266, 662
Massey, P. 2002, ApJS, 141, 81
Misselt, K. A., Clayton, G. C., & Gordon, K. D. 1999, ApJ, 515, 128
Mathis, J. S., Rumpl, W., & Nordsieck, K. H. 1977, ApJ, 217, 425
Nandy, K., Morgan, D. H., Willis, A. J., Wilson, R., Gondhalekar, P. M., & Houziaux, L.
1980, Nature, 283, 725
-- 20 --
Oestreicher, M. O., Gochermann, J., & Schmidt-Kaler, T. 1995, A&AS, 112, 495
Rachford, B. L., Snow, T. P., Tumlinson, J., Shull, J. M., Roueff, E., Andr´e, M., D´esert, J.
-M., Ferlet, R., Vidal-Madjar, A., & York, D. G. 2001, ApJ, 555, 839
Rachford, B. L., Snow, T. P., Tumlinson, J., Shull, J. M., Blair, W. P., Ferlet, R., Friedman,
S. D., Gry, C., Jenkins, E. B., Morton, D. C., Savage, B. D., Sonnentrucker, P.,
Vidal-Madjar, A., Welty, D. E., & York, D. G. 2002, ApJ, 577, 221
Rocca-Volmerange, B., Pr´evot, L., Pr´evot-Burnichon, M. L., Ferlet, R., Lequeux, J. 1981,
A&A, 99, L5
Sahnow, D. J., Moos, H. W., Ake, T. B., Andersen, J., Andersson, B.-G., Andr´e, M., Artis,
D., Berman, A. F., Blair, W. P., Brownsberger, K. R., Calvani, H. M., Chayer, P.,
Conard, S. J., Feldman, P. D., Friedman, S. D., Fullerton, A. W., Gaines, G. A.,
Gawne, W. C., Green, J. C., Gummin, M. A., Jennings, T. B., Joyce, J. B., Kaiser,
M. E., Kruk, J. W., Lindler, D. J., Massa, D., Murphy, E. M., Oegerle, W. R., Ohl,
R. G., Roberts, B. A., Romelfanger, M. L., Roth, K. C., Sankrit, R., Sembach, K.
R., Shelton, R. L., Siegmund, O. H. W., Silva, C. J., Sonneborn, G., Vaclavik, S. R.,
Weaver, H. A., Wilkinson, E. 2000a, ApJ, 538, L7
Sahnow, D. J., Gummin, M. A., Gaines, G. A., Fullerton, A. W., Kaiser, M. E., & Siegmund,
O. H. 2000b, Proc. SPIE, 4139, 149
Sasseen, T. P., Hurwitz, M., Dixon, W., V., & Airieau, S. 2002, ApJ, 566, 267
Savage, B. D., Bohlin, R. C., Drake, J. F., & Budich, W. 1977, ApJ, 216, 291
Savage, B. D., & Mathis, J. S. 1979, ARA&A, 17, 73
Schwering, P. B. W., & Israel, F. P. 1991, A&A, 246, 231
Shull, J. M., Tumlinson, J., Jenkins, E. B., Moos, H. W., Rachford, B. L., Savage, B. D.,
Sembach, K. R., Snow, T. P., Sonneborn, G., York, D. G., Blair, W. P., Green, J. C.,
Friedman, S. D., & Sahnow, D. J. 2000, ApJ, 538, L73
Snow, T. P., Jr., Allen, M. M., Polidan, R. S. 1990, ApJ, 359, L23
Snow, T. P., Jr., & York, D. G. 1975, Ap&SS, 34, 19
Sofia, U. J., Wolff, M. J., Rachford, B. L., Gordon, K. D., Clayton, G. C., Cartledge, S. I.
B., Martin, P. G., Draine, B. T., Mathis, J. S., Snow, T. P., & Whittet, D. C. B.
2005, ApJ, accepted (Paper I)
-- 21 --
Tumlinson, J., Shull, J. M., Rachford, B. L., Browning, M. K., Snow, T. P., Fullerton, A.
W., Jenkins, E. B., Savage, B. D., Crowther, P. A., Moos, H. W., Sembach, K. R.,
Sonneborn, G., & York, D. G. 2002, ApJ, 566, 857
Weingartner, J. C., & Draine, B. T. 2001, ApJ, 548, 296
Welty, D. E., Hobbs, L. M., & York, D. G. 1991, ApJS, 75, 425
Welty, D. E., Lauroesch, J. T., Blades, C., Hobbs, L. M., & York, D. G. 2001, ApJ, 554, L75
York, D. G., Drake, J. F., Jenkins, E. B., Morton, D. C., Rogerson, J. B., & Spitzer, L. 1973,
ApJ, 182, L1
Zubko, V. G., Menella, V., Colangeli, L., & Bussoletti, E. 1996, MNRAS, 282, 1321
This preprint was prepared with the AAS LATEX macros v5.2.
-- 22 --
1050
2e-13
1075
1100
1125
1150
1175
)
1
-
Å
1
-
s
2
-
m
c
g
r
e
(
x
u
l
F
1e-13
0
2e-12
1e-12
2e-13
1e-13
0
2e-12
1e-12
0
1050
1075
1100
1125
Wavelength (Å)
1150
1175
0
Fig. 1. -- Sample FUSE spectra.
Sample FUSE spectra are plotted above, including an example of poor quality data in the
upper panel (for AzV 456) and fine quality data in the lower panel (for AzV 70); these
spectra also comprise an extinction pair. Of further note, the AzV 456 continuum drops
dramatically between 1080 and 1120 A, giving rise to a peculiar kink in the extinction curve
discussed later in the text.
-- 23 --
0
2
4
6
8
10
0
2
4
6
8
10
0
2
4
6
8
10
)
V
A
(
/
)
λ
(
A
6
4
2
0
6
4
2
0
6
4
2
0
AzV 18
RV = 2.90
AzV 456
RV = 2.19
Sk -67° 2
RV = 3.75
Sk -68° 26
RV = 3.45
Sk -68° 155
RV = 2.81
Sk -68° 129
RV = 3.37
Sk -69° 228
RV = 3.54
Sk -68° 140
RV = 3.34
Sk -69° 279
RV = 3.54
6
4
2
0
6
4
2
0
6
4
2
0
2
4
6
8
10
0
2
4
6
1/λ [µm-1]
8
10
0
2
4
6
8
0
10
Fig. 2. -- Magellanic Cloud Extinction Curves.
The extinction curves for each of the reddened/unreddened star pairs in Table 1 are plotted
above; IR, optical, UV (from IUE), and FUV (from FUSE) data are all included. A CCM
curve based on the RV value derived from the IR and optical portions of each extinction
curve is also plotted using a dashed line, and the FM fit to the full curves are indicated in
each panel by solid grey lines. In each case but AzV 456 and Sk -69◦228, the FM fit for only
IR through UV data almost precisely overlaps the FM fit to the full curve. For these two
exceptions, the IR-to-UV fit is represented by a solid dark line. The vertical lines identify
central wavelengths for Lyα and Lyβ.
-- 24 --
7.6
7.8
8.0
8.2
8.4
8.6
8.8
9.0
3e-13
2e-13
1e-13
0
3e-13
2e-13
1e-13
0
7.6
7.8
8.0
8.4
8.2
1/λ [µm-1]
8.6
8.8
3e-13
2e-13
1e-13
0
3e-13
2e-13
1e-13
0
9.0
]
1
-
Å
1
-
s
2
-
m
c
g
r
e
[
x
u
l
F
Fig. 3. -- Sk -69◦228 Spectral Match.
IUE and FUSE spectra in a region that includes these instruments' overlapping wavelength
coverage have been plotted above for the star Sk -69◦228; IUE fluxes are represented by the
heavy solid line and the light grey line signifies FUSE data. The top and bottom panels
depict the flux levels before and after the linear corrections were applied. These corrections
generally reduced the size of extinction curve offsets between UV and FUV data. The vertical
dashed lines delimit portions of the spectrum contaminated by Lyα absorption that have
been eliminated from the curves plotted in Figure 2; the central portions have also been set
to zero in this plot.
-- 25 --
Current LMC/SMC Data
LMC/SMC (Tumlinson et al. 2002)
LMC (Gunderson et al. 1998)
MW (Savage et al. 1977)
0
-2
]
)
2
H
f(
[
0
1
g
o
l
-4
-6
-8
0
0.05
0.1
0.15
E(B-V)
0.2
0.25
0.3
Fig. 4. -- Galactic and Magellanic Cloud Molecular Hydrogen Fractions.
The variation of molecular hydrogen fraction f (H2) with color excess E(B −V ) is plotted
above for sight lines passing through material associated with the Magellanic Clouds and
the Milky Way. The current sample includes stars that are more heavily reddened that
the Tumlinson et al. (2002) sample and that emphasize the similarities between how f (H2)
relates to E(B −V ) in these galaxies. The label E(B −V ) used in this caption and for the
x-axis in the plot refer to unadjusted values in the Galaxy but E ′(B −V ) for Magellanic
Cloud sight lines.
-- 26 --
0
0.05
0.1
0.15
0.2
0.25
)]
V
-
B
(
'
E
/
)
I
H
N(
[
0
1
g
o
l
22.5
22.0
21.5
)]
V
-
B
(
'
E
/
)
l
a
t
o
T
H
N(
[
0
1
g
o
l
22.5
22.0
21.5
LMC
Milky Way
LMC
Milky Way
SK -67° 2
New data
Tumlinson et al (2002)
Gunderson et al (1998)
22.5
22.0
21.5
22.5
22.0
21.5
0.00
0.05
0.10
0.15
E'(B-V)
0.20
0.25
Fig. 5. -- LMC Gas-to-Dust Ratios.
Gas-to-dust ratios for LMC extinction pairs are plotted above as a function of E(B − V ).
Of particular note, Sk -67◦2 alone among the reddened LMC stars is characterized by a
log10[N(H I)/E ′(B −V )] ratio consistent with that of a Galactic star (dotted line; Diplas
& Savage 1994), whereas the corresponding value of log10[N(HTotal)/E ′(B −V )] agrees with
other LMC ratios. The LMC value for log10[N(H I)/E ′(B −V )] (dashed line; Koornneef
1982) is also plotted in both graphs for reference.
-- 27 --
2
4
6
8
10
SMC Bar
LMC2
SMC Wing
LMCAvg
10
5
0
10
5
0
10
5
0
10
5
)
V
(
λ)/A
(
A
0
0
2
4
6
1/λ [µm-1]
8
0
10
Fig. 6. -- Comparing the LMC, SMC, and Galactic Average Extinction Curves.
Average extinction curves for the Magellanic Cloud regions LMC2 and SMC bar, those more
distinct in character from Galactic extinction, are shown in the upper panel of the above
plot; the lower panel depicts the mean curves for the LMCAvg and SMC wing groupings.
Both panels include the CCM RV =3.1 curve in the role of a Galactic reference. The LMC
and SMC curves are offset 1 and 2 units in A(λ)/A(V ), respectively, from the Galactic curve
in each panel.
-- 28 --
Fig. 7. -- MEM Average Extinction Curve Fits.
The MEM fits of astrophysical silicate, amorphous carbon, and graphite dust components
to FM parameterizations of the average extinction curves in Figure 6 are plotted above.
It should be noted that the MEM models were adjusted to reproduce the FM fit to each
extinction curve; consequently, the SMC wing curve does not match the AzV 456/AzV 70
extinction plot of Figure 2 in full detail. The error bars are indicative of the deviation
between the FM fit and the extinction curve in nearby bins. For comparison, a typical
Galactic sight line is represented in this plot by HD 14250 (RV =2.98±0.14; Paper I).
-- 29 --
Fig. 8. -- MEM Dust Grain Mass Distributions.
The MEM grain size vs m(a)da functions for each of the average extinction curves of Figure 6
are plotted above. Of particular note, the SMC grain distributions do not include silicate
or amorphous carbon grains as large as are required to fit extinction curves in the LMC or
the Galaxy. Also, the graphite grain distribution for the SMC wing is strongly peaked by
the constraints implied in the bump strength and FUV rise of its extinction curve. As in
Figure 7, HD 14250 stands in for a typical Galactic sight line.
Table 1. Extinction Curve Pairs
Reddened Comparison Spectral ∆(B −V )
Star
(mag)
Star
Type
RV
AV
(mag)
N(H I)/AV
(1021 H I atoms/AV )
AzV 462
AzV 18
AzV 70
AzV 456
Sk -66◦35
Sk -67◦2
Sk -66◦35
Sk -68◦26
Sk -68◦129 Sk -68◦41
Sk -68◦140 Sk -68◦41
Sk -68◦155 Sk -67◦168
Sk -69◦228 Sk -65◦15
Sk -69◦279 Sk -65◦63
B2 Ia
O9.7 Ib
B2 Ia
B3 Ia
O9 Ia
B0 Ia
O8 Ia
B2 Ia
O9 Ia
0.17±0.03
0.26±0.03
0.15±0.05
0.19±0.03
0.17±0.05
0.20±0.05
0.20±0.05
0.15±0.05
0.21±0.05
2.90±0.42
2.19±0.23
3.75±0.36
3.45±0.27
3.37±0.29
3.34±0.25
2.81±0.22
3.54±0.34
3.54±0.25
0.49±0.11
0.57±0.08
0.56±0.23
0.64±0.16
0.57±0.22
0.67±0.22
0.56±0.18
0.53±0.23
0.75±0.23
17.27±4.30
7.01±1.22
1.78±0.67
5.46±1.08
6.98±2.36
5.99±1.75
8.88±9.21
6.60±2.50
5.38±1.50
--
3
0
--
Note. -- The Galactic foreground extinction is considered to be comparable for the stars in each
pair; consequently, the properties listed in this table are appropriate only to the Magellanic Cloud
dust component (Misselt, Clayton, & Gordon 1999). Spectral types refer to the UV classification
of each star; the sources are Gordon & Clayton (1998) and Misselt, Clayton, & Gordon (1999).
-- 31 --
Table 2. FUSE LMC and SMC Observations
Sight Line FUSE Data Set
Date
Exposure Time (s)
AzV 18
AzV 70
AzV 456
AzV 462
Sk -65◦15
Sk -65◦63
Sk -66◦35
Sk -67◦2
Sk -67◦168
Sk -68◦26
Sk -68◦129
Sk -68◦140
Sk -68◦155
Sk -69◦228
Sk -69◦279
A1180101000
B0890101000
A1180202000
A1180203000
B0900601000
Q1070101000
Q1070104000
Q1070106000
Q1070102000
Q1070103000
P2210201000
A1180301000
B0861001000
A0490701000
M1142001000
B0861101000
B1280101000
B0860301000
B0860101000
B0860901000
B0860201000
B0860501000
B0860601000
B0860701000
B0860401000
B0860801000
2000 May 29
2001 Jun 13
2000 Oct 03
2000 Oct 05
2001 Jun 15
2000 Oct 04
2000 Oct 06
2000 Oct 09
2000 Oct 10
2000 Oct 12
2001 Jun 14
2000 Jul 03
2001 Nov 17
1999 Dec 16
2000 Sep 26
2001 Sep 22
2001 Oct 25
2002 Sep 22
2001 Aug 14
2001 Sep 22
2001 Sep 17
2001 Sep 22
2001 Sep 17
2001 Sep 22
2001 Sep 23
2001 Sep 22
9293
45245
2831
1971
6152
3691
4903
5641
2687
4352
8278
5550
4293
5927
5581
4125
4030
4480
6935
4102
11391
6681
10712
8020
9422
5991
-- 32 --
Table 3. Molecular Hydrogen Models: Milky Way
log10 N (H2)MW log10 N (0)
(cm−2)
(cm−2)
log10 N (1)
(cm−2)
log10 N (2)
(cm−2)
log10 N (3)
(cm−2)
log10 N (4)
(cm−2)
log10 N (5)
(cm−2)
b
km s−1
a
T01
(K)
b
Tex
(K)
16.57(0.64)
18.55(0.05)
15.92(0.16)
18.15(0.06)
18.00(0.07)
17.78(0.46)
16.83(0.36)
16.38(0.35)
16.02(0.35)
15.31(0.27)
16.73(0.59)
19.01(0.12)
18.35(0.50)
18.00(0.20)
16.17(0.16)
16.33
18.42
15.55
17.79
17.75
17.24
16.56
15.60
15.61
14.31
15.76
18.54
18.26
17.29
15.70
16.17
17.95
15.48
17.86
17.54
17.60
16.19
16.25
15.32
15.14
16.39
18.82
17.64
17.80
15.93
SMC
LMC
14.82
15.90
14.88
16.75
16.66
16.34
15.98
15.24
15.58
14.39
16.15
17.37
15.77
17.22
14.88
14.27
15.46
14.98
16.04
16.56
15.74
15.81
14.74
14.67
14.31
15.87
16.28
15.36
16.14
14.71
14.66
14.08
15.13
14.37
14.39
8.1+2.9
−1.9
4.2+0.9
−1.6
5.9+2.3
−4.9
2.4+1.2
−1.4
1.4+1.1
−0.4
4.8+2.6
−2.5
6.3+5.9
−5.2
3.5+1.2
−2.2
5.4+6.0
−4.4
4.3+4.0
−3.3
2.9+6.0
−1.9
2.5+2.4
−1.5
4.6+2.2
−3.6
2.5+5.9
−1.5
8.3+5.9
−6.0
66±192
52±6
72±34
187±296
357±786
419±223
84±6
64±8
125±310
56±90
200±95
60±62
326±230
228±868
110±42
47±133
167±189
102±59
165±330
303±263
210±227
277±197
200±95
354±519
326±230
409±868
227±50
330±183
224±620
277±308
14.17
--
3
3
--
MC Star
AzV 18
AzV 70
AzV 462
Sk −65◦15
Sk −65◦63
Sk −66◦35
Sk −67◦2
Sk −67◦168
Sk −68◦26
Sk −68◦41
Sk −68◦129
Sk −68◦140
Sk −68◦155
Sk −69◦228
Sk −69◦279
aThe H2 kinetic temperature is assumed to be equivalent to T01, the temperature derived from N (J=0) and N (1) assuming a Boltzmann distribution
(T01 = E1−E0
k
log(N(0)/g0 )−log(N(1)/g1) ).
log(e)
bThe excitation temperature Tex reflects the sum of effects such as UV photon-pumping and excited state formation, and is identified in this table
with the slope of the log[N (J)/gJ] vs. E(J) plot for J ≥ 2.
Note. -- This table describes the Milky Way component of the molecular hydrogen models constructed for each target star from the FUSE data.
Table 4. Molecular Hydrogen Models: Magellanic Clouds
MC Star
log10 N (H2)M C
(cm−2)
log10 N (0)
(cm−2)
log10 N (1)
(cm−2)
log10 N (2)
(cm−2)
log10 N (3)
(cm−2)
log10 N (4)
(cm−2)
log10 N (5)
(cm−2)
b
km s−1
T01
(K)
Tex
(K)
AzV 18
AzV 456
AzV 462
Sk −66◦35
Sk −67◦2
Sk −68◦26
Sk −68◦129
Sk −68◦140
Sk −68◦155
Sk −69◦228
Sk −69◦279
20.36(0.07)
20.93(0.09)
17.65(0.13)
19.13(0.19)
20.95(0.08)
20.38(0.08)
20.05(0.10)
19.50(0.13)
19.99(0.24)
18.70(0.14)
20.31(0.07)
20.13
20.85
16.65
19.02
20.86
20.20
20.00
19.09
19.75
18.28
20.10
19.97
20.15
17.54
18.46
20.21
19.90
19.08
19.25
19.62
18.37
19.90
SMC
LMC
17.58
17.67
16.56
17.02
18.41
18.50
18.06
18.13
17.21
17.79
16.46
17.79
15.62
16.16
15.89
16.43
17.76
16.65
17.45
16.37
17.13
16.11
16.06
15.36
14.26
15.88
15.55
15.45
14.84
14.78
15.40
16.43
15.87
15.97
15.42
15.34
8.1+2.9
−1.9
6.0+5.9
−5.0
2.1+1.2
−1.1
4.4+3.0
−3.1
5.6+6.0
−4.6
2.3+1.0
−1.3
8.0+3.8
−4.2
5.4+2.6
−2.0
8.2+3.3
−4.9
3.1+6.0
−2.1
10.0+6.0
−6.0
66±9
45±4
· · ·
49±23
46±5
59±7
40±13
93±35
68±42
86±37
64±9
270±52
82±175
234±564
302±152
159±75
253±260
309±157
248±71
341±152
249±555
389±437
Note. -- This table details the Magellanic Cloud component of the molecular hydrogen models constructed for each target star from the FUSE
data. The temperatures T01 and Tex possess the same characteristics as those in Table 3.
--
3
4
--
-- 35 --
Table 5. Flux Shifts
Star
IUE/FUSE Flux Ratio Star
IUE/FUSE Flux Ratio
AzV 18
AzV 70
AzV 456
AzV 462
Sk −65◦15
Sk −65◦63
Sk −66◦35
Sk −67◦2
1.10
1.05
0.85
1.00
1.20
0.85
1.05
0.65
Sk −67◦168
Sk −68◦26
Sk −68◦41
Sk −68◦129
Sk −68◦140
Sk −68◦155
Sk −69◦228
Sk −69◦279
1.05
0.85
0.95
1.00
1.10
1.20
1.25
1.30
Note. -- The table ratios are derived from the calibrated fluxes recorded by
each instrument in the spectral overlap from 1150 -- 1185 A.
Table 6. FM Fit Parameters
Curve
c1
c2
c3
c4
x0
γ
AzV 18
−4.902±1.036
2.255±0.436
AzV 456
−0.419±0.130
0.908±0.094
−4.66
Sk −66◦19a
Sk −67◦2
−3.479±1.585
Sk −68◦26
0.003±0.020
Sk −68◦129 −2.174±0.784
−1.704±0.253
Average
2.02
1.723±0.648
0.937±0.150
1.393±0.443
1.268±0.041
SMC Bar
0.165±0.213
SMC Wing
5.026±1.625
LMC
1.21
3.108±1.296
2.581±0.513
1.184±0.351
2.902±0.187
LMC2
0.001±0.026
4.697±0.078
0.738±0.012
0.513±0.100
4.770±0.079
1.470±0.025
0.85
4.54
0.73
0.881±0.313
0.203±0.054
0.207±0.082
0.266±0.030
4.573±0.048
4.638±0.049
4.567±0.071
4.602±0.023
0.935±0.016
0.898±0.015
0.688±0.041
0.930±0.015
Sk −68◦140 −1.929±0.820
Sk −68◦155 −2.842±0.982
Sk −69◦228 −2.800±1.219
Sk −69◦270a
Sk −69◦279 −2.696±0.857
−2.487±0.303
Average
−3.51
1.323±0.387
1.615±0.458
1.449±0.529
0.161±0.055
1.093±0.525
0.897±0.336 −0.012±0.013
0.641±0.237 −0.418±0.164
4.440±0.074
4.611±0.062
4.714±0.079
0.811±0.014
0.684±0.028
0.609±0.054
1.52
0.97
0.24
4.62
0.78
1.364±0.364
1.443±0.054
0.853±0.232 −0.124±0.057
0.844±0.140 −0.055±0.016
4.603±0.077
4.586±0.058
0.664±0.047
0.697±0.012
aIn the interest of providing additional extinction curves for comparison, the data for sight lines previ-
ously observed by Hopkins Ultraviolet Telescope (Sk −66◦19; Sk −69◦270; Clayton et al. 1996; Gunderson,
Clayton, & Green 1998) have been included in this table.
--
3
6
--
Table 7. Properties of the Magellanic Cloud ISM
Sight Line or
Extinction Pair
N(H I)
(cm−2)
N(H2)
(cm−2)
E ′(B −V )a N(H I)/E ′(B −V ) N(H2)/E ′(B −V )
log10fH2
(mag)
(cm−2 mag−1)
(cm−2 mag−1)
AzV 18
AzV 456
AzV 462
AzV 18/AzV 462
AzV 456/AzV 070
Sk −66◦35
Sk −67◦2
Sk −68◦26
Sk −68◦129
Sk −68◦140
Sk −68◦155
Sk −69◦228
Sk −69◦279b
9.0×1021
1.5×1021
6.0×1020
8.5×1021
4.0×1021
4.0×1020
1.0×1021
3.5×1021
5.2×1021
5.5×1021
4.2×1021
4.0×1021
3.5×1021
Sk −66◦19/Sk −69◦83
Sk −67◦2/Sk −66◦35
Sk −68◦26/Sk −66◦35
Sk −68◦129/Sk −68◦41
7.0×1021
1.0×1021
3.5×1021
4.0×1021
2.3×1020 −1.35
8.5×1020 −0.28
4.5×1017 −2.83
2.3×1020 −1.29
8.5×1020 −0.53
SMC: individual targets
0.174
0.332
0.007
SMC: extinction pairs
0.17
0.26
LMC: individual targets
0.055
0.190
0.181
0.201
0.215
0.217
0.145
0.221
LMC: extinction pairs
0.25
0.18
0.19
0.17
1.3×1019 −1.20
8.9×1020 −0.19
2.4×1020 −0.92
1.1×1020 −1.38
3.0×1019 −1.97
9.8×1019 −1.35
4.9×1018 −2.61
2.0×1020 −0.98
1.6×1020 −1.64
8.9×1020 −0.19
2.3×1020 −0.94
1.1×1020 −1.27
--
3
7
--
5.2×1022
4.5×1021
8.6×1022
5.0×1022
1.5×1022
7.3×1021
5.3×1021
1.9×1022
2.6×1022
2.6×1022
1.9×1022
2.8×1022
1.6×1022
2.5×1022
5.6×1021
1.8×1022
2.3×1022
1.3×1021
2.6×1021
6.4×1019
1.4×1021
3.3×1021
2.4×1020
4.7×1021
1.3×1021
5.5×1020
1.4×1020
4.5×1020
3.4×1019
9.0×1020
6.4×1020
5.0×1021
1.2×1021
6.6×1020
Table 7 -- Continued
Sight Line or
Extinction Pair
N(H I)
(cm−2)
N(H2)
(cm−2)
E ′(B −V )a N(H I)/E ′(B −V ) N(H2)/E ′(B −V )
log10fH2
(mag)
(cm−2 mag−1)
(cm−2 mag−1)
LMCAvg curvec
Sk −68◦140/Sk −68◦41
Sk −68◦155/Sk −67◦168
Sk −69◦228/Sk −65◦15
Sk −69◦270/Sk −67◦78
Sk −69◦279/Sk −65◦63
LMC2 curvec
· · ·
4.0×1021
5.0×1021
3.5×1021
3.5×1021
4.0×1021
· · ·
· · ·
−1.00
3.0×1019 −1.83
9.8×1019 −1.42
4.9×1018 −2.56
0.7×1020 −1.69
2.0×1020 −1.04
−1.50
· · ·
· · ·
0.20
0.20
0.15
0.19
0.21
· · ·
1.1×1022
2.0×1022
2.5×1022
2.3×1022
1.8×1022
1.9×1022
1.9×1022
5.9×1020
1.5×1020
4.9×1020
3.3×1019
3.7×1020
9.7×1020
3.1×1020
aAs defined in the text, E ′(B −V ) refers to the portion of E(B −V ) for each star arising in the Magellanic
Clouds; the value listed for each extinction pair is equivalent to the quantity ∆(B −V ) in Table 1.
bThe atomic hydrogen column density for Sk −69◦279 is based on N(H I)LM C from nearby stars, since no
previously-published measurement could be found.
cGas-to-dust ratios for each of the two LMC group mean extinction curves are included. In order to derive
representative molecular gas-to-dust ratios, an f (H2)-value was assumed for each curve and compared with
the atomic gas-to-dust ratio determined from our averaging code; the results of these calculations are roughly
consistent with the sight line properties in each group. RV -values determined for the LMCAvg and LMC2 mean
--
3
8
--
curves are 3.49 ± 0.17 and 3.24 ± 0.13, respectively.
Note. -- Comparison of our H2 measurements with Tumlinson et al. (2002) required estimates of N(H I)LM C
and E′(B −V ), the Magellanic Cloud portion of the color excess; our sources are Fitzpatrick (1985a,b). The
properties for extinction pairs studied by Gunderson, Clayton, & Green (1998), Sk -66◦19 and Sk -69◦270, are
listed for comparison.
--
3
9
--
-- 40 --
Table 8. MEM Curve Abundance Requirements
Extinction
Grouping
Silicon Carbon
AS
Total AMC Graphite
SMC Wing
SMC Bar
LMCAvg
LMC2
HD 14250
170% 44% 15%
78%
19% 14%
45% 100% 77%
19%
78% 66%
114% 75% 51%
29%
5%
23%
12%
24%
Note. -- Our MEM modelling reproduced
curves corresponding to the FM parameters fit to
the average extinction curves for each sight line
grouping including wavenumbers from 0.455 to
9.575 µm−1; a typical Galactic sight line is repre-
sented here by HD 14250 (RV =2.98±0.14; Paper
I). The demands on each element are expressed as
a percentage of the total interstellar abundance
for each of the astronomical silicate (AS), amor-
phous carbon (AMC), and graphite grain compo-
nents.
|
astro-ph/9604085 | 1 | 9604 | 1996-04-16T13:27:41 | The Supernova Remnant G11.2-0.3 and its central Pulsar | [
"astro-ph"
] | The plerion inside the composite Supernova Remnant G11.2-0.3 appears to be dominated by the magnetic field to an extent unprecedented among well known cases. We discuss its evolution as determined by a central pulsar and the interaction with the surrounding thermal remnant, which in turn interacts with the ambient medium. We find that a plausible scenario exists, where all the observations can be reproduced with rather typical values for the parameters of the system; we also obtain the most likely period for the still undetected pulsar. | astro-ph | astro-ph | The Supernova Remnant G11.2 -- 0.3 and its central Pulsar
Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, 50125 Firenze, Italy
R. Bandiera
F. Pacini
Osservatorio Astrofisico di Arcetri, and
Dipartimento di Astronomia e Scienza dello Spazio, Universit`a degli Studi,
Largo E. Fermi 5, 50125 Firenze, Italy
and
M. Salvati
Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, 50125 Firenze, Italy
Received 3 February 1996;
accepted 11 April 1996
6
9
9
1
r
p
A
6
1
1
v
5
8
0
4
0
6
9
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
-- 2 --
ABSTRACT
The plerion inside the composite Supernova Remnant G11.2 -- 0.3 appears to
be dominated by the magnetic field to an extent unprecedented among well
known cases. We discuss its evolution as determined by a central pulsar and the
interaction with the surrounding thermal remnant, which in turn interacts with
the ambient medium. We find that a plausible scenario exists, where all the
observations can be reproduced with rather typical values for the parameters of
the system; we also obtain the most likely period for the still undetected pulsar.
Subject headings:
ISM: Supernova Remnants -- ISM: Individual: G11.2 -- 0.3 --
Stars: Pulsars: General
-- 3 --
1.
Introduction
Vasisht et al. (1996, hereafter VA) have recently discussed the nature of the remnant
G11.2 -- 0.3 (Historical Supernova 386 AD) and have shown that it is likely to be of a
composite nature, as already suggested on the basis of radio data by Morsi & Reich (1987).
The hard X-ray emission observed by ASCA, combined with previous Einstein data and
observations at other wavelengths, strongly suggests (despite the fact that a pulsar has not
yet been discovered) the presence of a plerion surrounded by a thermal shell. The data
across the electromagnetic spectrum can then be combined with the theory of the evolution
of pulsar-powered remnants (Pacini & Salvati 1973 (PS); Bandiera, Pacini & Salvati 1984;
Reynold & Chevalier 1984) and allow a determination of the physical parameters of the
system. The main outcome of the interpretation proposed by VA is that the plerion
should be characterized by a magnetic energy much higher than the energy channelled into
relativistic electrons, at variance with, e.g., the Crab Nebula.
In our paper we study the evolution of the plerion coupled with that of the surrounding
shell. We show that the magnetic field determination of VA is only marginally consistent,
from a dynamical point of view, if the thermal and non-thermal remnants are in pressure
equilibrium. We identify, on the other hand, a scenario that is in agreement with the
observations, and derive the initial and present parameters of the central pulsar. In
particular, we show that the apparent preponderance of the magnetic component in the
pulsar output may be a natural outcome of the evolution, and does not imply a strong
unbalance in the mechanism of production.
Following VA we assume that G11.2 -- 0.3 is the remnant of SN 386 AD and therefore has
an age t = 1610 yr = 5 × 1010 s. For its distance we use a reference value of 5 kpc and define
d5 = d/(5 kpc). The outer thermal shell has radius Rs = 3.3 d5 pc, electron temperature
Ts = 0.73 keV = 8.5 × 106 K, X-ray luminosity Ls ∼ 1036d2
5 erg s−1 (in the 0.6 -- 3.3 keV
-- 4 --
band), from which VA derive an ambient density no ∼ 1.5 d−1/2
5
cm−3, by assuming a Sedov
(1959) expansion in the interstellar medium. The central plerionic component has a radius
Rp ∼ 1 d5 pc and its spectrum is equal to
Lν ≃ 3 × 1016(cid:18)
ν
2.5 × 1018 Hz(cid:19)−0.8
d2
5 erg s−1Hz−1,
(1)
corresponding to an X-ray luminosity Lp ≃ 1 × 1035d2
5 erg s−1, in the 0.5 -- 4 keV band. eq. (1)
is consistent with the observations down to 32 GHz, but overestimates the plerion emission
at 1 GHz. Accordingly, the spectrum must have a break at a frequency νo between 1 GHz
and 32 GHz.
2. The model
The assumption of the PS model is that the central pulsar feeds both magnetic field
and relativistic particles into the expanding plerion. Taking into account the adiabatic
losses and the rate of production, the evolution of the field strength is given by
d
p
dt B2R4
6 ! = p ERp.
(2)
Here p is the fraction of the pulsar energy loss E that goes into magnetic energy; it is
assumed that p is constant. According to standard pulsar electrodynamics Ω ∝ −Ωn, with
n being the so-called braking index, and
E =
Eo
(1 + t/τo)α ,
(3)
with α = (n + 1)/(n − 1) = 2 in a pure dipole field. Measured values of n range roughly
between 1.4 for the Vela pulsar (Lyne, private communication) and 2.8 for PSR 1509-58
(Kaspi et al. 1994). The pulsar is assumed to inject also relativistic particles, whose
energies are distributed according to a power law with constant index γ, at a rate (1 − p) E.
-- 5 --
The evolution of the electron energies is determined by both adiabatic and synchrotron
losses.
An immediate implication of the theory is that the emitted spectrum of the plerion
should show a break at a frequency νb such that particles radiating at νb have a lifetime
equal to the age of the nebula. Since νb ∝ B−3t−2, if the observed break is identified with
νb one directly derives the field strength in G11.2 -- 0.3
B = 2.1 × 10−3(cid:18) νb
32 GHz(cid:19)−1/3
G,
which corresponds to a total field energy
WB = 2.1 × 1049(cid:18) νb
32 GHz(cid:19)−2/3
d3
5 erg.
The synchrotron spectrum above νb, up to a maximum frequency νmax, is given by
Lν =
2 − γ
γ − 1
(1 − p) E
2νmax (cid:18) ν
νmax(cid:19)−γ/2
.
(4)
(5)
(6)
From the observations γ = 1.6 and νmax > 2.5 × 1018 Hz; by combining eqs. (1) and (6)
we find the present-time energy loss from the pulsar:
E = 2.24 × 1035(cid:18)
νmax
2.5 × 1018 Hz(cid:19)0.2 d2
1 − p
5
erg s−1.
(7)
We describe the relative importance of the magnetic and particle components in terms
of the dimensionless quantity Q = B2R3
p/6(1 − p) Et. By substituting eq. (3) into eq. (2)
and integrating, we find
Q =
p
(1 + t)α
1 − p
t1+r
t
Z
0
xr
(1 + x)α dx;
(8)
here we put t = t/τo, and assumed a power law for the expansion, Rp ∝ tr. Note that Q is
a measure of the balance at production rather than a gauge of equipartition: in fact, when
p ≈ 0.5, Q ≈ 1 until t ≤ τo (and later on as well if α = 1 + r). The ratio of magnetic to
-- 6 --
particle energy, instead, can be arbitrarily high if the radiative lifetime of the energetically
important particles is sufficiently shorter than t. On the observational side, eqs. (5) and (7)
give
Q = 1860(cid:18) νb
32 GHz(cid:19)−2/3(cid:18)
νmax
2.5 × 1018 Hz(cid:19)−0.2
d5 :
(9)
the magnitude of Q in the present case, to be compared with Q ≈ 1 for the Crab plerion, is
a quantitative measure of the discrepancy which we want to address. For G11.2 -- 0.3, then,
either p is very close to unity (1 − p < 10−3), or t ≫ 1, (α − 1 − r) > 0. We regard the
former possibility as very unlikely, and in the following we shall limit ourselves only the
latter.
When t ≫ τo, it is appropriate to assume flux conservation for the evolution of the
magnetic field, B ∝ t−2r; furthermore, the plerion spectrum shows two breaks (PS), of
which the higher one, νb, corresponds to the radiative lifetime at the current time t, and
the lower one, νc, is the signature of the radiative lifetime at τo. If we identify the observed
break with νb, eq. (4) holds, and the pressure interior to the plerion is relatively very high.
As we show in the following, the magnetic pressure corresponding to eq. (4) exceeds the
shell pressure by a large factor; this is possible only if the plerion expansion is limited by
the inertia of some cold matter tied to the plerion itself, in which case we expect r ≥ 1.
Then r ≥ 1 and α ≤ 3, as observed in all measured cases, give p of order 0.5 only at the
expense of t ≥ 1000, and an initial magnetic energy of order t × WB > 1052d3
5 erg. We
consider very unlikely a value so much larger than the canonical supernova output, and
propose that: i) the plerion and the shell are in pressure equilibrium, so that r < 1; and
ii) the observed break is to be identified with νc, so that νb = νc × t10r−2 ≫ νo, and Q
is decreased with respect to eq. (9). However, the break at νb entails a change of slope
ǫ = (2.8r − 0.4α)/(5r − 1) (see PS); since G11.2 -- 0.3 shows a straight spectrum from the
microwaves to the X rays, we must require ǫ = 0, α = 7r.
-- 7 --
Within this framework more detailed results can be obtained by coupling the plerion
evolution to the hydrodynamics of the thermal shell. We can distinguish two different cases,
according to the distribution of the ambient density being of the interstellar type (ρo ∼
constant, henceforth IS), or circumstellar type (ρo ∼ r−2, henceforth CS). If the ambient
medium has a constant density, Rs ∝ t2/5, Rp ∝ t3/10, and, because of the condition on ǫ,
α = 2.1. If the shell expands inside the pre-supernova wind, one has Rs ∝ t2/3, Rp ∝ t4/7,
and α = 4.0.
We have computed in detail the two cases, and give the results in Table 1. In particular,
we have adjusted the unknown distance d5 so as to make the total energy of the shell
remnant equal to the canonical 1051 erg. We have chosen the value of νo within the allowed
interval so as to bring our results as close as possible to 'typical' values. And, finally, we
have neglected the work done by the plerion on the shell remnant, which is justified a
posteriori by the relatively small plerion energetics resulting from the computations. The
precise value of the ambient density in the IS and CS cases was computed so as to reproduce
the temperature and flux of the thermal X rays given in VA.
Note that the CS assumption leads to very 'palatable' estimates for all the unknowns,
with one exception: the slowing down exponent α implies a braking index n = 1.67. We
note that such value is still in the acceptable range between a purely dipolar (α = 3, n = 3)
and a purely monopolar geometry (α = ∞, n = 1), but that for most pulsars the braking
index is between 2 and 3. On the other hand, the IS hypothesis carries along some deviation
from perfect balance at production (p ∼ 0.9) and a pulsar magnetic field at the extreme of
the observed range (B∗ = 1.3 1013 G, like in PSR 1509-58). All in all, we regard the latter
interpretation as the most plausible one, also because it places G11.2 -- 0.3 exactly on the
empirical relation between Lp and E given by Seward & Wang (1988); we suggest that the
corresponding values in Table 1 be taken as a reasonable guess for further observing efforts.
-- 8 --
A major improvement in the modelling could ensue if future observations were to give
the precise frequency of the break(s) in the spectrum and the related change(s) of slope, or
the detection of the pulsar.
3. Conclusions
If G11.2 -- 03 is a composite remnant deriving from the historical explosion of SN 386
AD, the main parameters of the nebula and of the central pulsar can be estimated in
the framework of the PS model. They appear to be consistent with the properties of the
remnant and the standard views on pulsars.
The apparent preponderance of the magnetic channel in the output of the central pulsar
is the consequence of the dynamical evolution under the influence of the external shell.
This has slowed down the expansion and therefore reduced the importance of the adiabatic
losses with respect to the synchrotron losses. Since most of the energy was released by the
pulsar during a relatively short initial phase, it is natural that the subsequent evolution
has led to a large value of Q, eq. (8), even if the pulsar output was always well balanced,
p ≈ 0.5. We strongly stress this point as relevant to all sorts of non-thermal sources: only
if the generation of particles and fields keeps occurring for a time scale longer or roughly
equal to the age of the source does Q reflect the balance at production between fields and
particles. This is why Q ≈ 1 in the Crab plerion, for which t ≈ 3 × τo. In addition, and
for quite independent reasons, the Crab is not far from equipartition, since its νb is not far
from the frequency where most of the power is emitted.
This work was partly supported by a Grant of the Italian Space Agency.
-- 9 --
Table 1. Parameters of the Models
Parameter
IS Model
CS Model
νo
d
p
Bp
P
E
B∗
τo
Po
Eo
4 GHz
9.2 kpc
0.9
32 GHz
7.4 kpc
0.3
9.3 × 10−4 G
1.8 × 10−4 G
0.34 s
2.73 s
1.55 × 1037 erg s−1
1.25 × 1036 erg s−1
2.5 × 1013 G
1.6 × 1012 G
5.8 × 108 s
7.0 × 109 s
0.029 s
0.139 s
2.4 × 1049 erg
1.0 × 1048 erg
-- 10 --
REFERENCES
Bandiera, R., Pacini, F., & Salvati, M. 1984, ApJ, 285, 134
Kaspi, V. M., Manchester, R. N., Siegman, B., Johnston, S., & Lyne, A. G. 1994, ApJ, 422,
L83
Morsi, H. W., & Reich, W. 1987, A&AS, 71, 189
Pacini, F., & Salvati, M. 1973, ApJ, 186, 249 (PS)
Reynolds, S. P., & Chevalier, R.A. 1984, ApJ, 278, 630
Sedov, L. I. 1959, Similarity and Dimensional Methods in Mechanics (New York: Academic)
Seward, F. D., & Wang, Z. -- R. 1988, ApJ, 332, 199
Vasisht, G., Aoki, T., Dotani, T., Kulkarni, S. R., & Nagase, F. 1996, ApJ, 456, L59
This manuscript was prepared with the AAS LATEX macros v4.0.
|
0704.2179 | 1 | 0704 | 2007-04-17T19:26:29 | Spitzer spectral line mapping of supernova remnants: I. Basic data and principal component analysis | [
"astro-ph"
] | We report the results of spectroscopic mapping observations carried out toward small (1 x 1 arcmin) regions within the supernova remnants W44, W28, IC443, and 3C391 using the Infrared Spectrograph of the Spitzer Space Telescope. These observations, covering the 5.2 - 37 micron spectral region, have led to the detection of a total of 15 fine structure transitions of Ne+, Ne++, Si+, P+, S, S++, Cl+, Fe+, and Fe++; the S(0) - S(7) pure rotational lines of molecular hydrogen; and the R(3) and R(4) transitions of hydrogen deuteride. In addition to these 25 spectral lines, the 6.2, 7.7, 8.6, 11.3 and 12.6 micron PAH emission bands were also observed. Most of the detected line transitions have proven strong enough to map in several sources, providing a comprehensive picture of the relative distribution of the various line emissions observable in the Spitzer/IRS bandpass. A principal component analysis of the spectral line maps reveals that the observed emission lines fall into five distinct groups, each of which may exhibit a distinct spatial distribution: (1) lines of S and H2 (J > 2); (2) the H2 S(0) line; (3) lines of ions with appearance potentials less than 13.6 eV; (4) lines of ions with appearance potentials greater than 13.6 eV, not including S++; (5) lines of S++. Lines of group (1) likely originate in molecular material subject to a slow, nondissociative shock that is driven by the overpressure within the supernova remnant, and lines in groups (3) - (5) are associated primarily with dissociative shock fronts with a range of (larger) shock velocities. The H2 S(0) line shows a low-density diffuse emission component, and - in some sources - a shock-excited component. | astro-ph | astro-ph | Spitzer spectral line mapping of supernova remnants:
I. Basic data and principal component analysis
David A. Neufeld1, David J. Hollenbach2, Michael J. Kaufman3, Ronald L. Snell4,
Gary J. Melnick5, Edwin A. Bergin6, and Paule Sonnentrucker1
ABSTRACT
We report the results of spectroscopic mapping observations carried out to-
ward small (1′ × 1′) regions within the supernova remnants W44, W28, IC443,
and 3C391 using the Infrared Spectrograph of the Spitzer Space Telescope. These
observations, covering the 5.2 − 37 µm spectral region, have led to the detection
of a total of 15 fine structure transitions of Ne+, Ne++, Si+, P+, S, S++, Cl+,
Fe+, and Fe++; the S(0) – S(7) pure rotational lines of molecular hydrogen; and
the R(3) and R(4) transitions of hydrogen deuteride.
In addition to these 25
spectral lines, the 6.2, 7.7, 8.6, 11.3 and 12.6 µm PAH emission bands were also
observed. Most of the detected line transitions have proven strong enough to map
in several sources, providing a comprehensive picture of the relative distribution
of the various line emissions observable in the Spitzer/IRS bandpass. A principal
component analysis of the spectral line maps reveals that the observed emission
lines fall into five distinct groups, each of which may exhibit a distinct spatial
distribution: (1) lines of S and H2 (J > 2); (2) the H2 S(0) line; (3) lines of ions
with appearance potentials less than 13.6 eV; (4) lines of ions with appearance
potentials greater than 13.6 eV, not including S++; (5) lines of S++. Lines of
group (1) likely originate in molecular material subject to a slow, nondissociative
shock that is driven by the overpressure within the supernova remnant, and lines
in groups (3) – (5) are associated primarily with dissociative shock fronts with a
range of (larger) shock velocities. The H2 S(0) line shows a low-density diffuse
emission component, and – in some sources – a shock-excited component.
7
0
0
2
r
p
A
7
1
]
h
p
-
o
r
t
s
a
[
1
v
9
7
1
2
.
4
0
7
0
:
v
i
X
r
a
1Department of Physics and Astronomy, Johns Hopkins University, 3400 North Charles Street, Baltimore,
MD 21218
2NASA Ames Research Center, Moffett Field, CA 94035
3Department of Physics, San Jose State University, 1 Washington Square, San Jose, CA 95192
4Department of Astronomy, University of Massachusetts, 710 North Pleasant Street, Amherst, MA 01003
5Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138
6Department of Astronomy, University of Michigan, 825 Dennison Building, Ann Arbor, MI 48109
– 2 –
Subject headings: ISM: Molecules - ISM: Abundances - ISM: Clouds – molec-
ular processes – shock waves
1.
Introduction
Supernovae have long been recognized as a critical source of energy input to the ISM (e.g.
Cox & Smith 1974; McKee & Ostriker 1977). They send shock waves propagating through
the ISM, creating large cavities filled with hot ionized material. Eventually, supernova-driven
shock waves become radiative, emitting strong line emissions – primarily at optical and ul-
traviolet wavelengths – that have been widely observed from supernova remnants (e.g. Weiler
& Sramek 1988). Where supernova remnants encounter molecular clouds, they drive slower
shock waves that compress and heat the molecular material and result in strong infrared line
emission. Depending upon the shock velocity, such shock waves may be either dissociative –
resulting in the destruction of molecules by collisional processes – or nondissociative, in that
molecules are heated but nevertheless survive the passage of the shock front. The theory
of interstellar shock waves, both in atomic and molecular media, has been the subject of
extensive investigation (reviewed, for example, by Draine & McKee 1993).
Using the Infrared Spectrograph (IRS) of the Spitzer Space Telescope, we have carried
out infrared spectroscopic observations of four regions in which a supernova remnant interacts
with a molecular cloud: W44, W28, 3C391, and IC443. The IRS covers the 5.2 – 37 µm
region with unprecedented sensitivity, providing access to the lowest eight pure rotational
transitions of molecular hydrogen together with fine structure lines of [SI] and a variety of
In the observations reported here, IRS spectroscopy indicated the presence
atomic ions.
of interstellar gas over a huge temperature range, from <
∼ 100 K (H2 S(0) 28 µm line) to
∼ 105 K ([NeIII] 15.6 and 36 µm lines).
>
In this paper, we present a general overview of the data set we have obtained toward
these four supernova remnants. In §2, below, we give a brief overview of the four supernova
remnants selected for this study. In §3, we discuss the observations and methods used to
reduce the observational data, and we present spectral line maps for a representative sample
of the transitions that we detected. In §4, we describe how Principal Component Analysis can
be used to identify distinct groups of transitions that share a common origin in a particular
component of the interstellar gas. In §5, we discuss the similarities and differences in the
spatial distributions of the various spectral lines we observed. More detailed analyses of the
different shocked gas components will follow in future papers.
– 3 –
2. Source selection
The four supernova remnants we have observed – W44, W28, 3C391 and IC443 – repre-
sent classic examples of the interaction between supernova-driven shock waves and molecular
clouds. Brief descriptions of each source are given below, and finder charts appear in Figures
1 – 4. In each source, we have centered the mapped region at a position that is known from
previous observations to be a luminous source of molecular emissions; these are all positions
that have been observed previously by the Short- and/or Long Wavelength Spectrometer of
the Infrared Space Observatory (ISO). All four sources lie close to the Galactic plane, with the
mapped regions lying at Galactic latitudes b ∼ −0.43◦, −0.30◦, 0.03◦ and 3.02◦ respectively
for W44, W28, 3C391 and IC443 (see Table 1).
2.1. W44
W44 lies at a distance of 2.5 kpc and has a radius of 11 – 13 pc (Cox et al. 1999). The
remnant is associated with the pulsar B1853+01, which is believed to be the progenitor star
and provides an age of 20,000 years (Wolszczan et al. 1991; based upon the ratio of the pulsar
period to the period derivative). W44 is a prototype for a new class of supernova remnants
called "mixed-morphology" SNRs, because the radio continuum images exhibit a shell-like
morphology while the center is filled in with X-ray emission (Jones et al. 1993; Rho & Petre
1998). Such morphologies have been attributed to the interaction between the supernova
shock front and a molecular cloud (Chevalier 1999; Yusef-Zadeh et al. 2003), and W44 is
indeed associated with a Giant Molecular Cloud, recently mapped in CO emission (Reach,
Rho, & Jarrett 2005: hereafter RRJ05).
RRJ05 and Cox et al. (1999) provide nice summaries of the available observations. In
general, there is strong evidence for supernova interaction with a dense molecular cloud
through the detection of OH 1720 MHz maser emission, broad (FWHM ∼ 20 – 30 km/s)
CO molecular line emission, vibrational and rotational lines of molecular hydrogen, and
atomic fine structure lines (Claussen et al. 1997; Seta et al. 1998; Reach & Rho 2000;
RRJ05). Comparison to shock models are consistent with a mixture of dissociative and
non-dissociative shocks impacting on gas with densities of 100 − 104 cm−3 (Reach & Rho
2000; RRJ05), although Cox et al. (1999) present a model that matches most data but
requires no dense cloud interaction. The region we have observed with Spitzer (Figure 1)
lies at the northeast edge of the W44 supernova remnant, coincident with OH maser Region
E (Claussen et al. 1997). Observations of CO and X-rays imply total hydrogen column
densities of 2 × 1022 cm−2 along the sight-line (Rho et al. 1994; RRJ05), corresponding to a
visual extinction AV ∼ 10 mag.
– 4 –
2.2. W28
The supernova remnant W28 is located in a complex region of the Galactic disk where a
number of nearby large H II regions (M8 and M20) and young clusters (e.g., NGC 6530) are
present (Goudis 1976). Kinematic considerations based on molecular line broadening place
the remnant at a distance of 1.9 ± 0.3 kpc (Vel´azquez et al. 2002). The distribution of radio
molecular emission lines indicates that the remnant has an irregular shell-like structure with
prominent peak emissions in the northern and northeast sides of the shell where an interaction
between the remnant and a nearby molecular cloud was suggested (Wootten 1981). Further
evidence for such an interaction arises from the detection of numerous discrete OH 1720
MHz maser spots along the northeast ridge where the remnant is believed to be interacting
with a molecular cloud (Claussen et al. 1997). This maser activity has been shown to
constitute a tracer of supernova remnant-molecular cloud interaction (Yusef-Zadeh et al.
2003). Detection of the H2 S(3) and S(9) lines as well as a number of ionic species with
ISO suggests the presence of shocks in the northern part of the remnant (Reach & Rho
2000) as well. In the optical, short emission filaments are observed superposed over a more
diffuse and centrally-peaked emission seen throughout most of the interior of W28 (van den
Bergh 1973; Long et al. 1991). Diffuse thermal X-ray emission is also seen throughout
the region delineated by the radio shells and shows a rather patchy but centrally-peaked
emission distribution (Dubner et al. 2000; Rho & Borkowski 2002). Like W44, W28 shows a
"mixed-morphology" that combines a shell-like radio structure with a centrally-filled diffuse
X-ray emission structure. Observations of optical [SII] line ratios (Long et al. 1991) suggest
a reddening of E(B − V ) ∼ 1 − 1.3 mag, corresponding to a visual extinction of 3 − 4 mag
for an AV /E(B − V ) ratio of 3.1.
2.3.
3C391
3C391 (G31.9+0.0) is an X-ray- and radio-bright supernova remnant that lies in the
Galactic plane. The X-ray emission is diffuse (i.e., no limb brightening) and centered within
the radio shell (Rho & Petre 1996). 12CO J = 1−0 maps toward 3C391 reveal the presence
of moderately dense (n(H2) ∼ 300 – 1000 cm−3) molecular gas along the northwest portion
of the remnant (Wilner, Reynolds & Moffett 1998). The radio continuum emission from the
remnant (Reynolds & Moffett 1993), shown in Fig. 3 with the region mapped by Spitzer
superposed, suggests evolution along a strong density gradient, with compression of the
supernova blast wave to the northwest in the direction of the strongest CO emission and
lower continuum surface brightness emission to the southeast away from the molecular gas. It
is likely that the progenitor star exploded within a dense molecular cloud and the expanding
– 5 –
gas has now broken out of the cloud to the southeast.
Evidence exists for both dissociative and nondissociative shocks toward 3C391 (Reach
et al. 2002). Narrowband filter images in the [Fe II] 1.64 µm line show a pronounced intensity
peak along the bright radio bar (18h 49m 16s, –0o 55′ 00′′) at the interface between the su-
pernova remnant and the molecular cloud. Similarly, the ISOCAM CVF 5 – 18 µm spectrum
of the same region is dominated by emission lines from atomic ions – such as [Fe II] 5.5 µm,
[Ar II] 6.9 µm, [Ne II] 12.8 µm, and [Ne III] 15.5 µm – while the mid-infrared H2 rotational
lines in the same spectrum are either weak or absent (Reach et al. 2002). Approximately
2.8′ to the south and east of this position, near α =18h 49m 22s, δ = −0o 57′ 22′′ (J2000),
continuum-subtracted H2 2.12 µm images show an ∼ 30′′ diameter area containing a cluster
of small (< 5′′) clumps of H2 emission. The ISOCAM CVF spectrum of this region shows
strong H2 S(2) through S(7) H2 line emission (Reach et al. 2002). Longer wavelength ISO
LWS spectra toward this location show thermal H2O, OH, high-J 12CO (Reach & Rho 1998)
and [O I] 63 µm (Reach & Rho 1996) emission. This so-called broad molecular line region
(see Reach & Rho 1999) is also the site of OH 1720 MHz maser (Frail et al. 1996) and broad
(FWHM ∼ 20 km s−1) 12CO (J = 2 − 1), CS (J = 3 − 2), CS (J = 5 − 4), and HCO+
(J = 1 − 0) emission (Reach & Rho 1999). The existence of strong and broad molecular
line emission has led to the conjecture that the supernova blast wave is encountering dense
(n(H2) ∼ 104 – 105 cm−3) cores present within the molecular cloud from which the progenitor
of 3C391 formed. The region that we have mapped with Spitzer (Figure 3) is centered on
the broad molecular line region.
The extinction to 3C391 is very high due to its location in the Galactic plane and its
9 kpc distance (Caswell et al. 1971; Radhakrishnan et al. 1972). Column densities inferred
from the spectral analysis of ROSAT data yield a foreground column density of 2 – 3.6 ×
1022 cm−2 (Rho & Petre 1996). Since this analysis has been applied by averaging spectra
obtained over relatively large areas of the remnant, it is likely that the higher value applies
to the denser broad molecular line region of interest here. Thus, assuming a foreground
column density of 3.6 × 1022 cm−2, a line-of-sight visual extinction of Av = 19 magnitudes
is derived (Reach et al. 2002).
2.4.
IC443
IC443 is a well-studied SNR located in the outer Galaxy at a distance of approximately
1.5 kpc (Georgelin 1975; Fesen & Kirshner 1980; Fesen 1984). It is one of the more striking
examples of the interaction of a SNR with the interstellar molecular medium. IC443C, the
position that we have targeted with Spitzer, is one of several shocked gas regions within the
– 6 –
SNR first identified by DeNoyer (1978) based on the presence of high velocity HI emission.
These shocked gas regions were later detected to have broad OH absorption and CO emission
lines (DeNoyer 1979a,b). More extensive observations of CO and HCO+ emission in IC443
by Dickman et al. (1992) revealed many more regions of broad molecular line emission and
showed that they formed an expanding, tilted ring. This ring of shocked gas presumably
delineates where the SNR shock has encountered the molecular interstellar material. This
ring is also seen in 2 µm H2 emission (Burton et al. 1988; Richter et al. 1995b).
Based upon the kinematics of the observed CO and HCO+ emissions, IC443C is believed
to be located on the front side of the expanding ring (Dickman et al. 1992). The unshocked
molecular gas in this direction has a vLSR of –3 km s−1. The broad velocity emission from
the shocked molecular gas extends from the velocity of the unshocked gas to a vLSR of –70
km s−1 (Snell et al. 2005). The region of most intense molecular broad-line emission is about
1 arcminute in size.
Emission from the shocked interstellar gas in IC443 has been extensively studied and
detected in numerous atomic and molecular tracers, including HI (Braun & Strom 1986)
vibrationally excited H2 (Burton et al. 1988), the 63 µm line of [OI] (Burton et al. 1990), the
pure rotational lines of H2 (Richter et al. 1995a; Cesarsky et al. 1999) and numerous molecular
species (DeNoyer & Frerking 1981; Ziurys et al. 1989; van Dishoeck et al. 1993). Most
recently, the detection of water was reported in several of these shocked regions, including
IC443C (Snell et al. 2005). Most studies have concluded that a combination of shocks are
needed to explain the emission seen in these atomic and molecular tracers. Snell et al. (2005)
suggested the presence in IC443C of both a fast J-shock and either a slow C-shock or slow
J-shock. Given the observed ratio of H2 vibrational lines, Richter et al. (1995b) derived 2.1
µm extinctions of 1.3 and 1.6 mag for two slit positions in IC443C, corresponding to visual
extinctions AV of 12 and 15 mag.
3. Observations and results
W44, W28, 3C391, and IC443C were each observed using the IRS as part of a Cycle
2 General Observer Program, the Short-Low (SL), Short-High (SH) and Long-High (LH)
modules being used to obtain complete spectral coverage over the wavelength range available
to IRS. The spectral resolving power of IRS, R = λ/∆λ, ranges from 60 to 127 for the SL
module and is ∼ 600 for the SH and LH modules. Spectral line maps of size ∼ 1′ × 1′ were
obtained by stepping the slit perpendicular to its long axis by one-half its width, and – in
the case of SH and LH – parallel to its long axis by 4/5 of its length. Table 1 summarizes
the observational details for each source; note that the map center positions are not in all
– 7 –
cases identical to the (0,0) positions defined in the captions to Figures 10 – 13.
The data were processed at the Spitzer Science Center (SSC), using version 13.2 of the
processing pipeline, and then reduced further using the SMART software package (Higdon
et al. 2004), supplemented by additional routines (Neufeld et al. 2006a), that provide for the
removal of bad pixels in the LH and SH data, the calibration of fluxes obtained for extended
sources, the extraction of individual spectra at each position sampled by the IRS slit, and the
creation of spectral line maps from the set of extracted spectra. No background subtraction
could be performed, since no off-source measurements were made; while background sub-
traction might have affected the measured continuum fluxes, it would be expected to have
no effect upon the derived intensities for spectral lines. There is no evidence for fringing in
any of the spectra.
For several transitions observed in the LH and SH modules, the resultant line maps
exhibited striping perpendicular to the long axis of the slit. We determined this effect to
be an artifact created by the standard algorithm used in the IRSCLEAN software to obtain
interpolated data for bad pixels; that algorithm interpolates in the dispersion direction,
causing a systematic underestimate in the intensity when a bad pixel falls close to the
central wavelength of a spectral line. The striping was successfully removed by applying a
correction factor, computed separately for each position along the slit and for each affected
spectral line. The required correction factors were obtained by considering – separately for
each of the five positions along the slit – the cumulative distribution functions for the line
fluxes obtained over the entire map. In other words, for each five possible positions along
the slit, we plotted the fraction of line flux measurements that were smaller than a given
value, as a function of that value. Making the assumption that the cumulative distribution
functions should be similar except for the highest flux values, we computed the necessary
correction factors needed to bring the five distribution functions into agreement.
In order to improve the signal-to-noise ratio beyond that obtained at a single position,
we have computed average spectra for each of the sources we observed. The resultant spec-
tra, plotted in Figures 5 – 8, are the averages of all spectra within a circular region, the
contributing spectra being weighted by a Gaussian taper (HPBW of 25′′) from the center of
each synthesized beam. The central positions are given in the figure captions.
The spectra plotted in Figures 5 – 8 reveal a total of 25 spectral lines, detectable in
one or more source. These include 15 fine structure transitions of the species Ne+, Ne++ (2
transitions), Si+, P+, S, S++ (2 transitions), Cl+, Fe+ (5 transitions), and Fe++; 8 transitions
of H2 (the S(0) through S(7) pure rotational transitions), and the R(3) and R(4) transitions
of HD (detected toward IC443C and reported previously by Neufeld et al. 2006b). Beam
averaged intensities are listed in Table 2. In addition to these emission lines, the 6.2, 7.7,
– 8 –
8.6, 11.3 and 12.6 µm PAH emission bands are readily apparent in the spectra of W44, W28,
and 3C391, and the 11.3 µm micron feature is detected in IC443C. A discussion of the PAH
emission features is deferred to a future paper.
The observed line transitions represent a wide variety of excitation conditions. In Figure
9, we show their position in a scatter plot, where the horizontal axis shows the appearance
potential – equal to zero for neutral species and equal to the ionization potential of X(n−1)+
for the case of ion Xn+ – and the vertical axis shows the critical density at which the rate
of collisional de-excitation equals the spontaneous radiative rate. In computing the critical
density, we assumed the dominant collision partner to be electrons for ionized species and
hydrogen molecules for neutral species.
In Figures 10 – 13, again for each source, we present maps of a representative selection
of six strong spectral lines: the H2 S(0), [SI] 25 µm, H2 S(2), [SiII] 35 µm, [NeII] 12.8 µm,
and [SIII] 33 µm lines. As demonstrated by the Principal Component Analysis described
in §4 below, each of the other transitions shows a spatial variation that is almost identical
to one of the transitions plotted in Figures 10 – 13. For example, except for a constant
scaling factor, the [FeII] 26 µm line maps are visually indistinguishable from those shown
here for [SiII] 35 µm. For the maps obtained with the SL module, the effective integration
time and resultant signal-to-noise ratio is the same everywhere in the map. For the SH and
LH modules, the effective integration time is constant, except in overlap regions between
adjacent pointings where the integration time is larger by a factor two.
4. Principal Component Analysis
A cursory inspection of Figures 10 – 13 reveals both similarities and differences in the
spatial distributions of the various emission lines that we have observed. The method of Prin-
cipal Component Analysis (PCA) provides a valuable tool for describing those similarities
and differences. The value of PCA in analysing multitransition maps was first demonstrated
by Ungerechts et al. (1997), who applied the method to molecular line maps of the Orion
Molecular Cloud and gave a comprehensive description of PCA that will not be repeated
here (see also Heyer & Schloerb 1997). To summarize, PCA creates an orthogonal basis
set of maps – known as principal components (PC) – such that each observed map can be
represented as a linear combination of the PC. The key feature of the new basis set is that
the PC are obtained in decreasing order of amplitude: the first PC contains as much of the
spatial variation as possible, the second PC - constructed orthogonal to the first - contains
as much of the remaining variation as possible, and so forth.
– 9 –
Following Ungerechts et al. (1997), we applied the Principal Component Analysis to
spectral line maps that had been mean-subtracted and divided by the standard deviation,
the latter being dominated by structure in the source rather than instrumental noise. This
procedure gives equal weighting to each line instead of allowing the brightest lines to domi-
nate the analysis. Because this analysis involves data obtained with all three modules (i.e.
SL, SH, and LH), it was applied only to the intersection of the regions mapped by the dif-
ferent modules. Applying PCA to the four sources we have observed, we find that the first
PC contains 58 − 73% of the total spatial variation in the maps, the first and second PCs
contain together 82 − 88%, and the first three PCs contain 90 − 93% of the information. In
Figures 14 – 17 we present the results of the PCA for each source. The upper panel shows
the fraction of the variance accounted for by each component (asterisks), along with the
cumulative fraction for the first n components as a function of n (open squares). The lower
panels show the coefficients for the 1st, 2nd and 3rd PC needed to approximate the maps of
each transition; the coefficients for the 1st and 2nd PC are shown in the left lower panel and
those for the 2nd and 3rd PC are shown in the right lower panel. (These figures were referred
to as h-plots by Ungerechts et al. 1997.) Because the PCA is applied to mean-subtracted
line maps, the coefficients may be either positive or negative. While the 3rd PC contains
information about the source – as evidenced by the fact that transitions with similar coeffi-
cients for the 2nd PC tend to show similar coefficients for the 3rd – the 4th and higher PCs
are likely dominated by noise.
The principal components are normalized such that quadrature sum of the coefficients
for each transition is unity. Thus, if a particular transition were completely accounted for by
the 1st and 2nd PC, its position in the lower left plot would lie on the circle of unit radius;
similarly, if a particular transition were completely accounted for by the 2nd and 3rd PC,
its position in the lower right plot would lie on the circle of unit radius. The proximity of
most points in the lower left plot to the circle of unit radius is a reflection of the fact that
the first two PCs account for ∼ 85% of the variance in the observed line maps. In the limit
where two points actually lie on the unit circle, the cosine of the angle between the two radii
measures the degree of linear correlation; points lying next to each other on the unit circle
would be perfectly correlated (cos θ = +1), while those lying exactly on opposite sides of the
circle would be perfectly anticorrelated (cos θ = −1). For example, in W44, the lower left
panel indicates that the [FeII] 26 µm and [SiII] transitions are very well correlated with each
other, strongly anticorrelated with both observed [SIII] transitions, and almost uncorrelated
with the H2 S(0) transition.
In W44, the PCA yields a clear separation into five groups of spectral lines: (1) lines of
S, and H2 (J > 2); (2) the H2 S(0) line; (3) lines of ions with appearance potentials less than
13.6 eV (Fe+, Si+, P+); (4) lines of ions with appearance potentials greater than 13.6 eV
– 10 –
(Ne+, Ne++, Fe++), not including S++; (5) lines of S++. In the other sources, some of these
groups share a common distribution, but in no case do the spectral lines within a group
defined above fail to be strongly correlated with each other. In W28, lines in groups (1) and
(2) share a similar distribution, which is markedly different from that of groups (3) – (5). In
3C391, lines in groups (1), (2), (3) and (5) show distributions that are readily distinguishable
from each other, but group (4) shows a distribution that is very similar to that of group (3).
In IC443, groups (1), (3), and (4) are readily distiguishable from each other; group (2) (the
H2 S(0) line) has a similar distribution to that of group (1); and group (5) ([SIII] transitions)
is not detected. These behaviors suggest that the line emissions we have observed originate
in at least five physically distinct components (some of which have similar distributions – as
projected onto the sky – in sources other than W44). The implications of these groupings
are discussed further in §5 below.
5. Discussion
5.1. Line emission from neutral species: H2 and atomic sulphur
Each of the sources we observed shows strong emission in the H2 S(0) through S(7) lines.
In Figure 18, we present rotational diagrams derived from the H2 line fluxes listed in Table
1. Here we corrected for extinction, adopting the foreground extinction estimates given
in §2 and the interstellar extinction curves of Weingartner & Draine (2001; "Milky Way,
RV = 3.1"). The rotational diagrams bear a strong similarity to those obtained previously
from Spitzer observations of the Herbig-Haro objects HH54 and HH7 (Neufeld et al. 2006a).
In each case, a positive curvature in the rotational diagrams indicates the presence of an
admixture of gas temperatures, while a characteristic zigzag pattern – more pronounced for
low-lying transitions – indicates a nonequilibrium ortho-to-para ratio (OPR), smaller1 than
3. In Neufeld et al. (2006a), we argued that the nonequilibrium H2 ortho-to-para ratios were
consistent with shock models in which the gas is warm for a time period shorter than that
required for reactive collisions between H and para-H2 to establish an equilibrium OPR.
As in Neufeld et al. (2006a), we obtained a two-component fit to the rotational diagrams,
in which a warm component with temperature, Tw, ortho-to-para ratio, OPRw, and column
density Nw coexists with a hot component with temperature, Th, ortho-to-para ratio, OPRh,
and column density Nh. The resultant parameters, listed in Table 3, are broadly similar to
1Except in IC443C, where – given the uncertainties – the data are consistent with an equilibrium OPR
of 3.
– 11 –
those obtained for HH7 and HH54, and are consistent with the presence of nondissociative
shocks2 of velocity 10 – 20 km s−1. Our two-component fits are based upon a simple LTE (lo-
cal thermodynamic equilibrium) treatment of the H2 excitation, a treatment that is accurate
for the lower-lying transitions but may be an oversimplification in the case of the highest
transitions (S(6) and S(7)). In a future paper, we will present a refined analysis, including
non-LTE effects, together with maps of the various physical parameters derived from the H2
(and, for IC443C, the HD) maps.
While the H2 S(1) through S(7) lines show a clear morphological similarity to each
other, a result that is apparent from inspection of the spectral line maps and is confirmed
by the PCA, the H2 S(0) line is typically more extended than – and, in W44 and 3C391,
essentially uncorrelated with – the other H2 rotational lines. The [SI] 25 µm line, by contrast,
is always well correlated with the H2 S(1) through S(7) lines, despite the fact that its upper
state energy is very similar to that of H2 S(0). These behaviors are demonstrated clearly
in Figure 19, which present scatter diagrams showing correlations between [SI] and H2 S(0).
Each point in Figure 19 shows the mean intensities within a 5′′ × 5′′ square region. The
results shown for W28 are very revealing: while [SI] and H2 S(0) are well correlated in W28
(with a linear correlation coefficient of 0.88), the best-fit linear regression yields a positive
x-intercept, indicating that there is an extended component of H2 S(0) emission (with typical
intensity ∼ 2.5 × 10−5erg cm−2 s−1 sr−1) that is unassociated with measurable [SI] emission.
Figure 19 suggests an explanation for why [SI] 25 µm and H2 S(0) are reasonably well-
correlated in IC433C and W28 but not in W44 and 3C391: in W28 and IC443C, the shock
excited contribution dominates the variations in the extended component, whereas in W44
and 3C391 it does not.
The much lower [SI]/H2 S(0) line ratio in the extended component could indicate a much
lower abundance of atomic sulfur, or - more likely - a lower gas density, the critical density for
the [SI] 25 µm line being much greater than that of the H2 S(0) line (see Fig. 9). The extended
H2 S(0) emission in these sources has an intensity comparable to that measured by Falgarone
et al. (2005), who observed H2 pure rotational emission along a long sight-line within the
Galaxy that did not intersect any region of high-mass star formation. Falgarone et al. argued
that the observed H2 emission originated in a warm gas component (unassociated with
supernova remnants or regions of star formation) that is present within the cold, quiescent
2Given the magnetic field strengths typical of the interstellar medium, such shocks are expected to be of
"continuous" or "C-type" in the designation of Draine (1980), and are qualitatively different from the "jump"
or "J-type" shock fronts associated with faster, dissociative shock waves. C-type shocks are characterized by
continuous variations in the gas velocity, large ion-neutral drift velocities, and lower peak gas temperatures
than those attained behind J-type shocks of the same velocity.
– 12 –
ISM. Alternatively, or in addition, the H2 S(0) emission within supernova remnants might
originate within low density photodissociation regions that are irradiated and heated by
UV radiation from nearby fast shocks.
In contrast to the continuum radiation incident
upon typical PDRs, the UV radiation emitted by fast shocks is dominated by spectral lines.
Although detailed calculations will be needed to model PDRs that are irradiated by UV
emission from fast shocks, we speculate that a line-dominated spectrum may be particularly
effective in exciting H2 rotational emissions since – except for spectral lines that happen
to coincide with an H2 Lyman or Werner band transition – line radiation can heat the gas
without photodissociating molecules.
It is also noteworthy that the [SI] emission is better correlated with the warm H2 than
with the ionized species in these sources. Haas et al. (1991) had previously suggested that
[SI] was a tracer of fast dissociative shocks, rather than the slower nondissociative shocks
responsible for the H2 emissions. At least in the supernova remnants that we have mapped
with Spitzer, the PCA strongly suggests that [SI] traces slow nondissociative shocks.
The atomic sulfur abundance can be estimated from the [SI] 25 µm line flux. Unfortu-
nately, the critical density for the [SI] 25 µm line is probably large in neutral gas (∼ 106 cm−3
according to Hollenbach & McKee 1989, Table 5), so the derived SI abundance depends on
both the gas density and the adopted rate coefficient for excitation of SI in collisions with
H2. IC443C presents the best case for determining the atomic sulfur abundance, because
a measurement of the HD R(4)/R(3) line allows the gas pressure to be determined for this
case (Neufeld et al. 2006b). Adopting the H2 column densities and temperatures obtained
from a two-component fit to the rotational diagram, assuming the warm and hot gas compo-
nents to be at a common pressure adjusted to match the HD R(4)/R(3) line ratio (Neufeld
et al. 2006b), and making use of Hollenbach & McKee's (1989; hereafter HM89) estimate
of the rate coefficient for excitation of [SI] fine structure collisions3, we derive an atomic
sulfur abundance of ∼ 10−6 relative to H2. This value corresponds to ∼ 4% of the solar
abundance of sulfur (1.4 × 10−5 relative to H nuclei; Asplund, Grevesse, & Sauval 2005).
The uncertainty in this estimate is large, not least because of the absence of a reliable esti-
mate for the excitation rate coefficient. Owing to the presence of a significant UV radiation
field produced by fast dissociative shocks in the vicinity of the warm molecular component,
Snell et al. (2005) argued that sulfur could be significantly ionized within the nondissociative
molecular shocks that are responsible for the pure rotational emissions from H2; thus the
observed SI abundance might be entirely consistent with a negligible depletion of sulfur onto
grains and a negligible fraction of the sulfur nuclei being bound within gas-phase molecules.
3This estimate is actually for excitation by atomic hydrogen, but – since we are not aware of any estimate
in the literature for collisional excitation of SI by H2 – we adopt it for excitation by H2 as well.
– 13 –
Further discussion of the atomic sulfur abundance, and its implications for the chemistry of
interstellar sulfur molecules, is deferred to a future paper.
5.2. Line emission from ionized species
The PCA shows a very tight correlation between the [FeII], [PII] and [SiII] line intensities
in all the regions we observed, a result confirmed by the correlation plots shown in Figures 20
([FeII] versus [SiII] intensity) and 21 ([PII] versus [SiII] intensity). These close correlations
are unsurprising, given the similar appearance potentials (all < 13.6 eV) of Fe+, P+ and Si+:
all three ions are expected behind shocks that are fast enough to cause significant ionization,
and can also be produced in the ionizing "precursors" of fast shocks, where ultraviolet
radiation propagates upstream and can photoionize the gas before it even enters the shock
front. For shocks propagating in molecular gas, the calculations of HM89 suggest that a
shock velocity of at least 35 kms−1 is required to produce detectable fine structure emissions
from [FeII] and [SiII].
One additional feature of the data, however, which is not revealed by our PCA of mean-
subtracted line maps, is that the correlation plots typically show positive intercepts on the
horizontal [SiII] axis. As in the case of H2 S(0), this behavior implies the presence of an
extended gas component within which [SiII] is enhanced relative to [FeII] and [PII].
Four ions of appearance potential greater than 13.6 eV are detected in one or more
region: Ne+, Ne++, Fe++ and S++. Such ions, which have been observed widely in supernova
remnants (e.g. Raymond et al. 1997), are produced only in faster shocks. For example, when
∼ 80 kms−1 are needed to ionize Ne
shocks propagate in molecular gas clouds, velocities >
(HM89). These lines are generally well-correlated with each other, with the curious exception
of the two [SIII] lines (19 and 33 µm); in W44, the [SIII] lines are anticorrelated with the
other emissions from ionized species. We speculate that this behavior could represent an
excitation effect, the critical density for the [SIII] transitions being smaller than those for
[NeII] and [NeIII]. The [SIII] 18.7/33.5 µm and [NeIII] 15.6/36.0 µm line ratios both provide
useful constraints on the electron density in the emitting region (Alexander et al. 1999). The
observed [SIII] 18.7/33.5 µm line ratio was 0.78, 0.54 and 0.55 respectively in W44, W28,
and 3C391, the three sources in which [SIII] emissions were detected. All three values are
consistent with the low density limit obtained by Alexander et al. (1999; 0.6 at a temperature
of 104 K, or 0.7 at a temperature of 2 × 104 K). Conservatively estimating the likely error
on the ratio as ±35%, assuming that the gas temperature is at least 104 K, and making use
of the results presented by Alexander et al. (1999; their Figure 3), we obtain upper limits
on the electron density of 600, 200, and 200 cm−3 for W44, W28, and 3C391. For all three
– 14 –
sources, our best estimate of the [SIII] 18.7/33.5 µm line ratio is somewhat smaller than – but
probably consistent with – that obtained by Oliva et al. (1989) from ISO observations of the
young supernova remnant RCW 103 (0.89 ± 0.18). The observed [NeIII] 15.6/36.0 µm ratio
was ∼ 11 in both W28 and 3C391 (and was consistent with 11 in W44 and IC443C where
the 36.0 µm transition was not detected). Once again, this value is similar to that observed
by Oliva et al. (1989) toward RCW 103 (10 ± 2), and is consistent with the low density limit
obtained by Alexander et al. (1999) (11 at temperature of both 104 K and 2 × 104 K), but
in this case the implied upper limit on the electron density is ∼ 104 cm−3 for both W44 and
3C391. Thus the observed 18.7/33.5 µm and [NeIII] 15.6/36.0 µm line ratios are entirely
consistent with a picture in which the [SIII] emissions originate in a spatially distinct gas
component of lower density than that responsible for the observed [NeIII] emissions. In a
future paper, we will present a more detailed analysis of the emission from ionized species
in these sources, in the context of theoretical models for fast shocks and PDRs.
6. Summary
1) We have carried out spectroscopic mapping observations, covering the 5.2−37 µm spectral
region, toward the supernova remnants W44, W28, IC443C, and 3C391, with the use of the
Infrared Spectrograph (IRS) of the Spitzer Space Telescope. In each case, a region ∼ 1′ × 1′
has been mapped. Except in 3C391, which is more distant and has a much smaller angular
size than the other objects we observed, the mapped regions cover less than 1 percent of the
total radio-emitting area of the supernova remnant.
2) These observations, performed using the Short-Low, Short-High and Long-High modules
of the IRS, have led to the detection of a total of 15 fine structure transitions of Ne+, Ne++
(2 transitions), Si+, P+, S, S++ (2 transitions), Cl+, Fe+ (5 transitions), and Fe++; the S(0)
– S(7) pure rotational lines of molecular hydrogen; and the R(3) and R(4) transitions of
hydrogen deuteride (reported previously by Neufeld et al. 2006b).
3) We have performed a principal component analysis (PCA) of the spectral line maps
obtained from our observations, with the goal of characterizing the differences and similarities
between the spatial distributions of the various line emissions. In W44, the PCA reveals that
the observed emission lines fall into five distinct groups, each of which exhibits a distinct
spatial distribution : (1) lines of S and H2 (J > 2); (2) the H2 S(0) line; (3) lines of ions with
appearance potentials less than 13.6 eV; (4) lines of ions with appearance potentials greater
than 13.6 eV, not including S++; (5) lines of S++. In the other sources, some of these groups
share a common distribution, but in no case do the spectral lines within a group defined
above fail to be strongly correlated with each other. This behavior suggests that the line
– 15 –
emissions we have observed originate in at least five physically distinct components (some of
which have similar distributions – as projected onto the sky – in sources other than W44).
4) The observed lines of S and H2 (J > 2) likely originate in molecular material subject to a
slow, nondissociative shock that is driven by the overpressure within the supernova remnant.
For the case of IC443C, where the gas pressure is constrained by the HD R(4)/R(3) ratio
(Neufeld et al. 2006b), we estimate the atomic sulfur abundance as ∼ 10−6 relative to H2
(with a large uncertainty), a value corresponding to ∼ 4% of the solar abundance of sulfur.
Given the possible presence of S+ in the [SI] emitting-region, this observed atomic sulfur
abundance might be entirely consistent with a negligible depletion of sulfur onto grains and
a negligible fraction of the sulfur nuclei being bound within gas-phase molecules.
5) The H2 S(0) line is typically more extended than any of the other spectral lines we
have observed. In W28, it shows clear evidence for both a shock-excited component and a
low-density diffuse emission component.
6) The fine structure emissions from singly- and doubly-charged ions originate primarily in
dissociative shock fronts with a range of shock velocities. Those ions with appearance poten-
tials greater than 13.6 eV originate in faster shocks than – and show a different distribution
from – ions with appearance potentials less than 13.6 eV. In W44 the S++ emissions show
a very different distribution from those of other ions of similar appearance potential (e.g.
Ne+), a behavior which may reflect the lower critical density for the observed S++ emission
lines.
This work, which was supported in part by RSA agreement 1263841, is based on ob-
servations made with the Spitzer Space Telescope, which is operated by the Jet Propulsion
Laboratory, California Institute of Technology, under a NASA contract. D.A.N. and P.S.
gratefully acknowledge the additional support of grant NAG5-13114 from NASA's Long Term
Space Astrophysics (LTSA) Research Program. We thank the anonymous referee for a very
careful reading of the manuscript and a number of valuable suggestions.
REFERENCES
Alexander, T., Sturm, E., Lutz, D., Sternberg, A., Netzer, H., & Genzel, R. 1999, ApJ, 512,
204
Asplund, M., Grevesse, N., & Sauval, A. J. 2005, ASP Conf. Ser. 336: Cosmic Abundances
as Records of Stellar Evolution and Nucleosynthesis, 336, 25
– 16 –
Braun, R., & Strom, R. G. 1986, A&AS, 63, 345
Burton, M. G., Geballe, T. R., Brand, P. W. J. L.,& Webster, A. S. 1988, MNRAS, 231, 617
Burton, M. G., Hollenbach, D. J., Haas, M. R., & Erickson, E. F. 1990, ApJ, 355, 197
Cesarsky, D., Cox, P., Pineau des Forets, G., van Dishoeck, E. F., Boulanger, F., & Wright,
C. M. 1999, A&A, 348, 945
Caswell, J.L., Dulk, G.A., Goss, W.M., Radhakrishnan, V. & Green, A.J. 1971, A&A, 12,
271
Chevalier, R. A. 1999, ApJ, 511, 798
Claussen, M. J., Frail, D. A., Goss, W. M., & Gaume, R. A. 1997, ApJ, 489, 143
Cox, D. P., & Smith, B. W. 1974, ApJ, 189, L105
Cox, D. P., Shelton, R. L., Maciejewski, W., Smith, R. K., Plewa, T., Pawl, A., & R´ozyczka,
M. 1999, ApJ, 524, 179
DeNoyer, L. K. 1978, MNRAS, 183, 187
DeNoyer, L. K. 1979a, ApJ, 228, L41
DeNoyer, L. K. 1979b, ApJ, 232, L165
DeNoyer, L. K. & Frerking, M. A. 1981, ApJ, 246, L37
Dickman, R. L., Snell, R. L., Ziurys, L. M., & Huang, Y.-L. 1992, ApJ, 400, 203
Draine, B. T. 1980, ApJ, 241, 1021
Dubner, G. M., Vel´azquez, P. F., Goss, W. M., & Holdaway, M. A. 2000, AJ, 120, 1933
Draine, B. T., & McKee, C. F. 1993, ARA&A, 31, 373
Falgarone, E., Verstraete, L., Pineau Des Forets, G., & Hily-Blant, P. 2005, A&A, 433, 997
Fesen, R. A. & Kirshner, R. P. 1980, ApJ, 242, 1023
Fesen, R. A. 1984, ApJ, 281, 658
Frail, D. A., Kulkarni, S. R., & Vasisht, G. 1993, Nature, 365, 136
– 17 –
Frail, D.A., Goss, W.M., Reynoso, E.M., Giacani, E.B., Green, A.J., and Otrupcek, R. 1996,
AJ, 111, 1651
Georgelin, Y. M. 1975, Ph.D. dissertation, Universit´e de Provence, Marseille, France
Giacani, E. B., Dubner, G. M., Kassim, N. E., Frail, D. A., Goss, W. M., Winkler, P. F., &
Williams, B. F. 1997, AJ, 113, 1379
Giveon, U., Sternberg, A., Lutz, D., Feuchtgruber, H., & Pauldrach, A. W. A. 2002, ApJ,
566, 880
Goudis, C. 1976, Ap&SS, 40, 91
Haas, M. R., Hollenbach, D., & Erickson, E. F. 1991, ApJ, 374, 555
Heyer, M. H., & Schloerb, F. P. 1997, ApJ, 475, 173
Higdon, S. J. U., et al. 2004, PASP, 116, 975
Hollenbach, D., & McKee, C. F. 1989, ApJ, 342, 306 (HM89)
Jones, L. R., Smith, A., & Angellini, L. 1993, MNRAS, 265, 631
Le Bourlot, J., Pineau des Forets, G., & Flower, D. R. 1999, MNRAS, 305, 802
Long, K. S., Blair, W. P., Matsui, Y., & White, R. L. 1991, ApJ, 373, 567
Leahy, D. A. 2004, AJ, 127, 2277
McKee, C. F., & Ostriker, J. P. 1977, ApJ, 218, 148
Neufeld, D. A., et al. 2006a, ApJ, 649, 816
Neufeld, D. A., et al. 2006b, ApJ, 647, L33
Oliva, E., Moorwood, A. F. M., Drapatz, S., Lutz, D., & Sturm, E. 1999, A&A, 343, 943
Radhakrishnan, V., Goss, W.M., Murray, J.D. & Brooks, J.W. 1972, ApJS, 24, 49
Raymond, J. C., Blair, W. P., Long, K. S., Vancura, O., Edgar, R. J., Morse, J., Hartigan,
P., & Sanders, W. T. 1997, ApJ, 482, 881
Reach, W.T. & Rho, J.-H. 1996, A&A, 315, L277
Reach, W.T. & Rho, J.-H. 1998, ApJ, 507, L93
– 18 –
Reach, W.T. & Rho, J.-H. 1999, ApJ, 511, 836
Reach, W. T., & Rho, J. 2000, ApJ, 544, 843
Reach, W.T., Rho, J.-H., Jarrett, T.H., & Lagage, P.-O. 2002, ApJ, 564, 302
Reach, W. T., J. Rho, and T. H. Jarrett 2005, ApJ, 618, 297 (RRJ05)
Reynolds, S.P. & Moffett, D.A. 1993, AJ, 105, 2226
Rho, J., Petre, R., Schlegel, E. M., & Hester, J. J. 1994, ApJ, 430, 757
Rho, J.-H. & Petre, R. 1996, ApJ, 467, 698
Rho, J., & Petre, R. 1998, ApJ, 503, L167
Rho, J., & Borkowski, K. J. 2002, ApJ, 575, 201
Richter, M. J., Graham, J. R., Wright, G. S., Kelly, D. M. & Lacy, J. H. 1995a ApJ, 449,
L83
Richter, M. J., Graham, J. R. & Wright, G. S. 1995b, ApJ, 454, 277
Seta, M., et al. 1998, ApJ, 505, 286
Snell, R. L., Hollenbach, D., Howe, J. E., Neufeld, D. A., Kaufman, M. J., Melnick, G. J.,
Bergin, E. A. & Wang, Z. 2005, ApJ, 758, 773
Ungerechts, H., Bergin, E. A., Goldsmith, P. F., Irvine, W. M., Schloerb, F. P., & Snell,
R. L. 1997, ApJ, 482, 245
van den Bergh, S. 1973, A&A, 28, 469
van Dishoeck, E. F., Jansen, D. J., & Phillips, T. G. 1993, A&A, 279, 541
Vel´azquez, P. F., Dubner, G. M., Goss, W. M., & Green, A. J. 2002, AJ, 124, 2145
Weiler, K. W., & Sramek, R. A. 1988, ARA&A, 26, 295
Weingartner, J. C., & Draine, B. T. 2001, ApJ, 548, 296
Wilner, D.J., Reynolds, S.P., & Moffett, D.A. 1998, AJ, 115, 247
Wolszczan, A., Cordes, J. M., & Dewey, R. J. 1991, ApJ, 372, L99
Wootten, A. 1981, ApJ, 245, 105
– 19 –
Yusef-Zadeh, F., Wardle, M., & Roberts, D. A. 2003, ApJ, 583, 267
Ziurys, L. M., Snell, R. L., & Dickman, R. L. 1989, ApJ,341, 857
This preprint was prepared with the AAS LATEX macros v5.2.
– 20 –
Table 1. Details of the observations
Source
Date
Module Observing Map center position in equatorial (J2000)
timea (s)
and Galactic (decimal degrees) coordinates
2005 Apr 24
W44
2005 Apr 24
W44
2005 Oct 11
W44
2005 Apr 19
W28
2005 Apr 19
W28
2005 Apr 19
W28
2005 Apr 17
3C391
2005 Apr 17
3C391
3C391
2005 Apr 18
IC443C 2005 Mar 14
IC443C 2005 Mar 14
IC443C 2005 Mar 14
SL
SH
LH
SL
SH
LH
SL
SH
LH
SL
SH
LH
5552
6625
2650
5552
6625
2650
5552
6625
2650
5552
6625
2650
(R.A., Dec.) = (18h56m28.4s, +01◦29′59.0′′)
(l, b) = (34.8407, −0.4361)
(R.A., Dec.) = (18h01m52.3s, −23◦19′26.0′′)
(l, b) = (6.6856, −0.2956)
(R.A., Dec.) = (18h49m21.9s, −00◦57′22.0′′)
(l, b) = (31.8447, 0.0252)
(R.A., Dec.) = (06h17m42.8s, 22◦21′29.0′′)
(l, b) = (189.2953, 3.0250)
Map size
(arcsec)
58 × 57
57 × 58
59 × 56
58 × 57
57 × 58
59 × 56
58 × 57
57 × 58
59 × 56
58 × 57
57 × 58
59 × 56
aOne cycle was used in each case, with a 60 sec ramp time for LH and SL, and a 30 sec ramp time for SH.
– 21 –
Table 2. Beam-averaged line intensities, in a 25′′ (HPBW) Gaussian beam
Transition
Wavelength
(µm)
Intensity (10−6 erg cm−2 s−1 sr−1)
W44
W28
3C391
IC443C
3/2
−2 P 0
H2 S(0) J = 2 − 0
H2 S(1) J = 3 − 1
H2 S(2) J = 4 − 2
H2 S(3) J = 5 − 3
H2 S(4) J = 6 − 4
H2 S(5) J = 7 − 5
H2 S(6) J = 8 − 6
H2 S(7) J = 9 − 7
HD R(3) J = 4 − 3
HD R(4) J = 5 − 4
S 3P1 −3 P2
Si+ 2P 0
P+ 3P2 −3 P1
Cl+ 3P1 −3 P2
Fe+ 4F9/2 −6 D9/2
Fe+ 4F7/2 −4 F9/2
Fe+ 4F5/2 −4 F7/2
Fe+ 6D7/2 −6 D9/2
Fe+ 6D5/2 −6 D7/2
Ne+ 2P 0
−2 P 0
Ne++ 3P1 −3 P2
Ne++ 3P0 −3 P1
S++ 3P2 −3 P1
S++ 3P1 −3 P0
Fe++ 5D3 −5 D4
1/2
1/2
3/2
7.6
113
135
324
247
808
...a
436
28.2188
17.0348
12.2786
9.6649
8.0251
6.9095
6.1086
5.5112
28.502
23.034
25.2490
34.8141
32.8709
14.3678
5.3403
17.9363
24.5186
25.9882
35.3491
12.8149
15.5545
36.0135 < 2.6 (3σ)
22.4
5.3
156
17.1
1.7
149
1.7
0.7 ± 0.12
1.6 ± 0.5
107
4.7
18.713
33.480
22.925
34.4
43.7
1.2 ± 0.2
36.2
275
392
622
233
589
...a
231
16.2
415
4.4
3.3
151
8.7
2.5
44.7
11.6
247
38.4
3.4 ± 1.1
40.4
74.2
3.5
12.7
60.3
81.7
151
143
661
...a
272
26.8
835
11.9
8.6
388
42.0
12.5
160.4
47.2
538
202
18.0
76.9
140
7.1
12.9
359
499
1930
896
2331
491
1087
5.2
2.1
30.3
97.3
2.8 ± 0.4b
< 1.2 (3σ)
31.6 ± 5.0
2.5
< 1.2 (3σ)
11.3
10.3
15.5
4.2
< 0.9 (3σ)
< 1.4 (3σ)
< 4 (3σ)
< 0.5 (3σ)
aThe H2 S(6) line intensity cannot be determined reliably, because the line is blended
with a strong 6.2 µm PAH feature.
bError bars, where given, are 1σ statistical errors. When error bars are not given,
the uncertainty is dominated by systematic errors, which we estimate to be ≤ 25% (see
Neufeld et al. 2006a).
– 22 –
Table 3. H2 parameters
W44 W28
3C391
IC443C
Rotational state
J=2
J=3
J=4
J=5
J=6
J=7
J=8
J=9
Two-parameter fits
Tw (K)
log10 (Nw/cm−2)
OPRw
Th (K)
log10 (Nh/cm−2)
OPRh
log10 [(Nw + Nh)/cm−2]
log10 (Column density in cm−2)
19.98
19.73
19.52
20.06
19.32
18.71
19.52
18.63
18.40
17.79
18.32
17.84
...a
17.33
17.39
16.96
19.97
19.28
18.54
18.41
17.60
17.78
...a
16.79
20.36
19.83
19.09
18.70
17.69
17.66
...a
16.63
347
20.39
1.58
1134
19.42
...a
20.44
322
20.94
0.93
1039
19.51
...a
20.95
266
20.56
0.65
1099
19.31
...a
20.58
617
20.67
2.42
1225
19.60
3.85
20.70
aThe H2 S(6) line intensity cannot be determined reliably, be-
cause the line is blended with a strong 6.2 µm PAH feature. In
these cases, the OPR cannot be determined for the hot compo-
nent.
– 23 –
W44 OH region E
1 31 00
)
0
0
0
2
J
(
n
o
i
t
a
n
i
l
c
e
D
30 30
30 00
29 30
29 00
28 30
18 56 32 30 28 26 24
Right ascension (J2000)
)
0
0
0
2
J
(
n
o
i
t
a
n
i
l
c
e
D
1 35
30
25
20
15
10
05
Spitzer/IRS
field
18 57 00
18 56 00
18 55 00
Right ascension (J2000)
Fig. 1 – Finder chart for W44. The region mapped by the SH module (black square in upper
panel: white square with arrow in lower panel) is shown relative to the OH maser emission
and radio continuum emission in W44. Upper left panel: the region mapped by the SH
module is marked on Figure 5e of Claussen et al. (1997), which shows the location of the OH
maser spots within W44 OH Region E as well as the 1720 MHz radio continuum intensity
(contours, labeled in mJy/beam). Lower right panel (from Claussen et al. 1997; their Figure
2): the location of the OH maser regions is shown on the 1442 MHz radio continuum image
of Giacani et al. (1997). For an assumed distance to the source of 2.5 kpc, the map in the
lower right panel has a linear size ∼ 25 × 25 pc, and the region mapped by Spitzer has a
linear size ∼ 0.7 × 0.7 pc.
– 24 –
W28 OH region F
-23 18 30
)
0
0
0
2
J
(
n
o
i
t
a
n
i
l
c
e
D
45
19 00
15
30
45
20 00
18 01 56 55 54 53 52 51 50
Right ascension (J2000)
-23 10
)
0
0
0
2
J
(
n
o
i
t
a
n
i
l
20
30
40
c
e
50D
Spitzer/IRS
field
18 03 00 18 02 00 18 01 00 18 00 00
17 59 00
Right ascension (J2000)
Fig. 2 – Finder chart for W28. The region mapped by the SH module (black square in upper
panel: white square with arrow in lower panel) is shown relative to the OH maser emission
and radio continuum emission in W28. Upper left panel: the region mapped by the SH
module is marked on Figure 4f of Claussen et al. (1997), which shows the location of the OH
maser spots within W28 OH Region F as well as the 1720 MHz radio continuum intensity
(contours, labeled in mJy/beam). Lower right panel (from Claussen et al. 1997; their Figure
1): the location of the OH maser regions is shown on the 327 MHz radio continuum image
of Frail et al. (1993). For an assumed distance to the source of 1.9 kpc, the map in the lower
right panel has a linear size ∼ 40 × 30 pc, and the region mapped by Spitzer has a linear
size ∼ 0.55 × 0.55 pc.
– 25 –
3C 391
Spitzer / IRS
)
0
0
0
2
(
n
o
i
t
a
n
i
l
c
e
D
18h 49m 40s
35s
30s
25s
20s
15s
10s
Right Ascension (2000)
Fig. 3 – Finder chart for 3C391. Radio continuum (1446 MHz) image of 3C391 obtained with
a spatial resolution of 6′′ (from Reynolds & Moffett 1993). The Galactic plane is indicated by
the dashed diagonal line. The box centered on the "broad molecular line region" at α = 18h
49m 21.s9, δ = −0o 57′ 22′′ (J2000) delineates the area mapped with the SH module. For an
assumed distance to the source of 9 kpc, the entire map has a linear size ∼ 20 × 20 pc, and
the region mapped by Spitzer has a linear size ∼ 2.5 × 2.5 pc.
– 26 –
Fig. 4 – Finder chart for IC443. An optical image of IC443 with contours of the 1420 MHz
continuum emission overlaid (Leahy 2004). The insert shows an expanded region of the
remnant with the distribution of H2 v = 1 − 0 emission obtained by Burton et al. (1988).
The region mapped by the SH module (black square) is shown in the insert overlaid on the
contours of the H2 emission. For an assumed distance to the source of 1.5 kpc, the insert has
a linear size ∼ 9 ×3.5 pc, and the region mapped by Spitzer has a linear size ∼ 0.45 ×0.45 pc.
– 27 –
Fig. 5 – Average spectra observed for a 25′′ (HPBW) diameter circular aperture centered
at α = 18h 56m 28.s4, δ = +0o 29′ 59′′ (J2000) in W44 (corresponding to offset (0′′, 0′′) in
Figure 10 below). The spectra in the upper two panels were obtained with the SL module,
those in the middle two panels were obtained with the SH module (λ ≤ 19.5 µm) and those
in the lower two panels with the LH module (λ ≥ 19.5 µm). (For each of the SH and LH
spectra, two panels are shown with different vertical scales.) Vertical dotted lines demark
the wavelengths of spectral lines listed in Table 2 (regardless of whether the line is actually
detected in this particular source.) The spectra shown here are not background-subtracted,
since no off-source measurements were made, and thus the continuum flux level must be
regarded as somewhat uncertain.
– 28 –
Fig. 6 – same as for Fig. 5, except for a 25′′ (HPBW) diameter circular aperture centered
at α = 18h 01m 52.s3, δ = −23o 19′ 25′′ (J2000) in W28 (corresponding to offset (0′′, 0′′) in
Figure 11 below).
– 29 –
Fig. 7 – same as for Fig. 5, except for a 25′′ (HPBW) diameter circular aperture centered
at α = 18h 49m 21.s9, δ = −0o 57′ 22′′ (J2000) in 3C391 (corresponding to offset (0′′, 0′′) in
Figure 12 below).
– 30 –
Fig. 8 – same as for Fig. 5, except for a 25′′ (HPBW) diameter circular aperture centered at
α = 06h 17m 42.s5, δ = +22o 21′ 29′′ (J2000) in IC443C (corresponding to offset (−24′′, −20′′)
in Figure 13 below).
– 31 –
Fig. 9 – Excitation conditions for the observed transitions. The horizontal axis shows the
appearance potential – equal to zero for neutral species and equal to the ionization potential
of X(n−1)+ for the case of ion Xn+ – and the vertical axis shows the critical density, ncr, at
which the rate of collisional de-excitation equals the spontaneous radiative rate. For neutral
species, the critical densities apply to de-excitation by H2 molecules at a temperature of
500 K; for ionized species, the critical densities apply to de-excitation by electrons at a
temperature of 104 K. The values adopted for ncr are from the papers of Giveon et al.
(2002), Le Bourlot, Pineau des Forets & Flower (1999) and HM89.
– 32 –
Fig. 10 – H2 S(0), H2 S(1), [SI], [SiII], [NeII], and [SIII] emission line intensities observed
toward W44. The inset boxes demark the regions within which each transition was mapped.
The horizontal and vertical axes show the R.A. (∆α cos δ) and declination (∆δ) offsets in
arcsec relative to α = 18h 56m 28.s4, δ = +0o 29′ 59′′ (J2000).
– 33 –
Fig. 11 – H2 S(0), H2 S(1), [SI], [SiII], [NeII], and [SIII] emission line intensities observed
toward W28. The inset boxes demark the regions within which each transition was mapped.
The horizontal and vertical axes show the R.A. (∆α cos δ) and declination (∆δ) offsets in
arcsec relative to α = 18h 01m 52.s3, δ = −23o 19′ 25′′ (J2000).
– 34 –
Fig. 12 – H2 S(0), H2 S(1), [SI], [SiII], [NeII], and [SIII] emission line intensities observed
toward 3C391. The inset boxes demark the regions within which each transition was mapped.
The horizontal and vertical axes show the R.A. (∆α cos δ) and declination (∆δ) offsets in
arcsec relative to α = 18h 49m 21.s9, δ = −0o 57′ 22′′ (J2000).
– 35 –
Fig. 13 – H2 S(0), H2 S(1), [SI], [SiII], and [NeII] emission line intensities observed toward
IC443C. The inset boxes demark the regions within which each transition was mapped. The
horizontal and vertical axes show the R.A. (∆α cos δ) and declination (∆δ) offsets in arcsec
relative to α = 06h 17m 44.s2, δ = +22o 21′ 49′′ (J2000).
– 36 –
Fig. 14 – Results of PCA analysis for W44. Upper panel: fraction of the variance accounted
for by each component (asterisks), along with the cumulative fraction for the first n
components as a function of n (open squares). Left lower panel: coefficients for the first
and second principal components needed to approximate the maps of each transition. Right
lower panel: coefficients for the second and third principal components. Where multiple
transitions of a single ion have been observed, the labels distinguish between the different
transitions by giving the wavelength rounded to the nearest micron.
– 37 –
Fig. 15 – same as for Fig. 14, except for W28.
– 38 –
Fig. 16 – same as for Fig. 14, except for 3C391.
– 39 –
Fig. 17 – same as for Fig. 14, except for IC443C.
– 40 –
Fig. 18 – H2 rotational diagrams, computed for 25′′ (HPBW) diameter circular apertures
centered at the positions given in the captions to Figures 5 – 8. As usual, the quantity
plotted on the vertical axis is the logarithm of the column density per magnetic substate,
log10(N/gJgN ), where gJ = (2J + 1), J is the rotational quantum number, gN = (2I + 1) is
the nuclear spin degeneracy, and I is the nuclear spin (equal to 3 for states of odd J and 1
for states of even J). Square symbols indicate the measured values, while the lines show the
best two-component fit to the observations (see Table 3). The J = 8 column density cannot
be measured reliably in W44, W28, or 3C391, the H2 S(6) line being blended with a strong
6.2 µm PAH emission feature.
– 41 –
Fig. 19 – Scatter diagrams showing correlations between [SI] and H2 S(0). Each point shows
the mean intensities within a 5′′ × 5′′ square region.
– 42 –
Fig. 20 – Scatter diagrams showing correlations between the [FeII] 26 µm and [SiII] fine
structure lines. Each point shows the mean intensities within a 5′′ × 5′′ square region.
– 43 –
Fig. 21 – Scatter diagrams showing correlations between the [PII] 26 µm and [SiII] fine
structure lines. Each point shows the mean intensities within a 5′′ × 5′′ square region.
|
astro-ph/0105499 | 1 | 0105 | 2001-05-29T16:03:38 | Spectroscopy and orbital periods of four cataclysmic variable stars | [
"astro-ph"
] | We present spectroscopy and orbital periods Porb of four relatively little-studied cataclysmic variable stars. The stars and their periods are: AF Cam, Porb = 0.324(1) d (the daily cycle count is slightly ambiguous); V2069 Cyg (= RX J2123.7+4217), 0.311683(2) d; PG 0935+075, 0.1868(3) d; and KUV 03580+0614, 0.1495(6) d. V2069 Cyg and KUV 03580+0614 both show HeII lambda 4686 emission comparable in strength to H beta. V2069 Cyg appears to be a luminous novalike variable, and the strong HeII suggests it may be an intermediate polar. The period of KUV 03580+0614 is similar to members of the SW Sex-type novalike variables, and it shows the phase-dependent absorption in the Balmer and He I lines typical of this subclass. AF Cam shows absorption features from a K-type secondary, as expected given its rather long orbital period. The secondary spectrum and the outburst magnitude both suggest that AF Cam is about 1 kpc distant. The spectrum of PG 0935+075 resembles that of a dwarf nova at minimum light, with a noticeable contribution from an M-dwarf secondary star. The secondary spectrum and a tentative outburst magnitude both suggest a distance near 500 pc. | astro-ph | astro-ph |
Mon. Not. R. Astron. Soc. 000, 000 -- 000 (1994)
Printed 27 October 2018
(MN plain TEX macros v1.6)
Spectroscopy and orbital periods of four cataclysmic
variable stars
John R. Thorstensen and Cynthia J. Taylor
Department of Physics and Astronomy, Dartmouth College, Hanover, New Hampshire, 03755 USA
ABSTRACT
We present spectroscopy and orbital periods Porb of four relatively little-studied
cataclysmic variable stars. The stars and their periods are: AF Cam, Porb = 0.324(1)
d (the daily cycle count is slightly ambiguous); V2069 Cyg (= RX J2123.7+4217),
0.311683(2) d; PG 0935+075, 0.1868(3) d; and KUV 03580+0614, 0.1495(6) d. V2069
Cyg and KUV 03580+0614 both show HeII λ4686 emission comparable in strength
to Hβ. V2069 Cyg appears to be a luminous novalike variable, and the strong HeII
suggests it may be an intermediate polar. The period of KUV 03580+0614 is similar
to members of the SW Sex-type novalike variables, and it shows the phase-dependent
absorption in the Balmer and He I lines typical of this subclass. AF Cam shows
absorption features from a K-type secondary, as expected given its rather long orbital
period. The secondary spectrum and the outburst magnitude both suggest that AF
Cam is about 1 kpc distant. The spectrum of PG 0935+075 resembles that of a dwarf
nova at minimum light, with a noticeable contribution from an M-dwarf secondary
star. The secondary spectrum and a tentative outburst magnitude both suggest a
distance near 500 pc.
Key words: stars: novae -- stars: cataclysmic variables -- binaries: close -- stars: indi-
vidual -- stars: variables -- stars: fundamental parameters
1
INTRODUCTION
Cataclysmic variable stars (CVs) are close binary systems in
which a red dwarf transfers matter onto a white dwarf, gen-
erally by overflowing its Roche critical lobe. Warner (1995)
gives an excellent comprehensive review of CVs.
The orbital period of a cataclysmic is a fundamental
observable which correlates with evolutionary state and out-
burst type. In addition, the Ritter & Kolb (1998) catalog,
which is heavily used by the CV community, includes only
those systems with known or suspected binary periods, with
the result that a system remains largely 'beneath the radar'
until a period is measured. Photometry yields incontrovert-
ible orbital periods when eclipses are present, but other mod-
ulations can masquerade as orbital periods. Radial-velocity
spectroscopy is therefore the most reliable technique for de-
termining orbital periods of non-eclipsing CVs. The long cu-
mulative exposures needed to find reliable periods are useful
for characterizing the stars in other ways as well.
We present here spectra and radial velocity periods for
four CVs. Section 2 details the techniques common to the
studies, Section 3 gives background information and results
for the individual stars, and Section 4 contains a brief dis-
cussion.
2 TECHNIQUES
All the observations (summarized in Table 1) are from the
MDM Observatory on Kitt Peak, Arizona. The 1998 Jan-
uary observations of KUV 03580+0614 are from the 1.3
c(cid:13) 1994 RAS
m McGraw-Hill telescope and Mark III spectrograph, and
all others are from the 2.4 m Hiltner telescope and modu-
lar spectrograph. We observed comparison lamps frequently,
and checks of the λ5577 night-sky line show that our wave-
length scale is typically stable to < 10 km s−1. When the
weather appeared photometric, we observed flux standards
in twilight. Even so, our absolute fluxes are not expected
to be accurate to much better than 30 per cent, because
of occasional clouds and variable losses at the spectrograph
slits (1 arcsec at the 2.4 m., 2.2 arcsec at the 1.3 m). Fur-
thermore, for unknown reasons the modular spectrograph
produces wavelike distortions in the continua; these appear
to average out in sums of many exposures. Table 2 contains
measurements of the average fluxed spectra.
The data were reduced to counts vs. wavelength at the
observatory, using standard procedures within IRAF. Ra-
dial velocities of emission lines were measured using convo-
lution algorithms described by Schneider & Young (1980)
and Shafter (1983). For absorption velocities, we used the
cross-correlation algorithms of Tonry & Davis (1979), as im-
plemented in the rvsao package (Kurtz & Mink 1998). To
search for periods we used the 'residual-gram' method de-
scribed by Thorstensen et al. (1996), and to test alias choices
in doubtful cases we used the Monte Carlo method explained
by Thorstensen & Freed (1985). With the rough period es-
tablished, we fit the velocities with sinusoids, the parameters
of which are in Table 3.
We note one simple innovation in these otherwise stan-
dard procedures. As part of the spectral reduction process,
2
J. R. Thorstensen & C. J. Taylor
Table 1. Observing Log
Run
Nights N exposure
V2069 Cyg
1997 Sep
1997 Dec
1998 Sep
1999 Jun
1999 Oct
AF Cam
2000 Jan
3
3
1
3
4
5
PG 0935+075
3
1996 Apr
2000 Jan
4
KUV 03580+0614
1997 Dec
1998 Jan
1999 Oct
3
4
4
(s)
360
480
480
300
240
23
12
3
9
7
Range
(A)
4000 -- 7500
4000 -- 7500
4040 -- 7500
4230 -- 7580
4230 -- 7550
∆λ
(A)
3.7
3.7
3.7
3.6
3.9
32
360
4210 -- 7560
3.6
27
21
41
7
57
480
480
480
900
480
4208 -- 6780
4210 -- 7560
4000 -- 7500
4845 -- 6865
4230 -- 7560
2.7
3.6
3.7
4.5
3.6
Exposure times listed are typical values for each run.
∆λ is the FWHM resolution determined from fits to
night-sky features.
IRAF can estimate the noise in each spectral bin from
the readout noise, detector gain, background, and so on.
We have modified the convolution algorithm to compute
a counting-statistics uncertainty in the radial velocities by
propagating these noise estimates through the calculation.
These a priori estimates allow the data to be more optimally
weighted in period finding and curve fitting.
3 THE INDIVIDUAL STARS
3.1 AF Camelopardalis
This object is listed in the General Catalog of Variable stars
(Kholopov et al. 1988) as a U Gem-type dwarf nova with
13.4 ≤ mpg ≤ 17.6. It is the only star here with a useful
period estimate in the literature. Early indications of a very
short period (Szkody & Mateo 1986, Howell and Szkody
1988) prompted an emission-line radial velocity study by
Szkody and Howell (1989). They obtained very limited data
which indicated a 5 -- 6 h period, leading Ritter & Kolb (1998)
to tabulate Porb = 0.23 : in their catalogue.
We observed this object in 2000 January, in quiescence.
The mean spectrum (Fig. 1) shows the absorption features
of a late-type secondary star, as would be expected given
the relatively long Porb. The emission lines are conspicuous
and fairly narrow. We quantified the secondary star's con-
tribution by taking K-type main-sequence spectra from the
library of Pickles (1998), scaling them by logarithmically-
spaced factors, and subtracting them from the mean fluxed
AF Cam spectrum, which had been smoothed to match the
lower resolution of Pickles' spectra. The resulting spectra
were plotted and examined by eye to find the best cancella-
tion of the secondary star features. We found acceptable sub-
tractions for stars of type K4, K5, and K7, and a marginal
match was possible for M0. The secondary contribution was
30 -- 40 per cent of the light near λ6500. For the secondary
alone we find V = 18.4, with an uncertainty estimated from
the decomposition and the flux calibration of perhaps ±0.5
mag.
Table 2. Spectral Features.
Wavelength Identification E.W. Flux FWHM
(A)
(A)
(A)
AF Cam (2000 January)
4339
4473
4860
5876
5893
6564
6680
7066
Hγ
HeI λ4471
Hβ
HeI λ5876
NaD
Hα
HeI λ6678
HeI λ7067
33 107:
9:
28:
125
33
37:
7:
−1.8 :
. . .
220
37
4
23
20
3:
V2069 Cyg (1997 December)
4340
4641
4684
4922
5014
5411
5780
5874
5889
5895
6562
6678
7064
Hγ
CIII/NIII
HeII λ4686
HeI λ4921
HeI λ5015
HeII λ5411
DIB
HeI λ5876
Na DI
Na DII
Hα
HeI λ6678
HeI λ7067
10
3:
15
2:
2:
2.5:
-0.4
49
-0.5
-0.9
32
3.8
2.7
PG 0935+075 (2000 January)
4340
4861
4921
5018
5168
5875
6562
6676
Hγ
Hβ
HeI λ4921
HeI λ5015
FeI λ5169
HeI λ5876
Hα
HeI λ6678
93
80
14
16:
9:
23
101
10
KUV 03580+0614 (1999 October)
6
3
10
8.6
0.6:
0.8
-0.4
1.3:
0.4:
0.4
23
2
1.3:
Hγ
CIII/NIII
HeII λ4686
Hβ
HeI λ5015
HeII λ5411
DIB
HeI λ5876
Na D1
Na D2
Hα
HeI λ6678
HeI λ7067
4338
4647
4684
4859
5011
5407
5780
5872
5890
5896
6562
6678
7063
120
35:
180
26
25
30
. . .
13
. . .
. . .
365
42
31
129
138
20:
25
16:
40:
199
18
158
67
220
185
13
14
. . .
22
. . .
. . .
303
27
15
18:
16:
15
15:
10:
16
19
24:
13
26
13
16:
12:
20:
4
3.2
6.3
15
16
15
18:
18
28:
18:
23
20:
25
21
29
20
19
. . .
20
5.8
22:
3.4
4.1
25
25
24:
Wavelengths are as observed. Positive equivalent
widths denote emission. Line fluxes are given in
units of 10−16 erg cm−2 s−1, and their esti-
mated accuracy is ±30 per cent. The secondary
absorption features in AF Cam are not tabulated,
with the exception of Na D. The HeI lines in
PG0935+075 were strongly affected by their cen-
tral absorption, and the Na D1 and HeI λ5876 lines
generally overlapped.
We cross-correlated the λλ5020 − 5860 and λλ5900 −
6500 regions against a composite of several G- and K-type
IAU radial velocity standards, and obtained velocities for
30 of our 32 target spectra, with typical estimated errors
∼ 20 km s−1. We measured the Hα emission line by con-
volving with the derivative of Gaussian, optimized for a 16 A
c(cid:13) 1994 RAS, MNRAS 000, 000 -- 000
Four Cataclysmics
3
Table 3. Fits to the Radial Velocities
Data
T0
(HJD)
P
(d)
K
γ
σ
(km s−1) (km s−1) (km s−1)
AF Cam (absn)
AF Cam (emn)
AF Cam (avg)
2451552.671(4) 0.3242(12)
2451552.497(4) 0.3230(12)
0.3236(9)
. . .
104(5)
58(3)
. . .
V2069 Cyg
2451066.783(2) 0.311683(2)
125(5)
20(4)
15(3)
. . .
12(3)
PG 0935+075
2451552.774(2)
0.1868(3)
KUV 03580+0614 2451470.668(3)
0.1495(6)
86(7)
67(8)
−20(5)
−53(6)
21
12
. . .
20
22
42
N
30
32
53
42
105
Fits are of the form v(t) = γ + K sin[2π(t − T0)/P ], and σ is the uncertainty of a typical
measurement judged from the scatter around the best fit. N is the number of velocities
used.
Fig. 1. Mean flux-calibrated spectrum of AF Cam, from 2000
January. The vertical scale is uncertain by perhaps 30 per cent.
FWHM. The counting-statistics velocity uncertainties here
were typically 10 km s−1. Fig. 2 shows periodograms for the
emission and absorption velocities, both of which favor a pe-
riod near 0.32 d, but with the inevitable daily cycle-count
aliases. The Monte Carlo tests give discriminatory powers
(defined in Thorstensen & Freed 1985) of 0.96 and 0.83 for
the emission and absorption time series respectively, with
correctness likelihoods somewhat higher than this, so the
cycle count is reasonably secure, but frequencies differing by
1 cycle d−1 cannot be excluded absolutely. The data span
7.8 h of hour angle, ordinarily enough to decide daily cycle
counts unambiguously, but the sampling is somewhat sparse
and the period is awkwardly close to an integer submultiple
of one day. On the face of it, Szkody & Howell's (1989) 5 -- 6
hr period favors the ∼ 4 cycle d−1 alias, rather than the ∼ 3
cycle d−1 we adopt; however, their data do not have enough
hour angle span to influence the choice.
Fig. 3 shows the velocities folded on the combined best
period. The sine fits (Table 3) show that the emission line
velocity modulation lags the absorption line velocities by
0.54±0.02 cycles, suggesting as usual that the emission lines
do not trace the white dwarf motion with any precision.
The secondary velocity amplitude suggests a fairly low
orbital inclination, as might be expected from the single-
peaked emission lines. As a purely illustrative example, the
observed 104 km s−1 velocity amplitude of the secondary
would be expected for a 0.6 M⊙ secondary orbiting a 0.7
M⊙ white dwarf with an inclination of 35 degrees. If the late-
type features arise preferentially on the hemisphere facing
c(cid:13) 1994 RAS, MNRAS 000, 000 -- 000
Fig. 2. Period searches of the absorption (top) and emission (bot-
tom) velocities of AF Cam. The figure of merit shown is the in-
verse of (1/N ) Pi(o − c)2/σ2
i , where o is the observed value, c
is the value computed from the best-fit sinusoid at the trial fre-
quency, and σi is the estimated uncertainty of that particular
velocity.
away from the white dwarf, the true K is even smaller, and
the inclination still lower.
The spectral type found here (K4 -- M0) is similar to that
of other relatively long-period CVs (Beuermann et al. 1998).
AF Cam appears very similar to the apparently brighter
system CH UMa. The orbital period of CH UMa is 0.343
d (Friend et al. 1990), only a few per cent longer than AF
Cam. Becker et al. (1982) estimate the spectral type of CH
UMa's secondary as K4 -- M0, identical to our estimate for
AF Cam.
4
J. R. Thorstensen & C. J. Taylor
Fig. 3. Velocities of AF Cam folded on the best period. All data
are plotted twice for continuity. The round dots show the absorp-
tion velocities, and the crosses the Hα emission velocities. The
best-fit sinusoids are overplotted.
The detection of the secondary and the orbital period
constrain the distance. Beuermann et al. (1999) tabulate ab-
solute magnitudes, colors, and estimated radii of a sample
of nearby K and M dwarfs (their Table 3). Scaling these
data, we find that hypothetical 1 R⊙ stars of type K4 and
type M0 should respectively have MV = +6 and +7.2.
The orbital period and Roche constraints (Beuermann et
al. 1998, eqn. 1) yield (R2/R⊙) = f (q)0.92(M2/M⊙)(1/3) at
AF Cam's period, where the subscript 2 refers to the sec-
ondary star, and the function f (q) is within 3 percent of
unity for q = M2/M1 ≤ 1. Because the secondary is likely
to be modified by mass transfer (see, e. g., Beuermann et
al. 1998), we do not assume a main-sequence mass-radius
relation, but rather use the range of evolutionary models cal-
culated by Baraffe & Kolb (2000) as a guide; this may not
cover all possibilities, but it is arguably better than guessing.
At Porb = 8 h their models span 0.36 ≤ M2/M⊙ ≤ 0.99, but
the more massive secondaries are calculated to have spec-
tral types distinctly earlier than observed here; more real-
istically, M2 ≤ 0.7 M⊙. Secondaries ranging from 0.36 to
0.7 M⊙ yield 0.65 < R2/R⊙ < 0.82, corresponding to stars
0.7 ± 0.3 magnitude fainter than otherwise identical stars
with R = R⊙. Combining these calculations and propagat-
ing the uncertainties in quadrature yields MV = 7.3 ± 0.7
for the secondary. Our detection of the secondary therefore
yields yields m − M = 11.1 ± 0.9.
Warner (1987, 1995) gives a relation for dwarf novae be-
tween MV at maximum light, orbital inclination, and orbital
period. The inclination is uncertain, but likely to be fairly
low as noted earlier. Taking i = 30 degrees, these relations
give MV (max) = 3.0 for AF Cam. For the apparent mag-
nitude we use the GCVS outburst magnitude mpg = 13.4,
which should be similar to V for typical outburst colors.
This then yields m − M = 10.4, in reasonable agreement
with the distance estimated from the secondary. The actual
distance is likely to be less, because of extinction corrections
at this low Galactic latitude (b = 2◦.2).
3.2 V2069 Cygni
V2069 Cyg (= RX J2123.7+4217) was discovered through
its X-ray emission by Motch et al. (1996). They presented a
spectrum and a 2.8-hour session of time-resolved CCD pho-
tometry, which showed flickering but no apparent periodic-
ity. It is listed as 'Cyg6' in Downes, Webbink, and Shara's
(1997) catalog and atlas of CVs (hereafter DWS97).
The spectrum (Fig. 4) appears similar to that in Motch
et al. (1996). The He II λ 4686 line is notably strong. The
equivalent width of NaD (which does not show orbital mo-
tion) and the continuum slope both suggest that the redden-
ing is not negligible. The Hα radial velocities were measured
using the derivative of a gaussian as the convolution func-
tion, optimized for 14 A FWHM. The period search (Fig. 5)
yielded a unique choice of cycle count for the 760-day span
of the observations. The resulting period, 0.311683(2) d, or
7.48 h, is determined with the greatest accuracy of those
reported here. The folded velocities (Fig. 6) and fits (Table
3) show why this went so well -- the velocity amplitude K
was was large, and the scatter relatively small, making the
periodic modulation very conspicuous.
V2069 Cyg shows several features in common with V405
Aur (= RX J0558+5353). They are: (1) relatively nar-
row, single-peaked lines; (2) strong HeII λ4686 emission; (2)
Porb > 4 h; (3) a quiet, large-amplitude emission-line radial
velocity curve. V405 Aur has proven to be a very interesting
DQ Her star (= intermediate polar; see Harlaftis & Horne
1999 and references therein). Although Motch et al.'s (1996)
brief light curve did not show obvious periodic pulsation, a
more sensitive search might be rewarded with success.
3.3 PG 0935+075
PG 0935+075 was discovered as an ultraviolet-excess object
in the Palomar-Green survey (Green, Schmidt, & Liebert
1986). Ringwald (1993) obtained a spectrum showing the
strong, broad Balmer and He I emission typical of a cat-
aclysmic binary. DWS97 list the object as 'Leo7'. It is
evidently a little-studied dwarf nova, since it appears at
B = 13.0 in the PG survey, but is much fainter in Ring-
wald's spectrum and in the observations described below.
We observed this object only in 1996 April and 2000
January. The 1996 April observations indicated a periodicity
near 0.19 d, but the daily cycle count was ambiguous. The
2000 January observations, though fewer in number, were ar-
ranged to span 7.35 h of hour angle and therefore decided the
daily cycle count. The periods derived from the two observ-
ing runs differ by almost twice their mutual standard devia-
tion, but are reasonably consistent, with a weighted average
of 0.1868(3) d. Because the two observing runs were so far
apart, there is no unique choice of cycle count between them,
but if the two runs are phase coherent the allowed periods
are expressed by (1371.151 ±0.003 d)/(7345 ±50), where the
denominator is an integer. Because of the mediocre agree-
ment of the individual runs' periods, the cycle-count uncer-
tainty is chosen to yield periods within ±4σ of the weighted
average. Fig. 7 summarizes the period search for the veloci-
ties, and Fig. 8 shows the folded velocities.
The mean spectrum (Fig. 9, Table 2) closely resembles
those of dwarf novae at minimum light. There is a contribu-
tion from an M-dwarf secondary star, which confirms that
the luminosity was fairly low when the data were taken.
c(cid:13) 1994 RAS, MNRAS 000, 000 -- 000
Four Cataclysmics
5
Fig. 4. Mean spectrum of V2069 Cyg. Note the relatively narrow
emission lines and the strength of HeII λ4686.
Fig. 6. Hα radial velocities of V2069 Cyg, folded on the best
period. Different symbols represent different observing runs as
follows: solid dots, 1997 Sept.; filled squares, 1997 Dec.; filled
triangles, 1998 Sept.; stars, 1999 June; and open circles, 1999
Oct.
Fig. 5. Period search of the Hα emission velocities of V2069 Cyg.
The vertical axis is the mean of the inverse square residual at each
frequency. A single frequency is strongly preferred.
It would appear from the spectrum, period, and the mini-
mal information available on variability that this object is a
UGSS.
To quantify the secondary star contribution we used the
subtraction technique described earlier, but this time with
a library of M-dwarf spectra classified by Boeshaar (1976)
and observed with the same instrumental setup. We found
the secondary contribution to be type M3 ± 1, with the
secondary contributing 45 ± 15 per cent of the continuum
at 6500 A. The secondary contribution has V = 19.7 ± 0.5.
Unfortunately, this was too faint for us to measure radial
velocities of the secondary.
Again, we can estimate a distance from the secondary
contribution. Adapting the data from Table 3 of Beuermann
et al. (1999), we find that a hypothetical 1R⊙ star of type M3
± 1 would have MV = 9.2 ± 1.0. The Roche lobe constraint
at this period yields R2/R⊙ = 0.64f (q)(M2/M⊙)(1/3). Turn-
ing again to the evolutionary models of Baraffe & Kolb
(2000), we find 0.17 ≤ M2/M⊙ ≤ 0.6 in the 4 -- 5 hour pe-
riod range, which gives 0.35 ≤ R2/R⊙ ≤ 0.54, ignoring the
slight variation in f (q). This makes the secondary 1.8 ± 0.5
mag fainter than a 1R⊙ star of the same spectral class, so
we estimate MV = 11.0 ± 1.2 for the secondary, which in
c(cid:13) 1994 RAS, MNRAS 000, 000 -- 000
Fig. 7. Period search of Hα radial velocities of PG0935+075.
The full periodogram shows ringing from the unknown number
of cycle counts between the widely-spaced observing runs, so the
curve shown here is formed by connecting the local maxima of
the periodogram with straight lines.
turn yields m − M = 8.7 ± 1.3. The uncertainty in the sec-
ondary's spectral type dominates the error budget. For com-
parison, Warner's (1987) maximum light relationship yields
MV (max) = 4.6 at this period. Taking the inclination cor-
rection as zero and V at minimum light as 13.0 (both quite
uncertain) yields m − M = 8.4. The good agreement should
probably be interpreted as supporting the dwarf nova clas-
sification of this star (that is, on the one occasion it was
seen to be bright, it was about as bright as expected for
a dwarf nova), rather than as corroborating the distance.
At b = +40.2, the extinction is unlikely to be significant.
If m − M = 8.7, PG0935+075 lies some 350 pc from the
Galactic plane.
6
J. R. Thorstensen & C. J. Taylor
Fig. 8. Hα radial velocities of PG0935+075 folded on the best
period. The period used reflects an essentially arbitrary choice
of cycle count between observing runs. Filled triangles are from
1996 April, and filled squares from 2000 January.
Fig. 10. Period search of Hα radial velocities of KUV 03580+
0614. The curve shown here is formed by connecting the local
maxima of the periodogram with straight lines.
Fig. 9. Mean spectrum of PG0935+075 from 1999 October. Note
the M dwarf features.
3.4 KUV 03580+0614
Wegner & Boley (1993) discovered that KUV 03580+0614
shows emission lines, and flagged it as a cataclysmic variable
candidate. DWS97 list it as 'Tau2'.
The 1997 December velocities indicated a period near
0.15 d, but the daily cycle count was not securely estab-
lished. The 1999 October data are slightly more exten-
sive and the velocities show less scatter; they unambigu-
ously confirm the 0.15 d period. Sinusoidal fits to velocities
from the two observing runs starting near this frequency
gave essentially the same period, the weighted average being
0.1495 ± 0.0006 d. A period search of the combined veloci-
ties yields a slightly longer best-fit period near 0.1502 d. The
cycle count between the observing runs is not determined,
but the periods within ±3σ = 0.0018 d of the best overall
period can be expressed as (671.772 ± 0.005 d)/(4472 ± 53),
the denominator being integer. Figures 10 and 11 show the
period search and the folded radial velocities.
Fig. 12 shows the mean spectrum from 1999 October;
the 1997 spectrum appeared generally similar with a slightly
Fig. 11. Hα radial velocities of KUV 03580+0614 folded on
the best period. The period used reflects an essentially arbitrary
choice of cycle count between observing runs. Filled squares are
from 1997 December, filled triangles from 1998 January, and solid
circles are from 1999 October.
lower flux level. The spectrum is that of a novalike vari-
able, with HeII λ4686 similar in strength to Hβ; the con-
tinuum is quite blue, approximately Fλ ∝ λ−1.7. Fig. 13
shows a single-trailed representation of the 1999 October
data, prepared using phase-averaging techniques (Taylor,
Thorstensen, & Patterson 1999) The He I features especially
show the distinctive phase-dependent absorption character-
istic of the SW Sex stars (Thorstensen et al. 1991, Taylor
et al. 1999, Dhillon, Marsh, & Jones 1998). In the SW Sex
stars the Balmer emission velocities typically lag the ex-
pected white-dwarf motion by ∼ 0.2 cycle. Assuming this
to be the case here, we infer that white dwarf superior con-
junction should occur near φ = 0.3 in the phase convention
we are using. In most SW Sex stars the phase of white dwarf
superior conjunction is marked by eclipses, and the absorp-
tion reaches maximum strength approximately opposite the
eclipse, at 'phase 0.5' in the eclipse ephemeris. This should
correspond to φ ∼ 0.8 in the present phase convention. The
c(cid:13) 1994 RAS, MNRAS 000, 000 -- 000
Four Cataclysmics
7
ACKNOWLEDGMENTS
We thank Joe Patterson and Bob Fried for communicat-
ing that KUV 03580+0614 does not eclipse, and Patrick
Schmeer for a discussion of the variability of PG0935 +075.
The US National Science Foundation supported this work
through grants AST 9314787 and AST 9987334. The MDM
Observatory staff provided cheerful and efficient help.
REFERENCES
Baraffe, I., Kolb, U. 2000, MNRAS, 318, 354
Becker, R. H., Wilson, A. S., Pravdo, S. H., Chanan, G. A. 1982,
MNRAS, 201, 265
Beuermann, K., Baraffe, I., Kolb, U., Weichhold, M. 1998, A&A,
339, 518
Beuermann, K., Baraffe, I., Hauschildt, P. 1999, A&A, 348, 524
Boeshaar, P. 1976, Ph. D. Thesis, Ohio State University
Dhillon, V. S., Marsh, T. R., Jones, D. H. P. 1998, MNRAS, 291,
694
Downes, R. A., Webbink, R. F., Shara, M. M. 1997, PASP, 109,
345
Friend, M. T., Martin, J. S., Connon Smith, R., Jones, D. H. P.
1990, MNRAS, 246, 654
Green, R. F., Schmidt, M., Liebert, J. 1986, ApJS, 61, 305
Harlaftis, E. T., Horne, K. 1999, MNRAS, 305, 437
Hellier, C. 1998, PASP, 110, 420
Howell, S. B., Szkody, P. 1988, PASP, 100, 224
Kholopov, P. N. et al., 1988, General Catalogue of Variable Stars,
Moscow: Nauka Publishing House.
Kurtz, M. J., Mink, D. J. 1998, PASP, 110, 934
Motch, C., Haberl, F., Guillout, P., Pakull, M., Reinsch, K.,
Krautter, J. 1996, A&A, 307, 459
Patterson, J., Skillman, D. R. 1994, PASP, 106, 1141
Pickles, A. J. 1998, PASP, 110, 863
Ringwald, F. A. 1993, PhD thesis, Dartmouth College
Ritter, H., Kolb, U. 1998, A&AS, 129, 83
Schneider, D. P., Young, P. 1980, ApJ, 238, 946
Shafter, A. W. 1983, ApJ, 267, 222
Szkody, P., Mateo, M. 1986, AJ, 92, 483
Szkody, P., Howell, S. B., 1989, AJ, 97, 1176
Taylor, C. J., Thorstensen, J. R., Patterson, J. 1999, PASP, 111,
184
Thorstensen, J. R., Freed, I. W. 1985, AJ, 90, 2082
Thorstensen, J. R., Patterson, J. Thomas, G., Shambrook, A.
1996, PASP, 108, 73
Tonry, J., Davis, M. 1979, AJ, 84, 1511
Warner, B. 1987, MNRAS, 227, 23
Warner, B. 1995, Cataclysmic Variable Stars (Cambridge: Cam-
bridge U. Press)
Wegner, G., Boley, F. I. 1993, AJ, 105, 660
This paper has been produced using the Royal Astronomical
Society/Blackwell Science TEX macros.
Fig. 12. Mean spectrum of KUV 03580+0614 from 1999 October.
Note the strong continuum and HeII λ4686 emission.
observed absorption is strongest around φ ∼ 0.73, in fair
agreement with expectation. Close examination of Fig. 13
also shows a red-to-blue drift of the absorption features as
they strengthen, a behaviour seen in other SW Sex stars.
The orbital period is also similar to other examples of the
class.
A search for eclipses by R. Fried and J. Patterson
(private communication, 2000) proved negative. Thus KUV
03580+0614 joins the ranks of stars which behave spectro-
scopically like SW Sex stars, but which do not eclipse (e.g.,
Taylor et al. 1999).
4 DISCUSSION
The spectra and orbital periods presented here show all
these objects to be fairly typical examples of their classes.
The two dwarf novae (AF Cam and PG0935+075) are of
interest largely because their secondaries are detected, which
allows us to determine distances and further characterise CV
secondaries. PG0935+075 should be monitored for further
outbursts to confirm its variability type.
The two novalikes, V2069 Cyg and KUV 03580+0614,
are potentially more interesting as individuals. V2069 Cyg
shows similarities with V405 Aur, which has proven to be a
very interesting DQ Her star. SW Sex stars frequently show
interesting 'permanent superhumps' (which might more ac-
curately be called persistent superhumps) in their light
curves (Patterson & Skillman 1994). Thus more thorough
time-series photometry of these stars may prove interesting.
KUV 03580+0614 is a good example of a star showing
the phase-dependent absorption of the SW Sex phenomenon,
but no eclipses. The growing number of non-eclipsing SW
Sex stars presents a challenge to scenarios which require the
line of sight to graze the rim of the disk (e.g., Hellier 1998).
If the light being absorbed arises near the disk center, the
absorbing material must be rather far from the disk plane (to
be visible at non-eclipsing inclinations) and not azimuthally
symmetric (to be visible only at certain phases). How this
material gets there is unknown.
Fig. 13. (separate jpg file) Single-trailed representation of the KUV 03580+0614 spectra from 1999 October. The greyscale in the lower
panel is chosen to emphasize weaker features. The upper panel shows the same data, scaled to make visible the behaviour of the stronger
emission lines. The scale is negative (black = bright). In each panel, the data are ordered by phase and shown twice for continuity. The
individual spectra were divided by a fitted continuum. Note the phase-dependent absorption features, especially prominent in the He I
lines (λλ5876, 4921, 5015, and 6678).
c(cid:13) 1994 RAS, MNRAS 000, 000 -- 000
This figure "thorfig13.jpg" is available in "jpg"(cid:10) format from:
http://arxiv.org/ps/astro-ph/0105499v1
|
astro-ph/0506011 | 1 | 0506 | 2005-06-01T08:45:17 | Direct detection of exo-planets: GQ Lupi | [
"astro-ph"
] | We present a comparison of our VLT/NACO K-band spectrum of the GQ Lupi companion with the new GAIA-dusty model atmosphere grid for T=2000 and 2900 K and log g from 0 to 4. Then, we discuss the mass estimate for GQ Lup companion. | astro-ph | astro-ph |
Direct detection of exo-planets: GQ Lupi⋆
Ralph Neuhauser1, Eike Guenther2, and Peter Hauschildt3
1 Astrophysikalisches Institut und Universitats-Sternwarte, Schillergasschen 2,
D-07745 Jena, Germany [email protected]
2 Thuringer Landessternwarte Tautenburg, D-07778 Tautenburg, Germany
3 Hamburger Sternwarte, Gojenbergsweg 112, D-21029 Hamburg, Germany
1 Introduction: GQ Lupi and its companion
Since several years, we have been searching for sub-stellar companions around
young (up to 100 Myrs) nearby (up to 150 pc) stars, both brown dwarfs and
giant planets. Young sub-stellar objects are self-luminous due to ongoing
contraction and accretion, so that young stars are good targets. We have
found several brown dwarf companion candidates and confirmed three of
them by both proper motion and spectroscopy: TWA-5 B, HR 7329 B, and
GSC 8047 B, all being few Myr young late M-type brown dwarfs.
With K-band imaging using VLT/NaCo, Subaru/CIAO, and HST/PC,
we detected a 6 mag fainter object 0.7′′ west of the classical T Tauri star GQ
Lup, which is a clearly co-moving companion (≥ 10 σ), but orbital motion
was not yet detectable. The NaCo K-band spectrum yielded ∼ L1-2 (M9-L4)
as spectral type. At 140±50 pc distance (Lupus I cloud), it can be placed into
the H-R diagram. For more details, see Neuhauser et al. (2005). According
to our own calculations following Wuchterl & Tscharnuter (2003), it has 1
to 3 Mjup, but according to Baraffe et al. (2002) and Burrows et al. (1997)
models, it is anywhere between 3 and 42 Mjup. The latter models are not
applicable for the young GQ Lup and its companion, while the former does
take into account the formation and collapse; the Wuchterl & Tscharnuter
(2003) convection model is calibrated on the Sun.
2 Comparison with model atmospheres
We use the new so-called GAIA-dusty grid (Brott & Hauschildt et al., in
preparation), which is an updated version of the models from Allard et al.
(2001). The updates include improved molecular dissociation constants as
well as more dust species and their opacities. In addition, the models were
computed for a convective mixing length parameter of 1.5 times the pressure
scale height Hp. It uses spherical symmetry, which is most important for
young and sub-stellar objects as GQ Lup A and b with low gravities. The
Allard et al. (2001) AMES-dusty sometimes had problems at very low gravity.
⋆Proceedings ESO Workshop on The power of optical/IR interferometry
2
Ralph Neuhauser, Eike Guenther, and Peter Hauschildt
We compare our K-band spectrum (see Neuhauser et al. 2005) with the
GAIA grid for temperatures Teff = 2000 and 2900 K and for gravities of
log g = 0, 2, and 4 (g in cgs units). See figure 1 for the comparison.
Fig. 1. Comparison of our GAIA-dusty model atmosphere grid with the observed
spectrum overplotted. The high T=2900 K in the upper two spectra do not repro-
duce the water steam absorption in the blue. The model with T=2000 K and log g
=2 fits best (bottom), see text. H2O, Na, and 12CO are indicated.
The model spectrum with Teff = 2000 K is much better than the hotter
temperature (where there is no water vapour absorption in the blue part),
again indicating an early L spectral type. For the gravities, log g = 2 fits
best (log g = 4 is better than log g = 0). This gravity is fully consistent with
the gravity-sensitive CO index measured in our spectrum to be 0.862 ± 0.035,
yielding log g = 2.5 ± 0.8 according to Gorlova et al. (2003). We conclude
log g ≃ 2.0 to 3.3. The good fit indicates that the new spherically symmetric
GAIA-dusty model is better for low gravities than the former AMES-dusty.
It is important to note that the GAIA-dusty models are applicable: They
are independant of age and stand-alone, even without interior models like
the Baraffe et al. (2002) models. They are used, however, as outer boundary
conditions by, e.g., Baraffe et al. (2002) to provide their evolution models with
a more realistic description of how an interior model loses energy through the
atmosphere. Models of the solar atmosphere usually used are also stand-alone
and independant of the interior and age.
Direct detection of exo-planets: GQ Lupi
3
3 Discussion: Mass estimate for the GQ Lup companion
At 140 ± 50 pc distance, with Teff = 2050 ± 450 K and the flux of the
companion (K = 13.10 ± 0.15 mag), we can estimate its radius to be 1.2 ±
0.5 Rjup. With this radius and log g ≃ 2.0 to 3.3, we can estimate its mass
to be ≤ 1 Mjup (for log g ≃ 4 and 2 Rjup, its ∼ 6 Mjup).
The Wuchterl & Tscharnuter (2003) calculations (Fig. 4 in Neuhauser et
al. 2005) indicate ∼ 1 to 3 Mjup, co-eval with the star at ∼ 1 Myr.
Mohanty et al. (2004a) measured gravities for isolated young brown dwarfs
and free-floating planetary mass objects. Their coolest objects have spectral
type M7.5 and gravities as low as log g = 3.125 (GG Tau Bb). This lead
Mohanty et al. (2004b) to mass estimations as low as ∼ 10 Mjup. GQ Lup
A is younger than the Mohanty et al. Upper Sco objects. Its companion is
at least as late in spectral type, probably even cooler. An object younger
and cooler must be lower in mass. The GQ Lup companion is fainter than
the faintest Mohanty et al. object (USco 128, ∼ 9 Mjup), so that the mass
estimate for the GQ Lup companion is ≤ 8 Mjup.
According to Burrows et al. (1997) and Baraffe et al. (2002), the mass
of the companion could be anywhere between 3 and 42 Mjup, as given in
Neuhauser et al. (2005). However, these models are not valid at the young age
of GQ Lup, so that they are not applicable. According to both Mohanty et al.
(2004b) and Close et al. (2005), the Baraffe et al. (2002) models overestimate
the masses of young sub-stellar objects below ∼ 30 Mjup, but underestimate
them above ∼ 40 Mjup. This is consistent with our results.
Hence, according to all valid estimations, the mass of the GQ Lup com-
panion is ∼ 1 to 8 Mjup, i.e. significantly below ∼ 13 Mjup, hence almost
certainly a planet imaged directly, to be called GQ Lup b.
Acknowledgements. We would like to thank Subu Mohanty and Gibor
Basri for very fruitfull discussion about GQ Lupi b. Especially, we would like
to acknowledge Gibor Basri for pointing us to the Mohanty et al. work.
References
1. Allard F., Hauschildt P.H., Alexander D.R., Tamanai A., Schweitzer A., 2001,
ApJ, 556, 357
2. Baraffe I., Chabrier G., Allard F., Hauschildt P.H. 2002, A&A 382, 563
3. Burrows A., Marley M., Hubbard W. et al. 1997, ApJ 491, 856
4. Close L.M., Lenzen R., Guirado J.C., et al., 2005, Nature 433, 286
5. Gorlova N.I., Meyer M.R., Rieke G.H., Liebert J., 2003, ApJ 593, 1074
6. Mohanty S., Basri G., Jayawardhana R., Allard F., Hauschildt P., Ardila D.,
2004a, ApJ 609, 854
7. Mohanty S., Jayawardhana R., Basri G., 2004b, ApJ 609, 885
8. Neuhauser R., Guenther E.W., Wuchterl G., Mugrauer M., Bedalov A.,
Hauschildt P., 2005, A&A 435, L13
9. Wuchterl G., Tscharnuter W.M. 2003, A&A 398, 1081
|
astro-ph/9501102 | 1 | 9501 | 1995-01-26T11:48:00 | The Tully-Fisher relation for low surface brightness galaxies - implications for galaxy evolution | [
"astro-ph"
] | We present the B band Tully-Fisher relation for Low Surface Brightness (LSB) galaxies. These LSB galaxies follow the same Tully-Fisher relation as normal spiral galaxies. This implies that the mass-to-light ratio (M/L) of LSB galaxies is typically a factor of 2 larger than that of normal galaxies of the same total luminosity and morphological type. Since the dynamical mass of a galaxy is related to the rotation velocity and scale length via M \propto V^2 h, at fixed linewidth LSB galaxies must be twice as large as normal galaxies. This is confirmed by examining the relation between scale length and linewidth for LSB and normal galaxies. The universal nature of the Tully-Fisher relation can be understood if LSB galaxies are galaxies with low mass surface density, \sigma. The mass surface density apparently controls the luminosity evolution of a galaxy such as to keep the product \sigma M/L constant. | astro-ph | astro-ph | Mon. Not. R. Astron. Soc. , { ( )
Printed January
The Tully{Fisher relation for low surface brightness
galaxies { implications for galaxy evolution
M.A. Zwaan,
J.M. van der Hulst,
W.J.G. de Blok
and S.S. McGaugh
Kapteyn Astronomical Institute, P.O. Box , AV Groningen, The Netherlands
Institute of Astronomy, Madingley Road, Cambridge CB HA, UK
Accepted date. Received date
5
9
n
a
J
6
2
2
0
1
1
0
5
9
/
h
p
-
o
r
t
s
a
ABSTRACT
We present the B band Tully{Fisher relation for Low Surface Brightness (LSB)
galaxies. These LSB galaxies follow the same Tully{Fisher relation as normal
spiral galaxies. This implies that the mass-to-light ratio (M=L) of LSB galax-
ies is typically a factor of larger than that of normal galaxies of the same
total luminosity and morphological type. Since the dynamical mass of a galaxy
is related to the rotational velocity and scale length via M / V
h, at (cid:12)xed
linewidth LSB galaxies must be twice as large as normal galaxies. This is con-
(cid:12)rmed by examining the relation between scale length and line width for LSB
and normal galaxies. The universal nature of the Tully{Fisher relation can be
understood if LSB galaxies are galaxies with low mass surface density, (cid:22)(cid:27). The
mass surface density apparently controls the luminosity evolution of a galaxy
such as to keep the product (cid:22)(cid:27) M=L constant.
Key words: galaxies: fundamental parameters { galaxies: evolution { galaxies:
spiral { galaxies: distances and redshifts { cosmology: distance scale.
INTRODUCTION
It has become clear in recent years that disk galaxies exhibit a large range in central surface
brightness and scale length: the two parameters that describe an exponential disk. Freeman's
( ) result that the central disk surface brightness of spiral galaxies falls within a narrow
range of (cid:22)
( ) = : (cid:6) : mag=ut
, has now been superseded (e.g., de Jong and van der
B
Kruit ; Davies ). At the faint end of the surface brightness range we (cid:12)nd the so-
called Low Surface Brightness (LSB) galaxies; galaxies with (cid:22)
( ) (cid:25) mag=ut
or fainter.
B
L M.A. Zwaan, J.M. van der Hulst, W.J.G. de Blok and S.S. McGaugh
The selection e(cid:11)ects imposed by the night sky and their low surface brightnesses make LSB
galaxies di(cid:14)cult to detect. Hence, they are underrepresented in catalogues and have until
recently been studied in less detail than normal, High Surface Brightness (HSB) galaxies,
i.e., those galaxies with surface brightnesses approximately equal to the Freeman value.
The Tully{Fisher relation (Tully and Fisher , hereafter T{F), the relation between
absolute magnitude and rotational velocity, is well-established for HSB spiral galaxies. In
this paper we examine whether LSB galaxies follow the same relationship, despite their
much lower central surface brightness (see also Sprayberry et al. ). The position of LSB
galaxies in the luminosity-line width plane is particularly interesting because it will give
information about the mass-to-light ratio (M=L) for these galaxies. This parameter is still
relatively uncertain for LSB galaxies since a detailed study of rotation curves has only been
performed for one LSB galaxy (Bosma, Athanassoula and van der Hulst ).
DATA
. The samples
We selected the data from several samples from the literature, which we brie(cid:13)y discuss here.
McGaugh (McGaugh and Bothun ) and de Blok (de Blok, van der Hulst and Bothun
) studied LSB galaxies with central surface brightnesses (cid:22)
( ) (cid:25) mag=ut
. These
B
galaxies are selected from the UGC (Nilson ) and the LSB catalogue of Schombert et
al. ( , hereafter SBSM). Bulge and disk components are deconvolved from the surface
brightness pro(cid:12)les and exponential functions are (cid:12)t to the disk components. The disk mag-
nitudes are calculated by integrating the surface brightness pro(cid:12)les to in(cid:12)nity. However, the
disk magnitudes ignore any contribution of light from the bulge (which is nevertheless small
for most LSB galaxies in this sample), and will introduce extra light at larger radii. Fortu-
nately, the data necessary to calculate sky-limited aperture magnitudes are available. We
calculated these and used them in our derivation of absolute magnitudes. The inclinations
are derived from the photometric data.
Knezek (Knezek ) investigated giant LSB galaxies with mean blue surface bright-
nesses fainter than mag=ut
. The consequence of this selection criterion is that galaxies
are included with very bright central surface brightnesses. From the total sample we selected
only those galaxies with central surface brightnesses fainter than mag=ut
. We calculat-
ed the disk magnitudes from the published structural parameters; aperture magnitudes are
Tul ly{Fisher relation for low surface brightness galaxies { implications for galaxy evolution
L
Table . Properties of samples
()
()
()
()
()
Sample
N
hW
i
hM
i
h(cid:22)
( )i
R;
B
B
i
b;i
McGaugh
(cid:6)
-. (cid:6) .
. (cid:6) .
De Blok
(cid:6)
-. (cid:6) .
. (cid:6) .
Knezek
(cid:6)
- . (cid:6) .
. (cid:6) .
Total
(cid:6)
-. (cid:6) .
. (cid:6) .
Notes:
() sample name
() number of galaxies in sample
() mean corrected linewidth measured at % of peak intensity in km s
(cid:0)
() mean corrected absolute blue magnitude
() mean blue central surface brightness
not available. We determined the inclinations from the ma jor-to-minor axis ratios from the
UGC. Large errors in inclination corrected linewidth are hence unavoidable.
From these samples, we selected all galaxies with inclinations larger than
and for
(cid:14)
which a linewidth at % of the peak intensity is available. The linewidths were obtained
from the catalogue by SBSM, from LEDA
or from Schneider et al. ( , ). Table
?
presents the characteristics of each sample, including standard deviations.
. Corrections
The two parameters which de(cid:12)ne the T{F relation, the magnitude and the velocity width,
need to be corrected for inclination and absorption e(cid:11)ects. In the following we brie(cid:13)y describe
the corrections which we have applied.
We corrected the magnitudes for Galactic and internal extinction to face-on orientation,
as outlined in Tully and Fouqu(cid:19)e ( , hereafter TFq). The optical depth (cid:28) and the param-
eter f , which describes to what extent the obscuring matter is mixed with the light, are not
well determined for LSB galaxies. Hence, we adopt the standard values derived by TFq for
normal spiral galaxies: (cid:28) = : and f = :.
We corrected the line widths for random motion e(cid:11)ects and inclination as outlined in
TFq. The parameter which describes the importance of random motion is not well known
for LSB galaxies, so once again the value for HSB galaxies is used: W
= km s
.
t
(cid:0)
?
Lyon-Meudon Extragalactic Database
L M.A. Zwaan, J.M. van der Hulst, W.J.G. de Blok and S.S. McGaugh
Figure . Tully{Fisher relation for the sample of LSB galaxies. The long and short dashes represent the (cid:27) and (cid:27) range of
the (cid:12)t to the Broeils sample. One galaxy that enters the sample twice is connected.
For determining the distances to the galaxies we used a Hubble constant of km s
Mpc
.
(cid:0)
(cid:0)
We corrected the redshifts for Galactic rotation and a Virgocentric (cid:13)ow of km s
.
(cid:0)
RESULTS
In Fig. we present the B band T{F relation for the sample of LSB galaxies. As a compar-
ison we use a sample of (cid:12)eld HSB galaxies compiled by Broeils ( ). We used the same
corrections as described in section .. The slope of the T{F relation for these galaxies is
(cid:0): , which is in good agreement with slopes found for cluster samples (e.g., Pierce and
Tully ). The dispersion for the Broeils sample is . mag. This is considerably larger
than what is found for cluster samples, probably due to uncertainties in distances. The (cid:27)
and (cid:27) ranges around the (cid:12)t to the Broeils sample are represented in Fig. by long and
short dashes, respectively. Obviously, the LSB galaxies are indistinguishable from normal
galaxies in the luminosity-line width plane, that is, the T{F relation for LSB galaxies is
identical to that for HSB galaxies.
. Morphological type dependence
For normal spiral galaxies, di(cid:11)erent morphological types have di(cid:11)erent positions in the
luminosity-line width plane (e.g., Kraan-Korteweg, Cameron and Tammann ). In order
Tul ly{Fisher relation for low surface brightness galaxies { implications for galaxy evolution
L
Figure . Tully{Fisher relations binned according to morphological type. The solid circles are LSB galaxies, the open circles
are HSB galaxies from Broeils. The line is a (cid:12)t to all LSB galaxies.
to test whether LSB galaxies follow a similar trend, we made T{F plots binned according
to morphological type, both for the LSB and the HSB sample. These are shown in Fig. .
Only those LSB galaxies which have been unambiguously classi(cid:12)ed have been plotted. From
this (cid:12)gure, it is clear that LSB galaxies follow the same trend with morphological type as
the HSB galaxies.
DISCUSSION
We have shown that LSB and HSB galaxies obey the same, apparently fundamental, T{F
relation. Below we discuss the consequences of this result.
. Mass-to-light ratios
The fact that LSB galaxies follow a normal T{F relation has important implications for
the M=L of these galaxies. We derive a simple relation between the luminosity, line width,
L M.A. Zwaan, J.M. van der Hulst, W.J.G. de Blok and S.S. McGaugh
central surface brightness and M=L to illustrate this. The mass M is proportional to V
h
max
and the total luminosity L is proportional to (cid:6)
h
, where (cid:6)
is the central surface brightness,
h the scale length of the disk and V
the maximum rotational velocity. From these two
max
relations it follows that
M
M
(cid:6)
V
/
/
;
()
max
h
L
so that
V
max
L /
;
()
(cid:6)
(M=L)
where one recognises the T{F relation, L / V
. However, LSB and HSB galaxies are
max
observed to have similar luminosities at a (cid:12)xed line width, in spite of the di(cid:11)erence in (cid:6)
.
This requires that the di(cid:11)erence in surface brightness be compensated by a di(cid:11)erence in
M=L so as to keep the product (cid:6)
(M=L)
constant. The mean central surface brightness
of our sample is h(cid:22)
( )i = :, i.e., . mag fainter than typical of the Broeils sample. In
B
order to account for this factor di(cid:11)erence in (cid:6)
, M=L must be a factor of greater than
for normal spiral galaxies of similar total luminosity. This is in good agreement with the
typical values of M=L for LSB galaxies found by van der Hulst et al. ( ).
. Scale lengths
The normal T{F relation for LSB galaxies and the implication for M=L should be considered
further. Since M / V
h at any given location in the T{F diagram (i.e., at (cid:12)xed luminosity
max
and line width), M=L depends solely on h. Therefore, LSB galaxies which have a higher
M=L must also have scale lengths larger by a factor of than those of the HSB counterparts
with the same line width. The scale length is the preferred parameter for making a fair
comparison between the sizes, as D
severely underestimates the size of a LSB galaxy. The
values of the scale lengths of the Broeils sample can be derived from D
which is related
to the scale length via D
= : h assuming that these galaxies can be represented by
exponential disks with central surface brightnesses of : mag=ut
.
In Fig. we show a plot of the scale length of LSB and HSB galaxies versus line width.
A strong relationship between scale length and line width exists and there is a signi(cid:12)cant
segregation between the LSB and HSB galaxies. Although a minor di(cid:11)erence between the
slopes is present, both groups of galaxies follow a similar trend. In order to derive a reliable
value for the o(cid:11)set between the two samples, we made double regression (cid:12)ts, requiring
Tul ly{Fisher relation for low surface brightness galaxies { implications for galaxy evolution
L
Figure . Scale length versus velocity width, the solid circles are LSB galaxies, the open circles are HSB galaxies from Broeils.
the slopes to be equal. The di(cid:11)erence between the intercepts of the (cid:12)ts is . dex, which
means that LSB galaxies have scale lengths which are . times larger than those of normal
galaxies with the same line width. This is in good agreement with the factor of found for
the di(cid:11)erence in M=L for this sample of galaxies.
. Mass surface density
In addition to comparing M=L ratios and scale lengths at (cid:12)xed line width, one can consider
the implications of a universal T{F relation for the mass surface density. From L / V
max
and V
/ M=h it follows that L / M
=h
, or that
max
M=L / h
=M = = (cid:22)(cid:27) ;
()
where (cid:22)(cid:27) is the average mass surface density. In other words, galaxies with similar mass surface
densities have similar M=L ratios, i.e., the product (cid:22)(cid:27) M=L is constant. The result that LSB
galaxies have higher M=L ratios than HSB galaxies with the same rotational velocity implies
that LSB galaxies have lower mass surface densities. Another way of phrasing this result
is to combine Eq. () and () to (cid:6)
/ (cid:22)(cid:27)
; the surface brightness of a galaxy apparently
is determined by its mass surface density, i.e., LSB galaxies must be less dense than HSB
galaxies. The implication is that the mass surface density may be an important parameter
controlling the luminosity evolution of a galaxy. Less dense galaxies evolve more slowly,
L M.A. Zwaan, J.M. van der Hulst, W.J.G. de Blok and S.S. McGaugh
forming relatively fewer stars in a Hubble time, which results in lower surface brightness
disks and higher M=L ratios. This makes sense given that the stability of a disk, and hence
the star formation activity, is primarily controlled by the mass surface density (e.g., Toomre
; Goldreich and Lynden-Bell ; Kennicutt ; van der Hulst et al. , ). In
this light, the T{F relation simply describes the stability of galaxies.
CONCLUSIONS
We have shown that the B band Tully{Fisher relation for LSB (cid:12)eld galaxies is indistin-
guishable from the relation for their high surface brightness counterparts. This implies that
the Tully{Fisher relation is rather fundamental to spiral disks. Basic theoretical considera-
tions require that LSB galaxies have M=L ratios which are a factor of larger than those
of normal spiral galaxies of comparable total luminosity and morphological type. This dif-
ference implies that at a (cid:12)xed rotational velocity, LSB galaxies should have twice as large
scale lengths as normal galaxies. This is con(cid:12)rmed by the data. The universal nature of the
Tully{Fisher relation requires that galaxies of similar mass surface density have similar M=L
ratios and it is suggestive that the mass surface density of galaxies is a crucial parameter
controlling their luminosity evolution.
ACKNOWLEDGEMENTS
We thank T.S. van Albada for helpful comments on this paper.
REFERENCES
Broeils A.H. , PhD thesis, University of Groningen
de Blok W.J.G., van der Hulst J.M., Bothun G.D., , MNRAS, submitted
Bosma A., Athanassoula E., van der Hulst J.M., , A&A, ,
Davies J.I., , MNRAS, ,
de Jong R.S., van der Kruit P.C., , A&AS, ,
Freeman K.C., , ApJ, ,
Goldreich P., Lynden-Bell D., , MNRAS, ,
Kennicutt R.C., , ApJ, ,
Knezek P.M., , PhD thesis, University of Massachusetts
Kraan-Korteweg R.C., Cameron L.M., Tammann G.A., , ApJ, ,
McGaugh S.S., Bothun G.D., , AJ, ,
Nilson P., , Uppsala General Catalog of Galaxies, Uppsala Astron. Obs. Ann.,
Pierce M.J., Tully R.B., , ApJ, ,
Tul ly{Fisher relation for low surface brightness galaxies { implications for galaxy evolution
L
Schneider S.E., Thuan T.X., Magri C., Wadiak J.E., , ApJS, ,
Schneider S.E., Thuan T.X., Mangum J.G., Miller J., , ApJS, ,
Schombert J.M., Bothun G.D., Schneider S.E., McGaugh S.S., , AJ, ,
Sprayberry D., Bernstein G.M., Impey C.D., Bothun G.D., , ApJ, in press
Toomre A., , ApJ, ,
Tully R.B., Fisher J.R., , A&A, ,
Tully R.B., Fouqu(cid:19)e P., , ApJS, ,
van der Hulst J.M., Skillman E.D., Kennicutt R.C., Bothun G.D., , A&A, ,
van der Hulst J.M., Skillman E.D., Smith T.R., Bothun G.D., McGaugh S.S., de Blok W.J.G., , AJ, ,
This paper has been produced using the Blackwell Scienti(cid:12)c Publications L
T
X style (cid:12)le.
a
E
|
astro-ph/0303441 | 1 | 0303 | 2003-03-19T18:13:40 | Distinct Abundance Patterns in Multiple Damped Ly-alpha Galaxies: Evidence for Truncated Star Formation? | [
"astro-ph"
] | (abridged) Following our previous work on metal abundances of a double damped Ly-alpha system with a line-of-sight separation ~2000 km/s (Ellison & Lopez 2001), we present VLT UVES abundances of 3 new systems spanning a total of \~6000 km/s at z~2.5 toward the southern QSO CTQ247. These abundances are supplemented with echelle observations of another `double' damped Ly-alpha system in the literature. We propose a definition in terms of velocity shift of the sub-class 'multiple damped Ly-alpha system', which is motivated by its possible connection with large-scale structure. We find that the abundance ratio alpha/Fe is systematically low in multiple systems compared with single systems, and with a small scatter. The same behavior is found in 2 more single DLA systems taken from the literature that show evidence of belonging to a galaxy group. After a careful investigation of possible sources of systematic errors, we conclude that the low alpha/Fe ratios in multiple DLAs have a nucleosynthetic origin. We suggest that they could be explained by reduced star formation in multiple damped Ly-alpha systems, possibly due to environmental effects. | astro-ph | astro-ph | Astronomy & Astrophysics manuscript no. lopez
(DOI: will be inserted by hand later)
March 31, 2017
Distinct Abundance Patterns in Multiple Damped Lyα Galaxies:
Evidence for Truncated Star Formation?⋆
Sebastian Lopez1 and Sara L. Ellison2
,
3
1 Departamento de Astronom´ıa, Universidad de Chile, Casilla 36-D, Santiago, Chile
e-mail: [email protected]
2 European Southern Observatory, Casilla 19001, Santiago 19, Chile
3 Pontificia Universidad Cat´olica de Chile, Casilla 306, Santiago 22, Chile
e-mail: [email protected]
Received / Accepted
Abstract. Following our previous work on metal abundances of a double damped Lyα system with a line-of-sight
separation ∼ 2 000 km s−1(Ellison & Lopez 2001), we present VLT UVES abundances of 3 new systems spanning
a total of ∼ 6 000 km s−1 at z ∼ 2.5 toward the southern QSO CTQ247. These abundances are supplemented with
echelle observations of another 'double' damped Lyα system in the literature. We propose a definition in terms of
velocity shift of the sub-class 'multiple damped Lyα system', which is motivated by its possible connection with
large-scale structure. We find that the abundance ratio [S/Fe] is systematically low in multiple systems compared
with single systems, and with a small scatter. The same behavior is found in 2 more single DLA systems taken
from the literature that show evidence of belonging to a galaxy group. Although [Si/Fe] ratios are also generally
lower in multiple DLAs than in single DLAs, the effect is less striking since the scatter is larger and there are
a number of low [Si/Fe] DLAs in the literature. We suggest that this can be explained with a combination of
detection bias and, to a lesser extent, the scatter in ionization corrections for different absorbers. We investigate
whether consistently low α/Fe ratios could be due to dust depletion or ionization corrections and find that the
former effect would emphasize the observed trend of low α/Fe in multiple systems even further. Ionization may
have a minor effect in some cases, but at a level that would not change our conclusions. We thus conclude that
the low α/Fe ratios in multiple DLAs have a nucleosynthetic origin and suggest that they could be explained by
reduced star formation in multiple damped Lyα systems, possibly due to environmental effects. There seems to
be independent evidence for this scenario from the mild odd-even effect and from the relatively high N/α ratios
we observe in these multiple systems.
Key words. Quasars: general -- quasars: absorption lines -- quasars: individual: CTQ247 -- galaxies: evolution --
galaxies: clusters: general
3
0
0
2
r
a
M
9
1
1
v
1
4
4
3
0
3
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
1. Introduction
Damped Lyman Alpha systems (DLAs) in the spec-
tra of high-redshift quasars have promised to be a
powerful probe of early metal enrichment in young
galaxies. However, despite the considerable effort in-
vested to develop this technique, its full potential has
yet to be realized. Although extensive observing cam-
paigns at intermediate resolution have swollen the list
of known DLAs (Ellison et al. 2001b, Lopez et al. 2001
and Peroux et al. 2001 are some of the most recent DLA
surveys), only a fraction have been followed-up at high
resolution to study their chemical properties, investigate
Send offprint requests to: S. Lopez
⋆ The work presented here is based on data obtained at the
ESO Very Large Telescopes, programs 68.A-0361(A) and 66.A-
0660(A).
dust content, photoionization and metallicity evolution.
With the present sample of some 60 abundance mea-
surements some broad trends have been identified (e.g.,
Prochaska & Wolfe 2002), but the global picture of star
formation (SF) history deduced from these patterns of el-
emental abundances -- likely affected by dust depletion pat-
terns (Hou, Boissier & Prantzos 2001) -- is still unclear. In
particular, extracting coherent scenarios of SF histories in
individual DLAs, or subsets of absorbers remains a chal-
lenging prospect with current data.
One of the great advantages of studying high redshift
galaxies in absorption is the lack of selection bias asso-
ciated with cosmological dimming: galaxies are selected
based only on gas cross section, irrespective of their intrin-
sic luminosity. However, this feature may turn out to be a
double-edged sword in that extracting coherent chemical
enrichment patterns is almost certainly hindered by the
2
Sebastian Lopez and Sara L. Ellison: Abundances in Multiple DLAs
Table 1. VLT Observations of CTQ 247
Mode
Wavelength
Exp. Time
Observing Date
FORS2 (GRISM 600B)
FORS2 (GRISM 600R)
UVES Dichroic (390+580)
UVES Dichroic (437+860)
[nm]
345-590
525-745
328-445,476-684
376-500,667-1040
[sec]
600
400
13 500
10 800
Dec. 18 2000
Dec. 18 2000
Oct. 10, 11 and Nov.17, 2001
Nov. 17 2001
mixed morphological representation. Undoubtedly, some
of the scatter in relative abundances comes from vari-
able dust depletion, although system-to-system differences
persist even after correcting for grain fraction (Vladilo
2002). In order to compare elemental trends in DLAs
in a similar manner to, for example, local stellar abun-
dance studies, a significant refinement in our approach is
required. Such a refinement might include isolating par-
ticular subsets of DLAs based either on their morphology
or some other observed property (e.g. spin temperature;
Chengalur & Kanekar 2000).
In this paper, we investigate whether a distinct enrich-
ment trend exists for absorption line systems with close
companions in velocity space, which we will refer to as
'Multiple Damped Lyman Alpha systems', or MDLAs.
2. The Sub-class 'MDLA'
Lopez et al. (2001) identified the existence of three
damped Lyα absorbers
spanning a total of v =
5 900 km s−1 at z ∼ 2.5 toward the southern quasar
CTQ 247, the first ever example of a triple DLA; Ellison
et al. (2001b) discovered a double absorption systems for
which v = 1 800 km s−1 at z ∼ 2 toward Q2314−409;
a similar velocity span as in CTQ 247 is covered by the
two DLAs toward Q2359−02 at z ∼ 2 (Wolfe et al. 1986).
Statistically, these are extremely improbable events if the
proximity of the systems is purely by chance, and only
these few cases are reported in the literature. It is there-
fore feasible to postulate that in these cases the multiple
absorption systems may be in some way related, rather
than chance alignments. However, such large velocities
are difficult to reconcile with today's galaxy groups and
clusters -- i.e., virialized entities -- or even with superclus-
ters. However, at z > 0 current work on small angular
fields shows evidence for very large structures on comoving
scales as large as d ∼ 100 h−1
100 Mpc (∆v ∼ 10 000 km s−1 if
purely due to Hubble flow) at various redshifts. Evidence
for large structure on many Mpc scales has been found
in absorbing gas both across (e.g., Williger et al. 2002;
d ∼ 100 h−1 Mpc at z ∼ 1.3 ) and along the line of sight
(Quashnock, vanden Berk & York 1996; d ∼ 100 h−1
in super-clusters (d ∼ 20 h−1 Mpc
Mpc at z ∼ 3),
at z ∼ 0.8; Haines, Campusano & Clowes 2003) and in
Lyman-break galaxies (Steidel et al. 1998; d ∼ 10 h−1
Mpc at z ∼ 3 ). Large structure has also been reproduced
in CDM N-body simulations (d ∼ 20 h−1 Mpc at z ∼ 1;
Evrard et al. 2002). It therefore seems a feasible (but not
exclusive) possibility that absorption systems separated
by several thousand km s−1 may be associated with very
large scale structure.
Motivated both by the uniqueness of MDLAs and the
possibility of large-structure, we have embarked on a pro-
gram to measure abundances in these systems. We will
define an MDLA as 2 or more DLAs spanning a velocity
range
500 < ∆v < 10 000 km s−1.
(1)
While the upper limit is set to match the limits of the
largest structures known at high redshift, the lower veloc-
ity roughly corresponds to the largest asymptotic veloci-
ties in massive spirals (e.g. Rigopoulou et al. 2002)1. Note
that this a working definition only; although it might be
possible that MDLAs probe some kind of large structure
(but see § 5), our definition does not necessarily imply that
they host the strong C iv absorption used by Quashnock,
vanden Berk & York (1996) or they are associated with
the LBGs identified by Steidel et al. (1996).
From an instrumental point of view,
identifying
MDLAs in Lyα -- in contrast to using metal lines -- will
certainly limit the identification of systems separated by
∆v <∼ 1 000 km s−1. If the discovery spectrum has poor
S/N, obtaining line parameters for close MDLAs will re-
quire both the detection of metal lines, and a good con-
tinuum estimation2.
For N (H i) criterion, we will relax the canonical limit
to include systems with N (H i) > 1020 cm−2. Not only
do systems down to this limit continue to exhibit clear
damping wings, but in the following we show that in such
range ionization does not yet play a significant role when
deriving abundances from low-ions.
The first in-depth study of an MDLA was presented by
Ellison & Lopez (2001; hereafter Paper I). There, the two
DLAs toward Q2314−409 were noted to have low [S/Fe]
ratios, a result which we suggested was due to the pos-
sible impact of environment on the chemical abundances
1 This lower limit will probably exclude the majority
of bound galaxy satellites since most local spirals' satel-
less than 300 km s−1,
lites have velocity differences of
Zaritsky et al. (1997)
2 Two N (H i) = 2 × 1020 cm−2 DLAs separated by ∆v =
1 000 km s−1 will mimic a single system in FWHM ∼ 5 A data
if S/N <
∼ 20, but for ∆v ∼ 500 km s−1 they will do so unless
S/N >
∼ 50.
Sebastian Lopez and Sara L. Ellison: Abundances in Multiple DLAs
3
in DLA protogalaxies. In the present work, we extend the
study of abundances in MDLAs by investigating the triple
DLA toward CTQ247 with new echelle data, and includ-
ing literature abundances of the double system toward
Q2359−02. These new abundances reinforce our previous
suggestion of peculiar relative abundances in MDLAs.
After presenting the new data in § 3, the discussion
of the abundances is given in § 4, where we attempt a
statistical comparison between the 2 populations, and as-
sess possible systematic effects that may bias our result.
A discussion on the nature of MDLAs is outlined in § 5
3. Data Analysis
3.1. Observations and Data Reduction
CTQ 247 was observed in service mode with the UVES
instrument at the ESO Kueyen telescope in October and
November 2001 under good seeing conditions (0.6′′ −1.0′′).
With the 390+580 and 437+860 dichroic modes we cov-
ered from 328 to 1020 nm with two gaps at 576-583 nm
and 852-866 nm. The exposure times were 13 500 seconds
for dichroic 1 and 10 800 seconds for dichroic 2 (Table 1).
After the usual fashion of bias-subtracting and flat-
fielding of the individual CCD frames, the echelle orders
were extracted and reduced interactively with the UVES
pipeline routines (Ballester et al. 2000). The reference Th-
Ar spectra used for wavelength calibration were taken af-
ter each science exposure. The wavelength values were
converted to vacuum heliocentric values and each order
of a given instrumental configuration was binned onto a
common linear wavelength scale of 0.04 A pixel−1. The
reduced orders were then added with a weight according
to the inverse of the flux variances. Finally, the flux values
were normalized by a continuum that was defined using
cubic splines over featureless spectral regions. The spec-
tral resolution is FWHM ∼ 6.7 km s−1, while the typical
signal-to-noise ratio per pixel is S/N∼ 35 − 40.
In addition to the UVES data, a lower resolution
(FWHM ≈ 3 A) spectrum of CTQ 247 was obtained
on December 18 2000 using FORS2 at the ESO Kueyen
Telescope. This spectrum was used to better define the
quasar continuum in the spectral region around the
damped Lyα lines.
3.2. Column densities
We used FITLYMAN and VPFIT3 to fit the line profiles
with theoretical Voigt profiles. All fits were unconstrained
in redshift, Doppler width and column density, unless oth-
erwise stated. In general, we prefer this approach over
the apparent optical depth method (AODM; Savage &
Sembach 1991) when velocity components are not resolved
as is sometimes the case. In addition, when dealing with
transitions that lie in the Lyα forest, such as the S ii
triplet, fitting provides important information to the ex-
tent of possible blending. Despite this, however, we did use
3 Available at http://www.ast.cam.ac.uk/rfc/vpfit.html
the AODM for transitions where the fit failed due to line
blending of many components or poor S/N. We adopted
the up-to-date f -values listed in Prochaska et al. (2001)
and the solar abundances by Grevesse & Sauval (1998),
with updates for O and N by Holweger (2001). Table 2
lists the column densities and derived abundances of all
elements covered by our observations.
The H i column density was determined by fitting 3
components to the damped Lyα and Lyβ profiles in the
FORS2 spectrum. The choice of the low resolution spec-
trum reduces the uncertainties inherent to determining
N (H i) from echelle data, such as badly defined contin-
uum where the damping wings extend over more than one
order. We obtained N (H i) = 1021.13, 1021.09, and 1020.47
cm−2, for the 3 systems centered at zabs = 2.5505, 2.5950
and 2.6215. The theoretical profiles are superposed on the
data in Fig. 1. The internal fit errors were 0.02 dex for all
3 H i measurements but we believe a more realistic error
is given by ±0.1, which is shown by the dotted curve in
the figure. Although the noisy data around Lyβ constrains
H i only marginally, this error seems quite safe, given the
higher S/N at Lyα.
We next give a short description of the metal line fits
in each of the 3 DLAs which we henceforth refer to as
CTQ 247A,B and C.
CTQ247A (zabs = 2.5505): For this DLA we have identi-
fied nonsaturated transitions of N i, Si ii, S ii, Cr ii, Fe ii,
and Zn ii, all of which show 4 velocity components. More
(weaker) components are revealed by stronger transitions,
e.g. Fe ii λ2344, but these do not contribute significantly
to the total column densities. Therefore, in order to avoid
introducing too many parameters at intrinsically weak
transitions in relatively noisy parts of the spectrum, we
proceeded by fitting 4 components. Fig. 2 shows the non-
saturated transitions used and their fits. From an analysis
of the weaker components we estimate that the obtained
column densities encompass ∼ 90 % of the total column
density; exclusion of weaker components will lead to only
a ∼ 0.05 dex difference at log N = 15 cm−2. The S ii 1260
triplet lines fall on the red wings of the damped Lyα lines;
consequently, they were fitted simultaneously with the H i
fixed at the values found for the FORS data, hence the
sub-unity values of the continuum in Fig. 2. In addition,
we determine an AODM measurement for Mn ii and an
upper limit for P ii.
CTQ247B (zabs = 2.5950): This DLA has a relatively sim-
ple velocity structure. Although the strongest Si ii and
Fe ii transitions have 5 velocity components, these span
only ∼ 80 km s−1. As expected for such a high H i, O i is
saturated and we consider it no further here. Conversely,
N i which is often barely detected exhibits a large equiva-
lent width in this system. In fact, two of the λ1134 triplet
transitions are saturated. However, the triplet is relatively
unblended, as are two of the S ii triplet lines. Fig. 3 shows
the nonsaturated transitions used and their fits. There is
4
Sebastian Lopez and Sara L. Ellison: Abundances in Multiple DLAs
Fig. 1. Normalized flux and 1σ VLT FORS2 spectrum of CTQ 247 at FWHM = 3 A resolution showing the Lyα and
Lyβ lines of the triple DLA. Overlaid is a synthetic Voigt profile of three systems with N (H 1)= 1021.13, 1021.09, and
1020.47 cm−2, centered at zabs = 2.5505, 2.5950, and 2.6215. The dotted lines are ±0.1 dex deviations for all N (H i)
values.
an absorption feature at the expected position of P ii λ963
but since we suspect contamination by a Lyα forest line we
only provide an upper limit on N (P ii). As for CTQ247A,
we also provide an AODM measurement for Mn ii.
only values obtained from echelle data. In cases where
numbers are present in several references, we have pre-
ferred the values by Prochaska et al. (2001) for consis-
tency.
CTQ247C (zabs = 2.6215): This is the lowest H i DLA
of the triplet. For this DLA the bulk of the low ions are
divided into 2 'sub-clumps' of 3 and 2 velocity components
separated by ∼ 120 km s−1. Fig. 4 shows the nonsaturated
transitions used and their fits. Since this is the DLA with
the lowest H i of the 3, fewer species are detected and limits
are not very stringent. For example, for sulphur our 3σ
detection limit is log N (S ii) < 14.3. We have attempted
a fit for the N i triplet, but a combination of many weak
components and Lyα blending makes this a very uncertain
result and we regard it as an upper limit due to possible
contamination. On the other hand, the low H i allows us
to profile fit 2 O i transitions, for which each one of the
5 components shows at least one nonsaturated line. Also,
we report one of the few existing measurements of Al ii in
DLAs (line positions were held fixed at the Si ii values),
and one AODM measurement for Al iii.
4. Abundances in MDLAs
In Paper I we focused on the relative α to Fe peak abun-
dances in order to establish whether MDLAs may ex-
hibit distinct SF histories. Similarly, in Fig. 5 the solid
circles depict [α/Fe] vs. [Fe/H] for the 3 MDLAs pre-
sented in this paper (CTQ247A, B, and C), the dou-
ble DLA Q2314-409A, B (Paper I), and Q2359−02 A,
B (Prochaska & Wolfe 1999). The stars are DLAs taken
from the literature. Due to the potential problem of blend-
ing between S lines and the Lyα forest, we have compiled
In the following we show statistically that the sub-
class MDLA is drawn from a different distribution to sin-
gle DLAs. We then discuss to which extent dust and ion-
ization effects might produce such distinction, and what
possible systematic effects may be present in the 2 sam-
ples.
4.1. Low α/Fe in MDLAs
As expected in DLAs, all points in Fig. 5 fall over the
[α/Fe] = 0 mark, as both S and Si are α-chain elements
which are created mainly in massive stars on much shorter
scales than the long-lived iron producing stars. The ra-
tio of [α/Fe]
is usually interpreted as an indicator of
star formation history and the rate at which the differ-
ent elements are released into the ISM. In addition to
the MDLAs in Fig. 5, we have also plotted DLAs with
other neighboring galaxies (usually detected via Lyman
break imaging) in the field. These systems are Q0201+112
(Ellison et al. 2001a) which has 4 identified galaxies in
the vicinity, and Q0000-262 (Molaro et al. 2001) for which
Steidel et al. (1996) found 2 Lyman break galaxies within
∆z = 0.05 of the absorber.
The remarkable property to note in Fig. 5 is that all
MDLAs systematically show low [S/Fe] values with a small
scatter (< 0.2 dex). To quantify this impression, we have
performed a modified KS test that provides the likelihood
of MDLAs with [S/Fe] measurements to be drawn from
a different distribution to single DLAs. We created 1000
realizations of a simulation which recreates the [S/Fe] dis-
Sebastian Lopez and Sara L. Ellison: Abundances in Multiple DLAs
5
CTQ247A
CTQ247B
CTQ247C
N(X)
21.13 ± 0.10
14.55 ± 0.03
[X/H]
...
−2.51 ± 0.10
N(X)
21.09 ± 0.10
15.07 ± 0.02
[X/H]
...
N(X)
20.47 ± 0.10
−1.95 ± 0.10
< 14.36
[X/H]
...
< −2.04
−2.02 ± 0.10
−1.97 ± 0.12b
...
−2.04 ± 0.12
...
< −1.33
...
...
15.19 ± 0.02
12.88 ± 0.07
12.33 ± 0.03a
13.99 ± 0.06
...
< 14.34
...
...
13.60 ± 0.02
−2.37 ± 0.10
...
...
...
...
...
...
X
H i
N i
O i
Al ii
Al iii
Si ii
P ii
S ii
Cr ii
Mn ii
Fe ii
Fe iii
Ni ii
Zn ii
...
...
12.81 ± 0.05a
15.32 ± 0.04
< 13.15
...
...
...
−1.37 ± 0.11
< −1.54
...
...
12.63 ± 0.04a
15.59 ± 0.03
< 12.98
...
...
...
−1.06 ± 0.10
< −1.67
−1.51 ± 0.12
14.82 ± 0.06
13.20 ± 0.03
−1.62 ± 0.10
12.86 ± 0.02a −1.80 ± 0.10
−1.68 ± 0.12
14.95 ± 0.06
...
...
13.94 ± 0.02
12.44 ± 0.05
−1.44 ± 0.10
−1.36 ± 0.11
−1.10 ± 0.11
15.19 ± 0.05
13.37 ± 0.04
−1.41 ± 0.11
12.87 ± 0.01a −1.75 ± 0.10
−1.44 ± 0.10
15.15 ± 0.02
< 13.66
13.86 ± 0.31
12.68 ± 0.02
...
−1.48 ± 0.33
−1.08 ± 0.10
Table 2. Abundance measurements (and 3σ upper limits) for CTQ247. a uses AODM; b N (Al) = N (Al ii) + N (Al iii)
tribution of literature DLAs and MDLAs including the 1
σ quoted errors. For each DLA, an error was drawn at
random from a Gaussian distribution and added to the
observed value4. A KS test was then run on each of the
1000 datasets and the likelihood that the KS probability
PKS was greater than a certain confidence level was calcu-
lated from the number of realizations. In this fashion, we
overcome 2 problems associated with a normal KS statis-
tics: the small number of MDLAs we have at our disposal
and the exclusion of measurement errors.
According to our simulations, the likelihood p that
the two samples are drawn from different distributions is
p = 1.00 for PKS > 95 % (i.e. every one of the 1000 realiza-
tions gives a PKS > 95 %), p = 1.00 for PKS > 98 %, and
even as high as 0.97 for PKS = 99 %. This indicates that
the consistently low [S/Fe] values in MDLAs and other
multiple systems are statistically significant. Since Si and
S are both produced in oxygen burning reactions, [S/Si]
= 0 in all observed Galactic disk stars (Chen et al. 2002)
and we may expect to see similar patterns between S and
Si in DLAs. However, the MDLA [Si/Fe] values have a
slightly larger dispersion and there are a number of low
literature [Si/Fe] values, which renders the two popula-
tions less distinguishable. This may be explained because
low N (S ii) lines are easily contaminated in the Lyα for-
est, whereas low N (Si ii) can be much more easily mea-
sured. If we consider only the literature DLAs with Si and
S measurements, the KS probability that the two [Si/Fe]
samples are consistent is 11%, as opposed to 32% if all
are included. There is an obvious outlier in the [Si/Fe]
distribution associated with system Q2359−02A; in the
following section we argue that this system has anoma-
lously large dust content compared with the rest of the
MDLAs and literature DLAs. Discounting this exception-
4 We treat the upper limit toward DLA B in Q2314−409
as a detection; a lower value than this does not affect the KS
probability
ally dust depleted system reduces the KS probability even
further of all [Si/Fe] measurements to 9%. Therefore, al-
though there is a hint that [Si/Fe] may be low in MDLAs,
the evidence is not as convincing as that for S. We explore
this further in the following section.
4.2. How dust and/or ionization might affect the
observed ratios
Before discussing the implications of the observed low
[α/Fe] ratios in terms of a nucleosynthetic origin, we must
first rule out other (non-intrinsic) systematics. In Paper
I, we have already discussed the issue of dust and pho-
toionization in the double DLA toward Q2314-409. Here,
we embark upon a similar assessment for the CTQ247
triplet. In particular, we investigate whether systematics
may cause (a) the α/Fe ratio to be lower in MDLAs than
in single DLAs; and (b) the α under-abundance in MDLAs
to be more evident in S rather than Si.
First, we consider dust depletion. Although disentan-
gling dust depletion from pure nucleosynthesic effects in
DLAs is still matter of debate, some broad trends can
be established: (1) In no instance is S known to be de-
pleted in the Galactic ISM (e.g. Savage & Sembach 1996).
Conversely, Si is mildly depleted in most environments and
Fe is very easily incorporated into dust. Therefore, dust
depletion will cause an increase in both [S/Fe] and [Si/Fe],
the effect being most pronounced in the former ratio, so
[S/Si] > 0. (2) Zn is undepleted; thus, assuming the same
nucleosynthetic origin as Fe, one expects [Zn/Fe] > 0 in
dusty systems.
Both in CTQ247A and B the [Zn/Fe] shows a signifi-
cant departure from solar, indicating that dust is present.
This is also true for Q2314−409, the other MDLA for
which a Zn abundance is determined; in all cases [Zn/Fe] >
0.3, typical of other DLAs. Therefore, any downward cor-
rection due to dust would further emphasize the low α/Fe
in these MDLAs. It is noteworthy that Q2359−02A, which
6
Sebastian Lopez and Sara L. Ellison: Abundances in Multiple DLAs
Fig. 3. Same as in Fig. 2 but for CTQ247B (zabs =
2.5950)
has the largest [Si/Fe] of all, also shows an unusually large
dust content, [Zn/Fe]=+0.88, explaining its anomalously
large [Si/Fe]. Not only is Q2359−02A relatively highly de-
Fig. 2. UVES spectrum showing nonsaturated transitions
in CTQ247A (zabs = 2.5505). The solid line in the P ii
and Mn ii panels indicates the velocity region used for the
AODM.
Sebastian Lopez and Sara L. Ellison: Abundances in Multiple DLAs
7
Fig. 5. Relative ratios of α/Fe for MDLAs (solid circles),
the DLAs toward Q0201+1120 and Q0000−262 (solid tri-
angles) and DLAs taken from the literature (stars). All
ratios have been corrected to the following solar val-
ues:
log
(Fe/H)⊙ = −4.50 (Grevesse & Sauval 1998). The data
points shown in this plot are listed in Table 3.
log (S/H)⊙ = −4.80, log (Si/H)⊙ = −4.44,
Fig. 4. UVES spectrum showing nonsaturated transitions
in CTQ247C (zabs = 2.6215)
pleted compared to other MDLAs, but also in comparison
with other DLAs which mostly have [Zn/Fe] < 0.6.
Can low dust content in MDLAs compared with sin-
gle DLAs be responsible for lower [S,Si/Fe] in the for-
mer? Fig. 6 shows [S/Si] vs. [S/Fe] for the (few) systems
where both elements are available, including 3 MDLAs.
It can be seen that apart from one DLA with an ex-
tremely large [S/Fe], there is no strong trend for sys-
tems with lower [S/Si] (supposedly less dusty) to show
lower [S/Fe], suggesting that low S/Fe ratios are not
strongly effected by dust. Indeed, the large S abundance
measured in Q0307−49 might be due to Lyα blending
(Bonifacio et al. 2001) and must be considered with cau-
tion. Therefore, low S/Fe is not obviously correlated with
low dust content, at least not to the accuracy we can
Fig. 6. [S/Si] vs. [S/Fe] in MDLAs (circles), single DLAs
(stars), and in the DLA toward Q0000−262 (triangle).
presently measure this ratio. In conclusion, it does not
seem from the present sample that a general trend of low
α/Fe values be driven by any effect of dust, nor do MDLAs
have atypical dust content.
8
Sebastian Lopez and Sara L. Ellison: Abundances in Multiple DLAs
QSO
[Fe/H]
[Zn/Fe]
[Si/Fe]
[S/Fe]
Ref.
Q0013−004
Q0100+13
Q0149+33
Q0201+3634
Q0216+0803A
Q0216+0803B
Q0255+00A
Q0255+00B
Q0307−49
Q0336−01
Q0347−38
Q0741+474
Q0836+11
Q0957+33A
Q0957+33B
HE1104−18
Q1108−07
Q1210+17
Q1215+33
Q1223+17
Q1331+17
Q1409+095A
Q1759+75
Q1946+7658
Q2206−19
Q2206−19
Q2230+02
Q2231−00
Q2237−060
Q2243−603
Q2343+1232
Q2344+12
Q2348−01A
Q2348−01B
CTQ247A
CTQ247B
CTQ247C
Q2314−409A
Q2314−409B
Q2359−02A
Q2359−02A
-1.83
-1.90
-1.77
-0.87
-0.97
-1.06
-1.44
-2.05
-1.96
-1.80
-1.80
-1.93
-1.40
-1.45
-1.87
-1.60
-2.12
-1.15
-1.70
-1.84
-2.06
-2.30
-1.21
-2.53
-0.86
-2.61
-1.17
-1.31
-2.17
-1.25
-1.23
-1.83
-1.39
-2.24
-1.68
-1.44
-2.37
-1.32
-1.87
-1.66
-1.88
1.09
0.31
0.10
0.59
...
...
...
...
...
...
−0.10
...
...
...
...
0.57
...
0.25
0.41
0.22
...
...
...
0.45
...
0.45
0.54
...
0.13
...
...
...
...
0.32
0.36
...
...
...
...
...
...
0.48
0.28
0.46
0.30
0.50
0.50
...
0.41
...
0.46
0.24
0.25
0.45
0.37
0.55
0.32
0.27
0.22
0.25
0.61
0.28
0.40
0.30
0.44
0.30
0.41
0.44
0.36
0.38
0.47
0.09
0.69
0.26
0.31
0.38
0.33
0.27
0.00
0.88
0.29
0.79
0.44
...
...
...
...
...
0.27
1.00
0.39
0.51
0.25
...
...
0.56
0.53
...
...
...
...
0.76
...
0.46
...
...
...
...
...
...
0.40
0.40
...
...
...
0.17
0.34
<1.04
0.32
<0.21
...
...
m
b,c
a,b
a,n
d
d
a
a
e,h
a
a,h,i
a
a
a
a
f
a
a
a
a
a,b,o
k
a
a
a
a
a,b
b
c,d
g
c,l
a
a
a
p
p
p
q
q
b
b
Table 3. Abundance ratios: aProchaska et al. (2001);
bProchaska & Wolfe (1999); cLu, Sargent & Barlow (1998);
dLu et al. (1996); eDessauges-Zavadsky et al. (2001);
f Lopez et al. (1999); gLopez et al. (2002); hBonifacio et al.
(2001); iLevshakov et al. (2002); jGe, Bechtold & Kulkarni
(2001); kPettini et al. (2002); lProchaska, Gawiser & Wolfe
(2001); mCenturion et al. (2000); nProchaska & Wolfe
(1996); oDessauges-Zavadsky, D'Odorico & Prochaska (in
prep.); pThis paper; qPaper I
We now turn to ionization. Given the large N (H i)
of CTQ247A and B,
it is highly unlikely that ioniza-
tion corrections are required (Viegas 1995). Nonetheless,
Prochaska et al (2002c) found significant corrections in a
DLA with log N (H i) = 20.8, so there are occasional cases
where photoionization plays a part even in relatively large
column density systems. In order to constrain the ion-
ization parameter for CTQ247A and B we use CLOUDY
(Ferland 1993) models with a log N (H i) = 21.1, a Haardt-
Madau ionizing spectrum (Haardt & Madau 1996) at z =
2.3 and a metallicity of −1.5 solar, and use the ob-
served ratios to constrain the ionization parameter. For
CTQ247A, R ≡ log N (Al iii)/N (Al ii) provides the most
stringent limit on log U , whereas the ratio of Fe iii/Fe ii
provides most information on the ionization of CTQ247B.
In both cases, the limiting ionization parameter is found to
be log U < −3.2, see upper panel of Fig. 7. Consequently,
this limits the correction to [S/Fe] to be less than 0.07 dex.
Note that a correction this extent is in agreement with the
subsolar [S/Si] observed for the 3 MDLAs in Fig. 6 and
brings this ratio into line with solar, within the measure-
ment errors.
CTQ247C has a significantly lower H i column density.
Our spectrum does cover the Fe iii λ1122 transition but
the limit is not very meaningful. However, we can com-
pare Al ii and Al iii whose ratio has been shown to exhibit
a steady trend with N (H i), a higher fraction of Al iii be-
ing present for lower N (H i) (Vladilo et al. 2001). We find
R = −0.55 which is typical of moderate column density
DLAs, log N (H i) > 20.5. Repeating the above CLOUDY
models with the appropriate H i confirms that the cor-
rections to [S/Fe] will not exceed 0.1 dex in this MDLAs
(bottom panel in Fig. 7). Therefore, we would need to
make a small upward correction to [S/Fe] of up to about
0.1 dex but this does not alter our conclusions.
We note that the MDLAs toward Q2359−02 have
both high R ≈ −0.3, indicating relatively high ionization.
According to our CLOUDY models, however, the correc-
tions for [S/Fe] are still negligible for the corresponding
ionization parameter, while [Si/Fe] requires downward cor-
rections of 0.1 -- 0.2 dex.
From a more general point of view, Fig. 7 tells us that
for a typical log N (H i) = 20.5 DLA the corrections are
downward for Si/Fe and upward for S/Fe, whereas at the
low end of H i column densities the [Si/Fe] ratio is signif-
icantly more sensitive to ionization than [S/Fe] and both
corrections are downward (cf. Prochaska et al. 2002c).
Ionization will therefore require downward corrections in
[Si/S] but these will not necessarily contribute to the wider
spread of MDLA [Si/Fe] values we observe in Fig. 5.
We can quantify the effect of ionization on the compar-
ative dispersion between [Si/Fe] and [S/Fe] by performing
Monte-Carlo simulations with a grid of CLOUDY mod-
els. To this end, we constructed 2 samples of N = 500
CLOUDY models each, with −4 < log U < −3 ('low-
ionization' sample) and −4 < log U < −2 ('high+low-
ionization' sample), and H i column density and metal-
licity values uniformly distributed in the ranges 19.4 <
log N (H i) < 22.0, −2.4 < log(Z/Z⊙) < −0.8 5. To simu-
late our observations we selected randomly 1 000 sets of 20
5 Of course, uniformly distributed values of H i and Z do
not represent the observed distribution of DLAs, this range is
purely for the purpose of investigating the effects of ionization.
Sebastian Lopez and Sara L. Ellison: Abundances in Multiple DLAs
9
Fig. 7. Ratio of doubly to singly ionized species as a
function of ionization parameter U from CLOUDY models
with a Haardt & Madau (Haardt & Madau 1996) ionizing
spectrum. Top panel: Model with log N (H i) = 21.1 and
gas metallicity log Z = log Z⊙ − 1.5 compared with data
points for CTQ247A (open square) and CTQ247B (filled
squares). Middle panel: The measured Al iii/Al ii ratio
for CTQ247C, and a model using log N (H i) = 20.5 and
metallicity log Z = log Z⊙ − 2.5. Bottom panel: A model
using log N (H i) = 20.0, the column density cutoff in our
definition of MDLAs, and metallicity log Z = log Z⊙−1.5.
In all panels the dashed lines indicate solar-corrected ra-
tios while the solid lines are for column density ratios.
DLAs each from the two samples. Their [X/Fe] distribu-
tions are shown in the upper panel of Fig. 8. The difference
between the scatters in [Si/Fe] and [S/Fe] was calculated
as ∆S ≡ σ[Si/Fe] − σ[S/Fe], where σ[X/Fe] is the stan-
dard deviation of each dataset. The bottom panel of the
figure shows the ∆S distribution for both samples. For the
low+high-ionization sample the probability P (∆S > 0)
of having a wider scatter in [Si/Fe] is roughly 0.7, while
P (∆S > 0) = 0.85 if the DLAs are drawn exclusively from
the low-ionization sample. This is an indication that ion-
ization does affect the distributions differentially, although
the differences are generally below measurement errors.
4.3. Other possible systematic effects
Finally, other possible systematic effects we have to be-
ware of are: (1) Column densities: our column densities
come from χ2 fits whereas the majority of literature DLAs
have AODM measurements. However, it is well established
that both approaches give essentially same results pro-
vided the lines are nonsaturated as is our case. (2) Solar
abundances: solar abundances have undergone some sig-
Fig. 8. Simulated [S/Fe] and [Si/Fe] distributions (top
panel) and difference between their scatters (bottom
panel). The solid (dotted) histogram shows DLA simula-
tions drawn from a sample of DLAs with −4 < log U < −3
(−4 < log U < −2 )
nificant revisions in recent years; all the points in figures 5
and 9 have been put on a common solar scale. (3) Atomic
data: some f -values have changed in recent years, most
notably Fe ii λ1611. However, this should not be an issue
as long as very few systems rely on only one transition
6. (4) Blending: S ii transitions are normally in the forest
and therefore particularly susceptible to contamination.
This is especially a worry if the AODM is used since it
would cause larger N (S) if contamination is present. Since
40% of literature S measurements comes from the UCSD
database which uses the AODM, in theory we could have
compared our data with overestimations of S/Fe. However,
checking in Prochaska et al. (2001) on a case-by-case ba-
sis shows that this is unlikely to be a problem because the
possibility of contamination was carefully assessed in each
case. Even excluding [S/Fe] points determined by AODM
6 e.g., Q2206−19A is the only DLA in the sample that relies
solely on Fe ii λ1611
10
Sebastian Lopez and Sara L. Ellison: Abundances in Multiple DLAs
from Fig. 5 still shows the low S trend in MDLAs. (5)
Instrument: almost all literature abundances come from
HIRES data whereas we have used UVES at a slightly
higher resolution. However, some high α/Fe UVES values
in Table 3 and the low α/Fe HIRES value in the MDLAs
toward Q2359−02 show that a possible bias due to differ-
ent data type is unfounded.
In summary, we have found that MDLAs exhibit signif-
icantly lower [S/Fe], and to a less striking extent, [Si/Fe]
ratios than single DLAs taken from the literature. We have
argued that these low values are not driven by ionization
differences or dust depletion, although an observational
bias may be partially responsible for the clearer abun-
dance distinction in S compared to Si. Having excluded
the main systematic effects, these results therefore imply
a nucleosynthetic origin for differences in abundance ratios
between MDLAs and single DLAs.
5. Discussion
5.1. The Nature of MDLAs
[α/Fe] of MDLAs are
If the typically low values of
truly nucleosynthetic in origin, this may indicate that
these systems either represent a particular galaxy type
or that they share a common evolutionary or environmen-
tal link. The first issue to address is whether the 'mul-
tiplicity' is indeed due to more than one distinct galaxy.
Schaye (2001) has suggested that some DLAs may be as-
sociated with winds, in which case MDLAs may be caused
by outflowing material from the parent DLA. Indeed,
Nulsen, Barcons & Fabian (1998) have shown that large
scale galactic winds can produce column densities equal
to those of DLAs. This could be tested by studying abun-
dances in systems that are effectively transverse multiple
systems, that is DLAs observed in more than one line of
sight separated by many tens of kpc. The closest exam-
ple we have of this in the current dataset is the case of
Q0201+1120 (Ellison et al. 2001a) which is located in a
concentration with at least four other galaxies and indeed
shows low [S/Fe], possible evidence that the abundance
trend is symptomatic of a group environment rather than
simply line of sight velocities within a single object.
However, this evidence remains vague in our context
of MDLAs as these imaging studies deal only with DLAs
without line-of-sight companions. Deep imaging and sub-
sequent spectroscopy of the fields toward CTQ247 and
Q2314-409 as well as control fields around QSOs har-
boring single DLAs is therefore required in order to
pursue a comparison of possibly different environments.
At higher redshifts, finding evidence for cluster envi-
ronments of single DLAs (and even the DLA absorbers
themselves) has proven difficult
(Gawiser et al. 2001;
Prochaska et al. 2002a), so much work is still to be done
in this direction. Similarly, although several DLA imaging
studies have been carried out (e.g., Le Brun et al. 1997),
we lack studies of the clustering properties in the DLA
fields at z < 1.0 and there are currently no known MDLAs
at low redshift.
5.2. Truncated star formation in MDLAs?
As suggested in Paper I, the low α/Fe abundances ob-
served in MDLAs might be due to suppressed SF in the
galaxies harboring the absorbing gas. We now raise the
question as to whether environment may cause these ob-
served abundance anomalies and speculate upon the pos-
sibility that MDLAs arise in 'protogalactic groups'.
There is a clear trend of galaxies in rich environments
out to z <∼ 0.5 to exhibit suppressed SF, not just in dense
cluster cores, but also in clusters out to 2-3 virial radii
(Balogh et al. 1997; Lewis et al. 2002) and perhaps out to
several Mpc (Pimbblet et al 2001). The line diagnostics
of these galaxies indicate that radial trends are consis-
tent with an age sequence in which the last episode of
star formation happened most recently in the outermost
galaxies (Balogh et al. 1999). However, the physical mech-
anisms governing this effect are not clear; violent processes
such as ram pressure stripping (e.g., Couch et al. 2001
and references therein) and passive exhaustion of gas sup-
ply (e.g., Balogh et al. 1999) are the two main possibil-
ities. The recent finding by Lewis et al. (2002) that lo-
cal density is the dominant parameter in suppressing SF
rates indicates that extreme cluster-scale processes such
as ram-stripping by the ICM play a minor role. These
results also point to the possibility that reduction in
SF need not take place within the gravitational bounds
of existing virialized clusters and may occur already in
looser groups before they are accreted onto larger struc-
tures. This ties in with cluster histories at lower redshifts,
because although Dressler et al. (1999) find a significant
post-starburst population in clusters, there is evidence
that this fraction is not greater than among the field pop-
ulation (Balogh et al. 1999).
If the same physical processes are on-going at high
redshifts, then suppressed SF may be evident in galaxies
associated with early proto-clusters, such as those identi-
fied by Steidel et al (1998), even though canonical, viri-
alized clusters have yet to form. Although the abundance
ratios for MDLAs that we have discussed here are distinct
from 'field' DLAs, they have typical metallicities for their
redshift (c.f. Pettini et al 1999) and moderate dust con-
tent. These properties are indications that some chemical
maturity has indeed taken place, and MDLAs are simi-
lar to other DLAs in terms of their general enrichment.
However, the low [α/Fe] is an indication that what sets
MDLAs apart is that they have been evolving quiescently,
without any further major SF episodes, over the last ∼
500 Myrs.
5.3. Other evidence for quiescent star formation
If the low α/Fe ratios observed in MDLAs are to be
explained by quiescent SF, then we should expect to
Sebastian Lopez and Sara L. Ellison: Abundances in Multiple DLAs
11
Dessauges-Zavadsky et al 2002). If these high ratios are
not due to dust depletion (Fe is more depleted in the local
ISM than Mn) or ionization, they show a mild odd-even
effect. Analogously, [Al/Si] = 0.07±0.12 in CTQ247C, the
highest value ever measured, and [Al/Si] = −0.28 ± 0.18
in B2314−409B (Paper I), which is among the highest
values after the z = 2.154 MDLA toward Q2359−02
(Prochaska & Wolfe 2002). Altogether, the odd-even ef-
fect seems to be mild in those MDLAs where we can
measure it. Since models of massive star yields give good
agreement with observations of Mn/Fe in Galactic stars
(Goswami & Prantzos 2000; and references therein) and
also DLAs (Ledoux, Bergeron & Petitjean 2002), the lack
of odd-even effect at low metallicity in MDLAs indicates
again that high-mass stars have not been the dominant
metal pollutant in MDLAs. This may be interpreted as
further evidence that there have not been several epochs
of SF building on the enriched interstellar media from pre-
vious episodes; further support of suppressed SF.
Secondary
6. Summary
Fig. 9. The N/α ratio in MDLAs (circles), 'field' DLAs
(stars) and H ii regions (dots; see Pettini et al. 2002 for
references). The triangles are for the 2 DLAs with com-
panion LBGs.
see its signature in other elemental ratios that also
act as 'cosmic clocks'. The most promising possibil-
ity is the N/α ratio which traces the primary and
secondary contributions from stars of different masses
(Henry, Edmunds and Koeppen 2000). Fig. 9 shows the
N/α ratio in MDLAs, 'field' DLAs and H ii regions (Pettini
et al. 2002). We only used measurements of O or S in the
plot and corrected S values by the solar S/O ratio (1.54
dex). Prochaska et al. (2002b) have suggested a bimodal
distribution of N/α and although this needs to be con-
firmed by more data, it is clear that a number of DLAs
exhibit low ratios. For MDLAs, the ratios are all high,
with no points (N/O) < −1.9, supporting the quiescent
SF scenario. It perhaps indicates more chemical maturity
than average, but also that the role of massive stars in
polluting the ISM of MDLAs may be less important.
The other established trend with metallicity in
Galactic stars is the odd-even effect. Although this
is not technically a chronometer in the same sense
as N and the alpha elements, there is a correlation
with metallicity which may be due to metallicity de-
pendent yields (Goswami & Prantzos 2000). At our dis-
posal we have [Mn/Fe] = −0.12 ± 0.08 (CTQ247A)
and [Mn/Fe] = −0.31 ± 0.03 (CTQ247B), the for-
mer of which is among the highest values in DLAs
of this metallicity (Ledoux, Bergeron & Petitjean 2002;
We have studied a sample of 7 DLAs with line-of-sight
companions, 'multiple' DLAs in our nomenclature, and 2
DLAs arising in transverse groups. We have shown that
the relative abundances in these two sub-samples are un-
usual as compared with single DLAs. In particular, the
[S/Fe] and [Si/Fe] ratios are statistically lower than liter-
ature DLAs, although the effect is more striking in [S/Fe].
We suggest that this difference is due to (1) an obser-
vational bias against measuring low S column densities
because weak S ii are usually lost in the forest; and (2),
to a lesser degree, the larger scatter induced in [Si/Fe]
by photoionization. Besides low α/Fe ratios, MDLAs also
exhibit a mild odd-even effect and relatively high N/α
ratios. We interpret these results as truncated star forma-
tion in MDLAs. If the multiplicity of these DLAs is due
to grouping in physical space, a thesis we cautiously sup-
port, environment may be the cause of the quiescent SF
scenario, just as is observed in more nearby clusters and
groups of galaxies.
Acknowledgements. We are grateful to the anonymous ref-
eree for many qualified comments and suggestions on a
first version of this paper. We also thank Eric Gawiser
for fruitful discussions and comments, Michael Rauch for
allowing us
to use the FORS data of CTQ247, Max
Pettini
for providing us with the template for Fig. 9,
Mirka Dessauges-Zavadsky for communicating abundances
in advance of publication, and Jason Prochaska, whose
database at http://kingpin.ucsd.edu/~hiresdla/ we have
consulted. SL acknowledges support from the Chilean Centro
de Astrof´ısica FONDAP No. 15010003, and from FONDECYT
grant No3 000 001.
References
Ballester, P., Modigliani, A., Boitquin, O., Cristiani, S.,
Hanuschik, R., Kaufer, A., Wolf, S., 2000, ESO Messenger,
101, 31
12
Sebastian Lopez and Sara L. Ellison: Abundances in Multiple DLAs
Balogh, M., Morris, S., Yee, H. K. C., Carlberg, R. G.,
Peroux, C. Storrie-Lombardi, L. J., McMahon, R. G., Irwin,
Ellingson, E., 1999, ApJ, 527,54
M. & Hook, I. M., 2001, AJ, 121, 1799
Balogh, M., Morris, S., Yee, H. K. C., Carlberg, R. G.,
Pettini, M., Ellison, S. L., Bergeron, J. & Petitjean, P., 2002,
Ellingson, E., 1997, ApJL, 488, L75
A&A, 391, 21
Bonifacio, P., Caffau, E., Centurion, M., Molaro, P. & Vladilo,
Pettini, M., Ellison, S. L., Steidel, C. C., Bowen, D. V. 1999,
G., 2001, MNRAS, 325, 767
ApJ, 510, 576
Centurion, M., Bonifacio, P., Molaro, P. & Vladilo, G. 2000,
Pimbblet, K, Smail, I., Edge, A., Couch, W., O' Hely, E.,
ApJ, 536, 540
Chen, Y. Q., Nissen, P. E., Zhao, G. & Asplund, M., 2002,
A&A, 390, 225
Chengalur, J. N. & Kanekar, N., 2000, MNRAS, 318, 303
Couch, W., Balogh, M., Bower, R., Smail, I., Glazebrook, K.,
& Taylor, M., 2001, ApJ., 549, 820
Dessauges-Zavadsky, M., Prochaska, J., & D'Odorico, S., 2002,
A&A, 391, 801
Dessauges-Zavadsky, M., D'Odorico, S., McMahon, R. G.,
Molaro, P., Ledoux, C., Peroux, C. & Storrie-Lombardi,
L. J, 2001, A&A, 370, 426
Dressler, A., Smail, I., Poggianti, B., Butcher, H., Couch, W.,
Ellis, R., Oemler, A., 1999, ApJS, 122, 51
Ellison, S. L., & Lopez, S., 2001, A&A, 380, 117, Paper I
Ellison, S. L., Pettini M., Steidel C. C., Shapley A., 2001a,
ApJ, 549, 770
Ellison, S. L., Yan, L., Hook, I., Pettini, M., Wall, J., Shaver,
P., 2001b, A&A, 379, 393
Evrard, A. E., MacFarland, T. J., Couchman, H. M. P.,
Colberg, J. M., Yoshida, N., White, S. D. M., Jenkins, A.,
Frenk, C. S., Pearce, F. R., Peacock, J. A. & Thomas, P.
A., 2002, ApJ, 573, 7
Ferland, G. J. 1993, University of Kentucky, Physics
Department Internal Report
Gawiser, E., Wolfe, A. M., Prochaska, J. X., Lanzetta, K. M.,
Zabludoff, A., 2001, MNRAS, 327, 588
Prochaska, J X., Gawiser, E., Wolfe, A. M., Quirrenbach, A.,
Lanzetta, K. M., Chen, H.-W, Cooke, J., & Yahata, N.
2002a, AJ, 123, 2206
Prochaska, J. X., Henry, R., O'Meara, J., Tytler, D., Wolfe, A.,
Kirkman, D., Lubin, D., Suzuki, N., 2002b, ApJ, accepted
Prochaska, J. X., Howk, J. C., O'Meara, J. M., Tytler, D.,
Wolfe, A., Kirkman, D., Lubin, D. & Suzuki, N. 2002c,
ApJ, 571, 693
Prochaska, J. X., & Wolfe, A., 2002, ApJ, 566, 68
Prochaska, J. X., Wolfe, A. M., Tytler, D., Burles, S., Cooke,
J., Gawiser, E., Kirkman, D., O'Meara, J. M., & Storrie-
Lombardi, L. ApJS, 2001, 137, 21
Prochaska, J. X., Gawiser, E. & Wolfe, A. M., 2001, ApJ, 552,
99
Prochaska, J.X., & Wolfe, A.M. 1999, ApJS, 121, 369
Prochaska, J.X., & Wolfe, A.M. 1996, ApJ, 470, 403
Quashnock, J. M., vanden Berk, D. E. & York, D. G., 1996,
ApJ 472, 69
Rigopoulou, D., Franceschini, A., Aussel, H., Genzel, R.,
Thatte, H., Cesarsky, C., 2002, ApJ, 580, 789
Savage, B. D. & Sembach, K. R., 1991, ApJ 379, 245
Savage, B. D. & Sembach, K. R., 1996, ARA&A, 34, 279
Schaye, J., 2001, ApJL, 559, L1
Steidel, C.C., Adelberger, K.L., Dickinson, M., Giavalisco, M.,
Yahata, N., & Quirrenbach, A. 2001, ApJ, 562, 628
Pettini, M. & Kellogg, M. 1998, ApJ, 492, 428.
Goswami, A., & Prantzos, N., 2000, A&A, 359, 191
Grevesse, N., & Sauval, A.J. 1998, Space Sci Rev, 85, 161
Haardt, F., & Madau, P. 1996, ApJ, 461, 20
Haines, C., Campusano, L. E. & Clowes, R., 2002,
Steidel, C., Giavalisco, M., Pettini, M., Dickinson, M.,
Adelburger, K., 1996, ApJL, 462, L17
Viegas, S. M., 1995 MNRAS, 276, 268
Vladilo, G., Centuri´on, M., Bonifacio, P., & Howk, J. C., 2001,
astro-ph/0301473
ApJ 557, 1007
Henry, R. B. C., Edmunds, M. G., & Koppen, J. 2000, ApJ,
541, 660
Vladilo, G., 2002, ApJ, 569, 295
Williger, G. M., Campusano, L. E., Clowes, R. G. & Graham,
Holweger, H., 2001, in Solar and Galactic Composition, ed. R.
M. J., 2002, ApJ, 578, 708
Wimmer-Schweingruber, (Berlin: Springer), 23
Hou, J. L., Boissier, S. & Prantzos, N. 2001, A&A, 370, 23
Le Brun, V., Bergeron, J., Boisse, P. & Deharveng, J. M., 1997,
Wolfe, A. M., Turnshek, D. A., Smith, H. E. & Cohen, R. D.
1986, ApJS 61, 249
Zaritsky, D., Smith, R., Frenk, C., White, S. D. M., 1997 ApJ,
478, 39
A&A, 321, 733
Ledoux, C., Bergeron J., & Petitjean, P., 2002, A&A, 385, 802
Levshakov, S. A., Dessauges-Zavadsky, M., D'Odorico, S. &
Molaro, P., 2002, ApJ, 565, 696
Lewis I., Balogh, M., De Propris, R., et al. 2002, MNRAS, 334,
673
Lopez S., Reimers D., Rauch M., Sargent W. L. W. & Smette
A., 1999, ApJ 513, 598
Lopez, S., Maza, J., Masegosa, J., & Marquez, I., 2001, A&A
366, 387
Lopez, S., Reimers, D. D'Odorico, S. & Prochaska, J. X., 2002,
A&A 385, 778
Lu, L., Sargent, W. L. W., Barlow, T.A. 1998, AJ, 115, 55
Lu, L., Sargent, W. L. W., Barlow, T. A., Churchill, C. W.,
Vogt, S. 1996, ApJS 107, 475
Molaro, P., Levshakov, S. A., D'Odorico, S., Bonifacio, P., &
Centurion, M., 2001, ApJ 549, 90
Nulsen, P. E. J., Barcons, X., Fabian, A. C., 1998, MNRAS,
301, 168
|
astro-ph/0307420 | 1 | 0307 | 2003-07-24T11:06:41 | On characterising the variability properties of X-ray light curves from active galaxies | [
"astro-ph"
] | We review some practical aspects of measuring the amplitude of variability in `red noise' light curves typical of those from Active Galactic Nuclei (AGN). The quantities commonly used to estimate the variability amplitude in AGN light curves, such as the fractional rms variability amplitude, F_var, and excess variance, sigma_XS^2, are examined. Their statistical properties, relationship to the power spectrum, and uses for investigating the nature of the variability processes are discussed. We demonstrate that sigma_XS^2 (or similarly F_var) shows large changes from one part of the light curve to the next, even when the variability is produced by a stationary process. This limits the usefulness of these estimators for quantifying differences in variability amplitude between different sources or from epoch to epoch in one source. Some examples of the expected scatter in the variance are tabulated for various typical power spectral shapes, based on Monte Carlo simulations. The excess variance can be useful for comparing the variability amplitudes of light curves in different energy bands from the same observation. Monte Carlo simulations are used to derive a description of the uncertainty in the amplitude expected between different energy bands (due to measurement errors). Finally, these estimators are used to demonstrate some variability properties of the bright Seyfert 1 galaxy Markarian 766. The source is found to show a strong, linear correlation between rms amplitude and flux, and to show significant spectral variability. | astro-ph | astro-ph | Mon. Not. R. Astron. Soc. 000, 1–14 (2003)
Printed 19 November 2011
(MN L ATEX style file v2.2)
On characterising the variability properties of X-ray light curves
from active galaxies
S. Vaughan,1,2 R. Edelson,3 R. S. Warwick2 and P. Uttley4
1 Institute of Astronomy, Madingley Road, Cambridge CB3 0HA
2X-Ray and Observational Astronomy Group, Department of Physics and Astronomy, University of Leicester, Leicester LE1 7RH
3Astronomy Department, University of California, Los Angeles, CA 90095-1562; USA
4Department of Physics and Astronomy, University of Southampton, Southampton SO17 1BJ
Accepted: 24/7/2003; Submitted: 23/7/2003; in original form: 3/3/2003
ABSTRACT
We review some practical aspects of measuring the amplitude of variability in ‘red noise’ light
curves typical of those from Active Galactic Nuclei (AGN). The quantities commonly used to
estimate the variability amplitude in AGN light curves, such as the fractional rms variability
amplitude, Fvar , and excess variance, σ2
XS , are examined. Their statistical properties, relation-
ship to the power spectrum, and uses for investigating the nature of the variability processes
are discussed. We demonstrate that σ2
XS (or similarly Fvar ) shows large changes from one part
of the light curve to the next, even when the variability is produced by a stationary process.
This limits the usefulness of these estimators for quantifying differences in variability ampli-
tude between different sources or from epoch to epoch in one source. Some examples of the
expected scatter in the variance are tabulated for various typical power spectral shapes, based
on Monte Carlo simulations. The excess variance can be useful for comparing the variability
amplitudes of light curves in different energy bands from the same observation. Monte Carlo
simulations are used to derive a description of the uncertainty in the amplitude expected be-
tween different energy bands (due to measurement errors). Finally, these estimators are used
to demonstrate some variability properties of the bright Seyfert 1 galaxy Markarian 766. The
source is found to show a strong, linear correlation between rms amplitude and flux, and to
show significant spectral variability.
Key words: galaxies: active — galaxies: Seyfert — galaxies: individua
galaxies — methods: data analysis
l (Mrk 766) — X-rays:
3
0
0
2
l
u
J
4
2
1
v
0
2
4
7
0
3
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
1 INTRODUCTION
One of the defining characteristics of Active Galactic Nucle i
(AGN) is that their X-ray emission is variable. X-ray light curves
from Seyfert 1 galaxies show unpredictable and seemingly aperi-
odic variability (Lawrence et al. 1987; McHardy 1989). Such ran-
dom variability is often referred to as noise, meaning that it is the
result of a stochastic, as opposed to deterministic, process. In this
context the ‘noise’ is intrinsic to the source and not a result of mea-
surement errors (such as Poisson noise), i.e. the signal itself is the
output of a noise process.
One of the most common tools for examining AGN variability
(and noise processes in general) is the fluctuation Power Spe ctral
Density (PSD) which represents the amount of variability power
(mean of the squared amplitude) as a function of temporal fre-
quency (timescale−1 ). The high frequency PSDs of Seyferts are
usually well-represented by power-laws over a broad range of fre-
quencies (P (f ) ∝ f −α , where P (f ) is the power at frequency f )
with slopes α = 1 − 2 (Green, McHardy & Lehto, 1993; Lawrence
& Papadakis, 1993; Edelson & Nandra, 1999; Uttley, McHardy &
Papadakis, 2002; Vaughan, Fabian & Nandra 2003; Markowitz et
al. 2003). Such a spectrum, with a slope α ∼> 1 is usually called
‘red noise’ (for an introduction to red noise see Press 1978).
If Seyfert 1 light curves are viewed as the product of a stochas-
tic (in this case red noise) process then the speci fic details of each
individual light curve provide little physical insight. Each light
curve is only one realisation of the underlying stochastic process,
i.e. it is one of the ensemble of random light curves that might be
generated by the process. Each new realisation will look different
and these changes are simply statistical fluctuations inher ent in any
stochastic process (as opposed to changes in the nature of the pro-
cess itself). Therefore one should expect two light curves to have
different characteristics (such as mean and variance) even if they
are realisations of the same process. On the other hand, data from
deterministic processes, for example the energy spectrum of a non-
varying source or the light curve of a strictly periodic source (such
as a pulsar), should be repeatable within the limits set by the mea-
surement errors.
It is the average properties of the variability (such as the PSD)
that often provide most insight into the driving process. For exam-
ple, the PSD of any real red noise process cannot continue as a
2
Vaughan et al.
steep power-law indefinitely to longer timescales or the int egrated
variability power would diverge. Therefore the PSDs of AGN vari-
ability must break to a flatter index at low frequencies; the p o-
sition of such a break would represent a characteristic variability
timescale and may yield information about the underlying driving
process. Recent timing studies have indeed found evidence that the
steep power-law PSDs of Seyfert 1s show a flattening, or turno ver,
at low frequencies (Edelson & Nandra 1999; Uttley et al. 2002;
Markowitz et al. 2003).
In many cases however the data are not adequate for PSD anal-
ysis. In these situations the variability is usually described in terms
of the statistical moments (e.g. the sample mean and variance, etc.).
However, due to the stochastic nature of red noise variability there
is a large degree of randomness associated with these quantities. In
practice this means that it is difficult to assign meaningful errors to
the variance. This in turn makes it difficult to quantitative ly com-
pare variances, say from repeated observations of the same source
(and thereby test whether the variability is stationary). Such an
analysis might be desirable; it could in principle reveal changes
in the ‘state’ of the source if its variability properties were found
to evolve with time. This paper discusses this and related problems
that are encountered when examining the variability properties of
AGN. Particular emphasis is placed on the mathematical properties
and implications of the inherent randomness in the variability. The
mathematical details are well understood from the general theory
of stochastic processes (e.g. Priestley 1981 for spectral analysis)
but some of the practical consequences for AGN observations have
not been discussed in detail. On the basis of simulated data some
recipes are developed that may serve as a useful guide for observers
wishing make quantitative use of their variability analysis without
the recourse to e.g. extensive Monte Carlo simulations.
The paper is organised as follows. Section 2 defines the esti-
mators to be discussed (namely the periodogram and the variance).
Simulated data are used to illustrate various aspects of these es-
timators; section 3 describes methods for producing arti fic ial red
noise time series. Section 4 discusses the stationarity of time se-
ries. Sections 5 and 6 discuss two sources of uncertainty associated
with measuring variability amplitudes, the first due to the s tochas-
tic nature of the variability and the second due to flux measur ement
errors. Section 7 gives an example using a real XMM-Newton ob-
servation of Mrk 766. Finally, a brief discussion of these results is
given in section 8 and the conclusions are summarised in section 9.
2 ESTIMATING THE POWER SPECTRAL DENSITY
The PSD defines the amount of variability ‘power’ as a functio n of
temporal frequency. It is estimated by calculating the periodogram1
(Priestley 1981; Bloomfield 2000).
For an evenly sampled light curve (with a sampling period
∆T ) the periodogram is the modulus-squared of the Discrete
Fourier Transform (DFT) of the data (Press et al. 1996). For a light
curve comprising a series of fluxes xi measured at discrete times
ti (i = 1, 2, . . . , N ):
xi e2π ifj ti (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)
DF T (fj )2 = (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)
N
Xi=1
1 Following Priestley (1981) the term “periodogram ” is used f or the dis-
crete function P (fj ), which is an estimator of the continuous PSD P (f ).
The periodogram is therefore specific to each realisation of
the process,
whereas the PSD is representative of the true, underlying process.
=
2
,
(1)
xi cos(2πfj ti ))2
xi sin(2πfj ti ))2
( N
+ ( N
Xi=1
Xi=1
at N/2 evenly spaced frequencies fj = j/N ∆T (where j =
1, 2, . . . , N/2), fN/2 = 1/2∆T is the Nyquist frequency, fNyq .
Note that it is customary to subtract the mean flux from the
light curve before calculating the DFT. This eliminates the zero-
frequency power. The periodogram, P (fj ), is then calculated by
choosing an appropriate normalisation A (see Appendix A for more
on periodogram normalisations). For example
2∆T
P (fj ) = ADF T (fj )2 =
N DF T (fj )2 .
If the time series is a photon counting signal such as normally
encountered in X-ray astronomy, and is binned into intervals of
∆T , the effect of Poisson noise is to add an approximately con-
stant amount of power to the periodogram at all frequencies. With
the above normalisation this constant Poisson noise level is 2 ¯x (as-
suming the light curve is not background subtracted).
(2)
2.1 Statistical properties of the periodogram
(3)
The periodogram of a noise process, if measured from a single time
series, shows a great deal of scatter around the underlying PSD. In
particular, the periodogram at a given frequency [P (f )] is scat-
tered around the PSD [P (f )] following a χ2 distribution with two
degrees of freedom (van der Klis 1989):
P (f ) = P (f )χ2
2/2,
where χ2
2 is a random variable distributed as χ2 with two degrees of
freedom, i.e. an exponential distribution with a mean and variance
of two and four, respectively. The periodogram is distributed in this
way because the real and imaginary parts of the DFT are normally
distributed for a stochastic process2 (section 6.2 of Priestley, 1981;
Jenkins & Watts 1968). The expectation value of the periodogram
is equal to the PSD but its standard deviation is 100 per cent, lead-
ing to the larger scatter in the periodogram (see Fig. 1). See Leahy
et al. (1983), van der Klis (1989), Papadakis & Lawrence (1993),
Timmer & K onig (1995) and Stella et al. (1997) for further discus-
sion of this point.
When applied to real data the periodogram is an inconsistent
estimator of the PSD, meaning that the scatter in the periodogram
does not decrease as the number of data points in the light curve
increases (Jenkins & Watts 1968). In order to reduce this scatter
the periodogram must be smoothed (averaged) in some fashion. As
the number of data points per bin increases (either by binning over
frequencies or averaging over many data segments) the scatter in
the binned periodogram decreases, i.e. the averaged periodogram
is a consistent estimator of the PSD (see Papadakis & Lawrence
1993 and van der Klis 1997 for more on binned periodogram esti-
mates). A further point is that periodograms measured from fi nite
data tend to be biased by windowing effects which further compli-
cate their interpretation (van der Klis 1989; Papadakis & Lawrence
1993; Uttley et al. 2002 and see below).
2 The DFT at the Nyquist frequency is always real when N is even so the
periodogram at this frequency is distributed as χ2
1 , i.e. with one degree of
freedom.
Properties of X-ray light curves from AGN
3
3 SIMULATING RED NOISE LIGHT CURVES
3.1 Algorithms
In order to elucidate the properties of the variance of red noise data,
random light curves were generated from power-law PSDs similar
to those of AGN. Fig. 1 shows two arti ficial time series and the ir
periodograms. It is worth reiterating that the large scatter in the
periodograms is an intrinsic property of stochastic processes – it
does not depend on the number of data points and is not related to
Poisson noise in the data.
These arti ficial time series were produced using the algorit hm
of Timmer & K onig (1995). This generates random time serieswith
arbitrary broad-band PSD, correctly accounting for the intrinsic
scatter in the powers (i.e. equation 3). Other methods of generating
random light curves include the related ‘summing of sines’ method
(Done et al. 1992). Note that it is not correct to randomise only the
phases of the component sine functions, their amplitudes must also
be randomised. Otherwise this method does not account for this
intrinsic scatter in the powers. Shot-noise models can produce red
noise time series with certain PSD shapes (see Lehto 1989). There
also exist various mathematical tricks for producing data with spe-
ci fic power-law PSD slopes. Data with a α = 1 PSD (often called
‘ flicker noise’) can be generated using the half-integral me thod out-
lined in Press (1978), while α = 2 (‘random walk’) data can be
generated using a first-order autoregressive process (AR[1 ]), essen-
tially a running sum of Gaussian deviates (see Deeming 1970 and
Scargle 1981 for more on such methods). The method of Timmer
& K onig (1995) is used below as this can generate time series from
an arbitrary PSD and is computationally efficient.
3.2 Simulating ‘realistic’ data
Some caution should be applied when using these routines to pro-
duce arti ficial time series. As mentioned briefly in the previ
ous sec-
tion, periodograms measured from real data tend to be biased by
windowing effects. For uninterrupted but finite observatio ns data
of red noise processes the most important of these effects is ‘red
noise leak’ – the transfer of power from low to high frequenci es
by the lobes of the window function (see e.g. Deeter & Boynton
1982; van der Klis 1997). If there is signi ficant power at freq uen-
cies below the lowest frequency probed by the periodogram (i.e. on
timescales longer than the length of the observation) this can give
rise to slow rising or falling trends across the light curve. These
trends contribute to the variance of the light curve. Thus variability
power ‘leaks’ into the data from frequencies below the observed
frequency band-pass. The degree to which this occurs, and the re-
sultant bias on the measured periodogram, depend on the shape of
the underlying PSD and the length of the observation (Papadakis &
Lawrence 1995; Uttley et al. 2002). For flat PSD slopes ( α < 1.5)
the amount of leakage from low frequencies is usually negligible.
Since AGN light curves usually contain signi ficant power on
timescales longer than those probed (see section 4.1) the effects
of red noise leak must be included in simulations of AGN light
curves. This can be achieved by using the Timmer & K onig (1995)
algorithm to generate a long light curve from a PSD that extends to
very low frequencies and then using a light curve segment of the re-
quired length. Data simulated in such a fashion will include power
on timescales much longer than covered in the short segment. The
effects of measurement errors (e.g. Poisson noise) can be included
in the simulations using standard techniques (e.g. Press et al. 1996).
Figure 1. Simulated time series (left) and their periodograms (right). The
upper panel shows a ‘flicker noise’ time series which has a f −1 PSD. The
lower panel shows a ‘random walk’ time series with a f −2 PSD. Note the
large scatter in the periodogram (dots) around the underlying PSD (solid
line). It is clear that the time series with the steeper PSD shows more power
in long-term variability while the time series with the flatt er PSD shows
relatively more power in short term variability (flickering ). The two series
were generated using the same random number sequence.
2.2 Integrated power
The integral of the PSD between two frequencies (f1 and f2 ) yields
the contribution to the expectation value of the (‘true’) variance
due to variations between the corresponding timescales (1/f1 and
1/f2 ). This result follows from Parseval’s theorem (see e.g. van der
Klis 1989)
hS 2 i = Z f2
f1
Correspondingly, for a discrete time series the integrated peri-
odogram yields the observed variance for that particular realisation
P (f )df .
(4)
S 2 =
P (fj )∆f ,
N/2
Xj=1
where ∆f is the frequency resolution of the DFT (∆f =
1/N ∆T ). The total variance of a real light curve is equal to its
periodogram integrated over the frequency range f1 = 1/N ∆T to
fNyq = 1/2∆T .
The sample variance (which will differ from observation to
observation) is given by:
(5)
S 2 =
N
1
Xi=1
(xi − ¯x)2 ,
N − 1
where ¯x is the arithmetic mean of xi . In the limit of large N these
two variance estimates are identical. The normalised variance3 is
simply S 2 / ¯x2 .
(6)
3 In AGN studies normalised quantities are often used in preference to ab-
solute quantities as they are independent of the flux of a spec ific source.
This means that, in principle, normalised amplitudes can be used to com-
pare sources with different fluxes.
4
Vaughan et al.
4 STATIONARITY
A stationary process is one for which the statistical properties (such
as mean, variance, etc.) do not depend on time. Fig. 2 shows an ar-
ti ficial red noise time series together with its mean and vari ance
measured every 20 data points. The simulation was produced from
a single, well-defined PSD (which did not vary). The process i s
therefore stationary. It would have been reasonable to expect the
resulting time series (a realisation of the process) to appear sta-
tionary. This is not the case however; both the mean and variance
change with time (panels 2 and 3). This has nothing whatsoever to
do with measurement errors - the simulation has zero errors. This
simulation demonstrates that, when dealing with red noise, fluctu-
ations in variance are not sufficient to claim the variabilit y process
is non-stationary.
As the purpose of time series analysis is to gain insight into
the process, not the details of any speci fic realisation, a mo re robust
approach is needed to determine whether these data were produced
by a time-stationary process or a non-stationary process. It would
be more insightful to consider whether the expectation values of
the characteristics (such as the variance) are time-variable. The ex-
pectation values should be representative of the properties of the
underlying process, not just any one realisation. See section 1.3 of
Bendat & Piersol (1986) for a discussion of this point.
4.1 Weak non-stationarity
For a process with a steep red noise PSD (α > 1), the integrated
periodogram will diverge as f → 0. This means that (following
equation 4) the variance of a red noise time series with a steep PSD
will diverge with time. In this case there is no well-defined m ean;
Press & Rybicki (1997) describe this form of variability as ‘weakly
non-stationary.’ For time series with power spectra flatter
than this
the variance converges as f → 0, thus for a white noise process
with a flat PSD ( α = 0), the variance will converge as the obser-
vation length increases, and there will be a well-defined mea n on
long timescales.
Of course, for any real process the PSD must eventually flat-
ten such that the power does not diverge (i.e. α < 1 on sufficiently
long timescales). Thus, weak non-stationarity is entirely due to ob-
servations sampling only the steep part of the PSD of a source.
But in AGN this flattening occurs on timescales much longer th an
those probed by typical observations. For instance, XMM-Newton
observations of AGN typically last for ∼ few × 104 s whereas in
many objects the PSD is steep until > 105 s and in some cases
probably much longer (Edelson & Nandra 1999; Uttley et al. 2002;
Markowitz et al. 2003). Therefore on the timescales relevant for
most X-ray observations, AGN light curves should be considered
weakly non-stationary.
4.2 Stochasticity
Fluctuations in the statistical moments (such as mean and variance)
of a light curve are intrinsic to red noise processes. Therefore, even
in the absence of measurement errors (e.g. no Poisson noise) the
means and variances of two light curves produced by exactly the
same process can be signi ficantly different. This can be seen in
Fig. 2 (panels 2 and 3), where each 20 point segment of the light
curve shows a different mean and variance. These random fluct u-
ations in variance are however governed by the normal statistical
rules of noise processes and can thus be understood in a statistical
sense.
Figure 2. Panel 1: Simulated red noise time series (with a f −2 PSD) with
N = 2800 points. Panel 2 and 3: mean and variance measured from seg-
ments of 20 points (calculated using equation 6). The variances follow a
distribution of the form shown in the bottom panel of Fig. 3. (Note the vari-
ance is plotted on a logarithmic scale.) Panel 4: averaged variance measured
by binning the individual variances into groups of 20 consecutive estimates.
The errors are the standard error on the mean (equation 4.14 of Bevington
& Robinson 1992). These averaged variances are consistent with staying
constant. In other words, although the instantaneous value of the variance
fluctuates, its expectation value is consistent with consta nt (i.e. a station-
ary process). Panel 5: fractional rms amplitude (pS 2 / ¯x2 ) measured from
segments of 20 points. Panel 6: averaged fractional rms amplitude measured
by binning the individual amplitudes into groups of 20. The fractional am-
plitude is anti-correlated with the light curve because hS 2 i is constant but
Fvar is normalised by the light curve flux.
Any given series is only one realisation of the processes and
its periodogram will show the scatter predicted by equation 3. The
integrated periodogram (which gives the variance; equation 5) will
therefore be randomly scattered around the true value for the PSD
of the process. The variance in a speci fic time series is given by
S 2 =
1
N ∆T
P (fi )χ2
2 /2,
N/2
Xi=1
i.e. the variance of a given realisation is a sum of χ2
2 distributions
weighted by the PSD4 . (This assumes biases such as red noise
leak are not signi ficant. If this is not true then these biases will
(7)
4 As noted earlier, the periodogram at the Nyquist frequency is actually
Properties of X-ray light curves from AGN
5
These points are illustrated by Fig. 3, which shows the distri-
butions of variances in random time series with three different PSD
slopes (α = 0, 1, 2). This plot was produced by generating 50,000
random time series (each 100 points long) for each PSD slope and
measuring the variance of each one. The scatter in the variance is
entirely due to random fluctuations between different reali sations
because the PSD normalisation was kept fixed and no instrumen tal
noise was added. The shape of the distribution of variances can be
seen to depend on the PSD slope.
Consider a white noise process (α = 0, i.e. P (f ) = const).
The periodogram of each realisation is randomly scattered around
its flat PSD. Following equation 7 the variance is simply the s um
2 -distributed powers in the periodogram, and these
of the N/2 χ2
2 distri-
are evenly weighted (PSD is constant). The sum of N/2 χ2
butions follows a χ2
N distribution. This tends to a normal distribu-
tion as N increases. Thus the variance of a white noise process is
approximately normally distributed (as can be seen in Fig. 3) and
converges as N increases. The fractional standard deviation of χ2
N
is given by p2/N , so for the 100 point light curves used the (1σ )
fractional width of the variance distribution is ≈ 14.1 per cent, in
agreement with the simulations (Fig. 3, top panel).
For time series with steeper PSDs the lower frequency peri-
odogram points contribute more strongly to the sum than the higher
frequency points. The variance of such a time series is therefore
dominated by a few low frequency powers5 and thus resembles a
ν distribution with low ‘effective degrees of freedom.’ The distri-
χ2
bution of variances in red noise data is dependent on the underlying
PSD and is, in general, non-Gaussian (Fig. 3). The fractional stan-
ν is p2/ν , which tends to unity as the PSD
dard deviation of χ2
gets steeper (i.e. as effective ν → 2). Thus the largest fluctuations
in variance (up to a limit of ∼ 100 per cent rms) are expected to
result from very steep PSD slopes.
5 INTRINSIC SCATTER IN VARIANCE
As discussed above, when examining AGN light curves one should
expect random changes in the mean and variance with time (be-
tween segments of a long observation or between observations
taken at different epochs). This is true even if the measurement
errors are zero and does not depend on the number of data points
used (due to the weak non-stationarity). However, it is also possi-
ble that the underlying process responsible for the variability itself
changes with time (e.g. the PSD changes), in which case the vari-
ability is non-stationary in a more meaningful sense – ‘stro ngly
non-stationary.’ Such changes in the variability process could pro-
vide insight into the changing physical conditions in the nuclear
regions. On the other hand the random changes expected for a red
noise process yield no such physical insight. The question thus
arises: how does one tell, from a set of time series of the same
source, whether they were produced by a strongly non-stationary
process? In other words, is it possible to differentiate between dif-
ferences in variance caused by real changes in the variability pro-
cess (physical changes in the system), and random fluctuatio ns ex-
pected from red noise (random nature of the process)?
If the process responsible for the variability observed in a
Figure 3. Distribution of variances in time series with three different PSD
shapes: f 0 (top), f −1 (middle) and f −2 (bottom). Each distribution is de-
rived from 50,000 realisations. As the PSD gets steeper the distribution of
variances becomes less Gaussian and more like a χ2 distribution with a low
effective degrees of freedom.
further distort the distribution of variances.) It is the expectation
value of the variance that is representative of the integrated power
in the PSD, and thus the average amplitude of the variability pro-
cess (eqn. 4). Thus, while the expectation value of the variance is
equal to the integrated PSD, each realisation (time series) of the
same process will show a different variance even if the parent vari-
ability process is stationary. This is particularly important for steep
PSD time series (i.e. weakly non-stationary data) since the vari-
ance does not converge as more data are collected. Only if the time
series spans timescales on which the integrated power converges
at low frequencies (i.e. α < 1) will the variance converge as the
length of time series increases.
distributed as χ2
1 for even N . However, for large N this will make a negli-
gible difference to the sum (cf. equation 2.9 of van der Klis, 1989).
5 The windowing effects mentioned above mean that fluctuation s at powers
above and below the frequency range of the periodogram may also affect the
variance.
6
Vaughan et al.
given source is stationary then its PSD is constant in time. The ex-
pectation value of the absolute (un-normalised) variance will there-
fore be the same from epoch to epoch, but the individual variance
estimates will fluctuate as discussed in section 4.2. This ma kes it
difficult to judge, from just the variances of two light curve s taken
at two epochs, whether they were produced by a stationary process.
Given sufficient data it is, however, possible to test whethe r the ex-
pectation values of the variance (estimated from an ensemble of
light curves) at two epochs are consistent with a stationary process.
5.1 Comparing PSDs
The methods most frequently employed involve comparing the
PSDs (estimated from the binned periodogram) at different epochs.
If the PSDs show signi ficant differences (at a given confidenc
e
level) the variability process can be said to be strongly non-
stationary. As an example of this, the PSDs of X-ray binaries
evolve with time, and the way in which the variability properties
evolve provides a great deal of information on the detailed work-
ings of these systems (see e.g. Belloni & Hasinger 1990; Uttley &
McHardy 2001; Belloni, Psaltis & van der Klis 2002; Pottschmidt
et al. 2002).
Papadakis & Lawrence (1995) suggested a method suitable for
testing whether large AGN datasets display evidence for strongly
non-stationary variability. Again this method works by comparing
the PSDs from different time intervals, in this case by determining
whether the differences between two periodograms are consistent
with the scatter expected based on equation 3. In particular, they de-
fine a statistic s based on the ratio of two normalised periodograms.
If s deviates signi ficantly from its expected value for stationa ry data
(if hsi = 0) then the hypothesis that the data are stationary can be
rejected (at some confidence level).
5.2 Comparing variances
A different approach is compare variances Si derived from M ob-
servations of the same source (either segments of one long obser-
vation or separate short observations). In order to test whether the
Si differ signi ficantly (i.e. more than expected for a red noise pro-
cess) a measure of the expected scatter is required. This error could
be obtained directly from the data (by measuring the standard de-
viation of multiple estimates) or through simulations (based on an
assumed PSD shape)6 .
5.2.1 Empirical error on variance
An empirical estimate of the mean and standard deviation of the
variance can be made given M non-overlapping data segments. The
M segments each yield an estimate of the variance7 , Si . Each of
these is an independent variable of (in general) unknown but identi-
cal distribution (unless the process is strongly non-stationary). The
central limit theorem dictates that the sum of these will become
6 It is assumed that the data segments being compared have identical sam-
pling (same bin size and observation length). Every effort should be made
to ensure this is the case, e.g. by clipping the segments to the same length.
The variances will then be calculated over the same range of timescale (fre-
quencies). As the variance can increase rapidly with timescale in red noise
data this is most important for steep PSD data such as AGN light curves.
7 Ideally each segment should contain at least N ∼> 20 data points in order
to yield a meaningful variance.
Figure 4. The average rms amplitude (σ = √S 2 ) as a function of flux for
the simulated light curve shown in Fig. 2. The individual rms estimates were
sorted by flux and binned to M = 20 estimates per bin. Errors correspond
to the error on the mean value. The amplitude is constant with flux.
normally distributed as M increases. Therefore by averaging the
M variance estimates it is possible to produce an averaged vari-
ance (hS 2 i) and assign an error bar in the usual fashion (e.g. equa-
tion 4.14 of Bevington & Robinson 1992). This gives a robust esti-
mate of the variance and the standard deviation of the M variances
around the mean gives an estimate of the uncertainty on the mean
variance.
If several sets of data segments are acquired it is therefore pos-
sible to compare the mean variance of each set statistically (since
each has an associated uncertainty). For example, with two long
XMM-Newton observations of the same source, taken a year apart,
one could measure the variance for each observation (by breaking
each into short segments and taking the mean variance of the seg-
ments). Thus it would be possible to test whether the variability
was stationary between the two observations. This method of esti-
mating the mean and standard deviation of the variance requires a
large amount of data. Of order N × M = 20 × 20 = 400 data
points are needed to produce a single well-determined estimate of
the mean variance and its error. A typical XMM-Newton observa-
tion of a bright Seyfert 1 galaxy (∼ 40 ks duration) is only likely
to yield enough data for one estimate of the mean variance. Thus
this method is suitable for testing whether the mean variance has
changed from observation to observation.
Fig. 2 (panel 4) demonstrates this empirically derived mean
variance and its error bar on a long, simulated time series. These
data were produced by calculating the variances Si in bins of
N = 20 data points (panel 3) and then averaging M = 20 vari-
ances to produce a mean variance with error bar (panel 4). These
averaged variances are consistent with constant, as expected; fit-
ting these data with a constant gave χ2
ν = 0.84. Figure 4 shows
the rms amplitude is constant with flux. These tests indicate
that
the integrated PSD is consistent with being constant with time; the
variance does not change signi ficantly from epoch to epoch (o r as
function of flux), as expected for a stationary process.
5.2.2 Estimating the error on the variance through simulations
The advantage of the above method is that it requires no assump-
tion about the shape of the PSD, The drawback is that it requires a
substantial amount of data to produce a single, robust variance esti-
mate. An alternative approach is to estimate the standard deviation
of the variances Si based on simulations.
Given an assumed shape for the PSD it is possible to calcu-
late the distribution of variances expected for a stationary process
(see section 4.2). Some example distributions are shown in Fig. 3,
which clearly demonstrates how the distribution depends on the
slope of the PSD. The distribution becomes more normal at flat -
ter slopes and more asymmetric at steep slopes. For a given PSD
shape these distributions are well defined (by eqn. 7) and can be
computed through Monte Carlo simulations. This makes it possible
to estimate limits within which one would expect the variance to be
distributed if the process is stationary.
The two primary factors that affect the distribution of variance
are the PSD shape and the sampling of the light curve (the length of
the data segments in the case of contiguously binned light curves).
Table 1 gives the expected confidence limits for four differe nt PSD
shapes and five different lengths for the data segments. Thes e val-
ues were computed by simulating one very long light curve with
the assumed PSD shape and breaking it into 1000 separate seg-
ments (of speci fied length). The variance within each segmen t was
measured and the distribution of the 1000 variances was calculated.
The 90 per cent confidence interval was calculated by finding the
5th and 95th percentiles of the variance distribution (in general
these upper and lower bounds will differ because the distribution
is asymmetric). The numbers given in the table are the boundaries
of the 90 and 99 per cent confidence regions estimated by aver-
aging the results from 50 runs. The limits are given in terms of
±∆ log(S 2 ) because they are multiplicative. That is, from a partic-
ular realisation the variance is expected to be scattered within some
factor of the true variance (for which the absolute normalisation
is irrelevant). The factors are tabulated in terms of their logarithms
(since multiplicative factors in linear-space become additive offsets
in log-space).
The PSD used for the simulations was chosen to match that ex-
pected for AGN, i.e. a steep power-law at high frequencies (with a
slope of α = 1.0, 1.5, 2.0, 2.5) breaking to a flatter slope ( α = 1.0)
at low frequencies. The frequency of the break was fixed to be
10−3 , in other words the break timescale was 1000 times the bin
size. (The absolute size of the time bins is arbitrary in the simula-
tions. When comparing the simulated results to real data sampled
with e.g. 25 s time resolution, the break timescale in the simulated
PSD is thus 25 ks.)
The numbers given in the table provide an approximate pre-
scription for the expected scatter in the variance of a stationary
process with a red noise PSD similar to that of AGN. The simu-
lated light curve shown in Fig. 2 was used to demonstrate the use
of this table. In this case the PSD is know to have a slope α = 2,
and the variances (shown in panel 3 of Fig. 2) were calculated ev-
ery 20 points. Therefore, the 90 interval for the expected variance
is given by log(S 2 )+0.45
−0.71 . Taking the mean variance as the expec-
tation value for S 2 , this translates to S 2 = 59.9 (11.7 − 168.8).
The interval boundaries were calculated by converting the logarith-
mic value into a linear factor and multiplying by the sample mean
(assumed to represent the true variance). This interval is shown on
Fig. 5 by the dotted lines. The corresponding 99 per cent confidence
interval is also marked.
As expected the individual variances fall within the expected
region. However, the 90 per cent region spans an order of magni-
tude in variance. Thus even order of magnitude differences in vari-
ance between short sections of a light curve are to be expected and
do not necessarily indicate that the underlying process is not sta-
tionary. Subtle changes in the PSD will thus be difficult to de tect
Properties of X-ray light curves from AGN
7
Table 1. Expected scatter in variance estimates. The 90 and 99 per cent
intervals are presented in terms of ±∆ log(S 2 ). (The 99 per cent interval
is given in bold.) The boundaries were calculated from Monte Carlo simu-
lations of light curves. The PSD was chosen to be a broken power-law with
a slope of α = 1 below the break (at a frequency 10−3 ) and a slope above
the break of α = 1.0, 1.5, 2.0, 2.5. The simulated data segments were cho-
sen to be 10, 20, 50, 100, 1000 points long (the timescale of the break in
the PSD being at 1000, in arbitrary units).
α
1.0
1.5
2.0
2.5
10
-0.84
-0.50
+0.33
+0.53
-0.96
-0.61
+0.39
+0.65
-1.16
-0.78
+0.46
+0.75
-1.49
-1.03
+0.52
+0.83
Number of data points
20
50
100
-0.58
-0.36
+0.28
+0.46
-0.75
-0.50
+0.36
+0.61
-1.01
-0.71
+0.45
+0.73
-1.37
-0.98
+0.52
+0.83
-0.40
-0.26
+0.23
+0.39
-0.62
-0.43
+0.34
+0.58
-0.93
-0.67
+0.44
+0.72
-1.31
-0.95
+0.51
+0.82
-0.32
-0.22
+0.20
+0.35
-0.57
-0.40
+0.33
+0.57
-0.90
-0.66
+0.43
+0.72
-1.28
-0.92
+0.50
+0.80
1000
-0.19
-0.13
+0.15
+0.27
-0.45
-0.32
+0.28
+0.49
-0.72
-0.50
+0.36
+0.59
-0.92
-0.63
+0.40
+0.66
Figure 5. Variance of the simulated data shown in Fig. 2 (panel 3) with the
90 (dotted line) and 99 per cent (dashed line) confidence intervals marked
(as calculated in section 5.2.2). Clearly the variances fall within these limits,
as expected for a stationary process. The solid line marks the mean variance.
by examining the raw variances as the intrinsic scatter is so large.
Such changes could be revealed by comparing averaged variances
or comparing the PSDs as described above.
6 EFFECT OF MEASUREMENT ERRORS
6.1 Excess variance and Fvar
The datasets considered thus far have been ideal, in the sense that
they are free from flux uncertainties. In real life, however,
a light
8
Vaughan et al.
(8)
(9)
(10)
.
1
N
σ 2
err =
σ 2
err,i .
curve xi will have finite uncertainties σerr,i due to measurement er-
rors (such as Poisson noise in the case of an X-ray photon counting
signal). These uncertainties on the individual flux measure ments
will contribute an additional variance. This leads to the use of the
‘excess variance’ (Nandra et al. 1997; Edelson et al. 2002) as an
estimator of the intrinsic source variance. This is the variance after
subtracting the contribution expected from measurement errors
XS = S 2 − σ 2
σ 2
err ,
err is the mean square error
where σ 2
N
Xi=1
The normalised excess variance is given by σ 2
XS/ ¯x2 and
NXS = σ 2
the fractional root mean square (rms) variability amplitude (Fvar ;
Edelson, Pike & Krolik 1990; Rodriguez-Pascual et al. 1997) is the
square root of this, i.e.
Fvar = r S 2 − σ 2
err
¯x2
NXS al-
The statistic Fvar is often chosen in preference to σ 2
though the two convey exactly the same information. Fvar is a
linear statistics and can therefore give the rms variability ampli-
tude in per centage terms. The choice of whether to quote Fvar
or σ 2
NXS is usually purely one of presentation. It is worth not-
ing that the Monte Carlo results given in section 5.2.2, to esti-
mate the expected scatter on the variance, can also be applied
to its square root. The expected boundaries of the confidence
re-
gion of the logarithm of the rms is approximately half those of
the variance. Speci fically, ∆ log(σ ) ≈ ∆ log(S 2 )/2 and similarly
NXS )/2.
∆ log(Fvar ) ≈ ∆ log(σ 2
6.2 Spectral variability
An X-ray light curve of an AGN can be split into different energy
bands. The light curves in each band will be strictly simultaneous
and can be used to test whether the X-ray variability is a function
of energy. For example, one might examine the ratio of a soft band
light curve to a hard band light curve. The statistical signi ficance
of any variations in the ratio can be quanti fied by propagatin g the
measurement errors and applying an appropriate test, such as the
χ2 test of the constant ratio hypothesis8 . If the ratio shows varia-
tions greater than those expected from the errors then the two light
curves are intrinsically different and the source does indeed show
spectral variability. Such changes in the energy spectrum with time
can in principle provide valuable clues to the nature of the X-ray
source. This test does not provide any quantitative description of
the spectral variability.
Another tool for investigating spectral variability is the rms
spectrum, i.e. the rms variability amplitude (or Fvar ) as a func-
tion of energy. See e.g. Inoue & Matsumoto (2001), Edelson et al.
(2002), Fabian et al. (2002) and Schurch & Warwick (2002) for
some examples of rms spectra from AGN. However, when exam-
ining rms spectra it is often not clear whether changes in the am-
plitude with energy reflect real energy-dependence of the in trinsic
variability amplitude or are caused by random errors in the light
s (e.g.
curves. The finite measurement errors on the individual fluxe
8 This does, of course, assume the light curves have been binned suffi-
ciently for the error bars to be approximately Gaussian.
due to Poisson noise) will introduce some uncertainty in the esti-
mated rms amplitudes. An estimate of this uncertainty would help
answer the question posed above, namely whether features in rms
spectra are the result of random errors in the data or represent spec-
tral variations intrinsic to the source.
The problem of how to assess the uncertainty on the excess
variance (or Fvar ) is a long-standing one (e.g. Nandra et al. 1997;
Turner et al. 1999; Edelson et al. 2002). The standard error formu-
lae presented in the literature (e.g. Turner et al. 1999; Edelson et al.
2002) are formally valid in the case of un-correlated Gaussian pro-
cesses. Typically AGN light curves at different X-ray energies are
strongly correlated and are not Gaussian. However, when searching
for subtle differences in amplitude between simultaneous and cor-
related light curves it may be more useful to have an indication of
the uncertainty resulting from the finite flux errors.
6.2.1 Uncertainty on excess variance due to measurement errors
. (11)
A Monte Carlo approach was used to develop a prescription of the
effect of measurement errors on estimates of Fvar (and σ 2
NXS ). A
short red noise light curve was generated. Poisson noise was added
(i.e. the individual flux measurements were randomised foll owing
the Poisson distribution) and the excess variance was recorded. The
fluxes of the original light curve were randomised again and t he ex-
cess variance recorded, this was repeated many times. The distribu-
tion of excess variances was then used to determine the uncertainty
in the variance estimate caused by Poisson noise. Full details of the
procedure are given in Appendix B
For these simulations it was found that the error on σ 2
NXS de-
creases as the S/N in the light curve is increased according to:
NXS) = vuut(r 2
¯x )2
¯x2 )2
+ (r σ 2
σ 2
2Fvar
err
err
err(σ 2
N ·
N ·
See appendix B for details of this equation and its equivalent in
terms of Fvar .
As this only accounts for the effect of flux measurement er-
rors (such as Poisson noise) in a given light curve it can be used
to test whether two simultaneously observed light curves of the
same source, but in different bands, show consistent amplitudes.
A demonstration of this using real data is given in the following
section. This uncertainty does not account for the random scatter
intrinsic to the red noise process, therefore the absolute value of
the rms spectrum will change between realisations (i.e. from epoch
to epoch). But if a source shows achromatic variability then the
values of Fvar calculated in each energy band (at a given epoch)
should match to within the limits set by the Poisson noise (i.e. the
fractional rms spectrum should be constant to within the uncertain-
ties given by the above equation). Differences in Fvar signi ficantly
larger than these would indicate that the source variability ampli-
tude is a function of energy. This would then mean the PSD ampli-
tude/shape is different in different energy bands, or there are multi-
ple spectral components that vary independently.
The above uncertainty estimates can be used to test the hy-
pothesis that the source variability is achromatic. If signi ficant dif-
ferences between energy bands are detected (as in the case of Mrk
766 presented below) then these errors should not be used to fi
t
the rms spectrum. The assumption that the differences are due only
to measurement errors is no longer the case. In such situations the
light curves in adjacent energy bands are likely to be partially cor-
related and so χ2 - fitting of the rms spectrum is not appropriate. The
Properties of X-ray light curves from AGN
9
Figure 7. Excess variance of the Mrk 766 data shown in Fig. 6 (panel 3)
with the 90 (dotted line) and 99 per cent (dashed line) confidence inter-
vals marked (as calculated in section 5.2.2). The variances fall within these
limits, as expected for a stationary process. The solid line marks the mean
variance.
7.1 Observation details
Mrk 766 was observed by XMM-Newton (Jansen et al. 2001) over
the period 2001 May 20 – 2001 May 21 (rev. 265). The present
analysis is restricted to the data from the pn European Photon Imag-
ing Camera (EPIC), which was operated in small window mode.
Extraction of science products from the Observation Data Files
(ODFs) followed standard procedures using the XMM-Newton Sci-
ence Analysis System (SAS) v5.3.3. Source data were extracted
from a circular region of radius 35 arcsec from the processed image
and only events corresponding to patterns 0–4 (single and do uble
pixel events) were used. Background events were extracted from
regions in the small window least effected by source photons, these
showed that the background rate increased dramatically during the
final ∼ 1.5 × 103 s of the observation. This section of the data
was excluded, leaving 1.05 × 105 s of uninterrupted data. The light
curves were corrected for telemetry drop outs (less than 1 per cent
of the total time) and background subtracted. The errors on the light
curves were calculated by propagating the Poisson noise.
7.2 Stationarity of the data
The broad band (0.2–10 keV) light curve extracted from the pn is
shown in Fig. 6 (panel 1). As was the case for the simulated data
shown in Fig 2, the mean and variance (calculated every 20 data
points)10 show changes during the length of the observation (pan-
els 2 and 3). The expected range for the excess variance, calculated
using the results of section 5.2.2 (and assuming a PSD slope of
α = 2.0), is marked in figure 7. Fig. 8 shows the same data in
terms of normalised excess variances. Neither of these show fluc-
tuations larger than expected for a stationary process. But given the
large expected scatter this is a rather insensitive test. In the case of
the Mrk 766 light curve however, there are sufficient data to e xam-
ine variations of the average variance with time, allowing a more
sensitive test for non-stationarity.
Figure 6. Top panel: 0.2–10.0 keV pn light curve of Mrk 766 (with 25 s
bins, in units of ct s−1 ). Panel 2 and 3: mean count rate and excess vari-
ance measured from segments of 20 points. Panel 4: averaged normalised
excess variance measured by binning the individual variance estimates into
groups of 20. This average variance is inconsistent with constant. Panel 5:
fractional rms amplitude measured from segments of 20 points. Panel 6:
averaged fractional rms amplitude measured by binning the individual am-
plitudes into groups of 20. This average fractional amplitude is consistent
with constant. This contrasts with the situation shown in Fig. 2
differences in excess variance will be a combination of intrinsic dif-
ferences and measurement errors. Their uncertainty will therefore
be more difficult to quantify.
7 CASE STUDY: AN XMM-Newton OBSERVATION OF
Mrk 766
In this section a long (∼ 105 s) XMM-Newton observation of the
bright, variable Seyfert 1 galaxy Markarian 766 is used to illus-
trate the points discussed above. The data were obtained from the
XMM-Newton Data Archive9 . Details of the observation are dis-
cussed in Mason et al. (2003) and an analysis of the PSD is pre-
sented in Vaughan & Fabian (2003).
9 http://xmm.vilspa.esa.es
10 These correspond to ‘instantaneous’ estimates of the source variance on
timescales of 50 − 500 s.
10
Vaughan et al.
Figure 8. As for Fig. 7 but using the normalised excess variance of the Mrk
766 data.
By averaging the excess variance estimates (in time bins con-
taining 20 excess variance estimates) signi ficant changes i n the
variance with time are revealed (panel 4). This contrasts with the
simulated data shown in Fig. 2. The binned excess variance is in-
consistent with a constant hypothesis: fitting with a consta nt gave
χ2 = 23.1 for 9 degrees of freedom (dof ), rejected at 99 per cent
confidence. The average variance is therefore changing with time,
indicating the variability is strongly non-stationary.
A careful inspection of Fig. 6 (panels 3 and 4) shows the indi-
vidual variance estimates have a tendency to track the source count
rate. This is difficult to discern from the individual varian ces (panel
3), due to the larger intrinsic scatter, but much clearer in the av-
eraged variances (panel 4). This can be seen clearly in Fig. 9 (top
panel) where the rms amplitude (pσ 2
XS ) is shown as a function of
count rate. To produce this plot the individual rms estimates (Fig. 6,
panel 3) were sorted by count rate and binned by flux (such that
there were 20 estimates per bin). The error on the mean rms was
calculated in the standard fashion (see above). This indicates that
the source does show a form of genuine non-stationarity: the abso-
lute rms amplitude of the variations increases, on average, as the
source flux increases. This effect has been noted in other Sey ferts
(Uttley & McHardy 2001; Edelson et al. 2002; Vaughan et al. 2003)
and is due to a linear correlation between rms and flux (see Utt
ley
et al. in prep. for further discussion of this effect). Non-stationarity
of this form can be ‘factored out’ by using the normalised ampli-
NXS ) instead of the absolute values. Normalising
tude (Fvar or σ 2
each variance (or rms) estimate by its local flux removes this
trend.
The bottom panel of Fig. 9 shows that Fvar is indeed constant with
flux ( fitting a constant gave
χ2 = 7.8 for 9 dof ). Fig. 6 (panels 5
and 6) shows Fvar and its average as a function of time; the aver-
age is consistent with staying constant (χ2 = 5.8 for 9 dof ). The
variability of Mrk 766 does show genuine (strong) non-stationarity,
in the sense that the absolute rms increases linearly with flu x, but
this trend can be removed by using normalised units – Fvar (and
therefore the normalised excess variance) is consistent with being
constant (with time and flux).
The above analysis suggests that, after accounting for the ef-
fect of the rms–
flux correlation, there is no other evidence f
or
strong non-stationarity in the rapid variability of Mrk 766. This
was confirmed using the s-test of Papadakis & Lawrence (1995; see
their Appendix A). A periodogram was calculated for three consec-
utive segments of 3.4 × 104 s duration, and normalised to fractional
Figure 9. Top panel: The average absolute rms amplitude (pσ2
XS ) as a
function of flux for the Mrk 766 light curve (compare with Fig. 4). Bottom
panel: The average it fractional rms amplitude (pσ2
NXS ) as a function
of flux. Clearly the absolute rms amplitude is a function of flu
x, but this
dependence is removed in the fractional rms.
units (see Appendix A). The s value was computed by comparing
periodograms at frequencies below 2 × 10−3 Hz (above which the
Poisson noise power becomes comparable to the source variabil-
ity). For each pair of periodograms the value of s was within the
range expected for stationary data (speci fically s < 1, within one
standard deviation of the expected value).
7.3 rms spectrum
The variability amplitude as a function of energy was calculated
by measuring Fvar from light curves extracted in various energy
ranges. The results are shown in Fig. 10 and the errors were cal-
culated using equation B2 to account for the effect of Poisson
noise. The variability amplitude is clearly a function of energy,
i.e. Mrk 766 shows signi ficant spectral variability. This wa s con-
firmed by a Fourier analysis of the light curves in different e nergy
bands (Vaughan & Fabian 2003) which revealed complex energy-
dependent variability.
The rms spectrum was re-calculated for light curves contain-
ing only single pixel (pattern 0) events and again for double pixel
(patterns 1–4) events. These two sets of data were extracted from
the same detector and using identical extraction regions etc. Af-
Properties of X-ray light curves from AGN
11
curves show large fluctuations in variance. These changes pr ovide
little insight as they are expected even when the underlying physical
process responsible for the variability is constant. Rather, they may
simply be statistical fluctuations intrinsic to the stochas tic process.
All red noise processes show random fluctuations in both mean
and variance and the variance will be distributed in a non-Gaussian
fashion with a large scatter (see sections 4.2 and 5).
Previous claims of non-stationary variability based on changes
in variance (e.g. Nandra et al. 1997; Dewangan et al. 2002; Gliozzi,
Sambruna & Eracleous 2003) should therefore be treated with cau-
tion since they did not account for this intrinsic scatter (see also
section 3.3.1 of Leighly 1999 for a discussion of this point). Real
changes in the PSD would indicate genuine non-stationarity and re-
flect real changes in the physical conditions of the variabil
ity pro-
cess. Such changes can be measured from the average properties of
the light curve, such as the averaged periodogram or the averaged
variance (see section 5).
A different issue is that differences between the variance of si-
multaneous light curves obtained in different energy bands can be
examined using the excess variance (or Fvar ) statistic. It is possible
to estimate the uncertainty in the excess variance due to errors in
the flux measurements. This uncertainty, accounting only fo r mea-
surement (e.g. Poisson) errors, can be used when testing for spectral
variability, as demonstrated in section 7.
Estimators such as the excess variance provide a useful, if
crude, means of quantifying the variability of AGN. Even though
the stochastic nature of AGN light curves makes it difficult t o es-
timate variability amplitudes robustly from short observations, the
excess variance can provide useful information. For example an
analysis of the excess variances measured from short observations
of Seyfert 1 galaxies demonstrated that the variability amplitude (at
a given range of timescales) is inversely correlated with the lumi-
nosity of the source (Nandra et al. 1997; Leighly 1999; Markowitz
& Edelson 2001). Although random fluctuations in variance ar e ex-
pected for AGN light curves the range of variances observed is far
larger than could be accounted for by this effect alone. Another
example is given in section 7.2 when it is demonstrated that the av-
erage variance of Mrk 766 is a function of the flux of the source .
A similar effect has been observed in X-ray binaries (Uttley &
McHardy 2001). A discussion of the implications of this result will
be given in Uttley et al. in prep.
9 CONCLUSIONS
This paper discusses some aspects of quantifying the variability of
AGN using simple statistics such as the variance. Various possible
uses of these are presented and some possible problems with their
signi ficance and interpretation are brought to light. The pr imary
issues are as follows:
(i) In order to search for non-stationary variability in an ensem-
ble of short light curves (or short light curve segments) one can test
whether the individual variances are consistent with their mean.
Two practical methods are presented (sections 5.2.1 and 5.2.2).
(ii) In the first method the mean variance and its error are cal -
culated at various epochs by binning the individual variance esti-
mates. This is most useful when searching for subtle changes in
variability amplitude but requires large datasets (in order that the
variance can be sufficiently averaged).
(iii) In the second method the individual variance estimates are
compared with the expected scatter around the mean. The expected
scatter is calculated using Monte Carlo simulations of stationary
Figure 10. rms spectrum of Mrk 766 measured using EPIC pn light curves
with 1000 s bins.
ter accounting for the difference in count rate between single and
double pixel events the two sets of light curves should be identi-
cal except for the effects of Poisson noise. The two rms spectra
should be the same except for Poisson errors. Comparing the ra-
tio of the two rms spectra using a χ2 -test (against the hypothesis
of unity ratio) gave χ2 = 25.3/20 degrees of freedom. Compar-
ing the difference of the two rms spectra to the hypothesis of zero
difference gave identical results and shows the two rms spectra are
indeed fairly consistent. This test indicates that for real data the er-
ror formula given above does provide a reasonable description of
the uncertainty induced by photon noise.
8 DISCUSSION
The analysis of stochastic processes, such as X-ray variability of
AGN, is conceptually different from the analysis of deterministic
data such as time-averaged spectra (see discussions in e.g. Jenkins
& Watts 1968; Priestley 1981; Bendat & Piersol 1986; Bloomfie ld
2000). For example, when observing the spectrum of a constant
source one expects repeatability of the data to within the limits set
by measurement errors, i.e. each new realisation of the spectrum
should be consistent within the errors. In AGN variability analy-
sis it is the signal itself that is randomly variable; one does not
expected repeatability of quantities such as the mean or variance.
These statistical moments will change (randomly) with each new
light curve even if there are no measurement errors.
The stochastic nature of red noise processes means that it is
usually only their average properties that can provide physical in-
sight. Non-deterministic data should be handled statistically. For
example, it is customary to examine the timing properties of X-ray
binaries using PSDs estimated from the average periodogram of an
ensemble of light curves (e.g. van der Klis 1995). Averaging over
many independent realisations reduces the random fluctuati ons in-
herent in the noise process.
In most AGN timing studies however there are rarely enough
data to construct averages in this way (but see Papadakis &
Lawrence 1993 and Uttley et al. 2002 for more on PSD estima-
tion for AGN). As a result of this relative lack of data, AGN tim-
ing studies often emphasise the properties of a single light curve.
But emphasis on the detailed properties of any single realisation
of a stochastic process can be misleading. For example, AGN light
12
Vaughan et al.
processes. The table gives some examples of the scatter expected
for various PSD shapes typical of AGN. This table can therefore
be used to provide a ‘quick look’ at whether the observed fluct u-
ations in the variance are larger than expected. One drawback is
that, because the intrinsic scatter in the variance is rather large for
red noise data, this method is only sensitive to very large changes
in the variability amplitude. Another drawback is that one has to
assume a shape for the PSD.
(iv) The excess variance can also be used to quantify how the
variance changes as a function of energy (section 6.2). An approx-
imate formula is presented (based on the results of Monte Carlo
simulations) that gives the expected error in the excess variance
resulting from only observation uncertainties ( flux errors
such as
Poisson noise). This can be used to test for signi ficant diffe rences
in variance between energy bands. If the normalised excess vari-
ances (or Fvar s) are found to differ signi ficantly between energy
bands this implies the PSD is energy dependent and/or there are
independently varying spectral components.
(v) Possibly the most robust yet practical approach to variabil-
ity analysis from AGN data is to test the validity of hypotheses
using Monte Carlo simulations. This approach has yielded reliable
PSD estimates for Seyfert galaxies (Green et al. 1999; Uttley et al.
2002; Vaughan et al. 2003; Markowitz et al. 2003) and has been
used to test the reliability of cross-correlation results (e.g. Welsh
1999) amongst other things. Section 3 discusses some methods for
simulating red noise data.
ACKNOWLEDGEMENTS
We are very grateful
to the referee, Andy Lawrence, for a
thoughtful report that prompted signi ficant improvements t o the
manuscript. SV and PU acknowledge support from PPARC. This
paper made use of observations obtained with XMM-Newton, an
ESA science mission with instruments and contributions directly
funded by ESA Member States and the USA (NASA).
REFERENCES
Belloni T., Hasinger G., 1990, A&A, 227, L33
Belloni T., Psaltis D., van der Klis M., 2002, ApJ, 572, 392
Bendat J. S., Piersol A. G., 1986, Random Data: Analysis and Measurement
Procedures, Wiley (New York)
Bevington P. R., Robinson D. K,. 1992, Data Reduction and Error Analysis
for the Physical Sciences, McGraw-Hill (New York)
Bloom field P, 2000, Fourier Analysis of Time Series, Wiley (New York)
Deeming T. J., 1970, AJ, 75, 1027
Deeming T. J., 1975, Ap&SS, 36, 137
Deeter J. E., Boynton P. E., 1982, ApJ, 261, 337
Dewangan G. C., Boller Th., Singh K. P., Leighly K. M., 2002, A&A, 390,
65
Done C., Madejski G. M., Mushotzky R. F., Turner T. J., Koyama K., Ku-
nieda H., 1992, ApJ, 400, 138
Edelson R., Pike G. F., Krolik J. H., 1990, ApJ, 359, 86
Edelson R., Nandra K. 1999, ApJ, 514, 682
Edelson R. et al. 2002, ApJ, 568, 610
Fabian A. C. et al. 2002, MNRAS, 335, L1
Green A. R., McHardy I. M., Lehto H. J. 1993, MNRAS, 265, 664
Green A. R., McHardy I. M., Done C., 1999, MNRAS, 305, 309
Gliozzi M., Sambruna R. M., Eracleous M., 2003, ApJ, 584, 176
Inoue, H., Matsumoto, C., 2001, AdSpR, 28, 445
Jenkins G. M., Watts, D. G., 1968, Spectral Analysis and its Applications,
Holden-Day (San Fancisco)
Lawrence A., Watson M. G., Pounds K. A., Elvis M., 1987, Nature, 325,
694
Lawrence A., Papadakis I., 1993, ApJ, 414, L85
Leahy D. A., Darbro W., Elsner R. F., Weisskopf M. C., Kahn S., Sutherland
P. G., Grindlay J. E., 1983, ApJ, 266, 160
Lehto H. J., 1989, in J. Hunt, B. Battrick, eds, Two Topics in X Ray Astron-
omy, (ESA SP-296; Noordwijk: ESA), p499
Leighly K., 1999, ApJS, 125, 297
Markowitz A., Edelson R., 2001, ApJ, 547, 684
Markowitz A. et al. 2003, ApJ, in press (astro-ph/0303273 )
Mason K. O., et al. 2003, ApJ, 582, 95
McHardy I. M., 1989, in J. Hunt, B. Battrick, eds, Two Topics in X Ray
Astronomy, (ESA SP-296; Noordwijk: ESA), p1111
Nandra K., George I. M., Mushotzky R. F., Turner T. J., Yaqoob T., 1997,
ApJ, 476, 70
Papadakis I. E., Lawrence A., 1993, MNRAS, 261, 612
Papadakis I. E., Lawrence A., 1995, MNRAS, 272, 161
Pottschmidt K., 2002, A&A, submitted (astro-ph/0202258 )
Press W. H. 1978, Comments on Astrophysics, 7, 103
Press W. H., Teukolsky S. A., Vetterling W. T., Flannery B. P., 1996, Nu-
merical Recipes, Cambridge Univ. Press. (Cambridge)
Press W. H., Rybicki G. B., 1997, in Astronomical Time Series, D. Maoz,
A. Sternberg, E. M. Leibowitz (eds.), Kluwer (Dordrecht), p61
Priestley M. B., 1981, Spectral Analysis and Time Series, Academic Press
(London)
Rodriguez-Pascual P. M. et al. , 1997, ApJS, 110, 9
Scargle J. D., 1981, ApJS, 45, 1
Schurch N. J., Warwick R. S., 2002, MNRAS, 334, 811
Stella L., Arlandi E., Tagliaferri G., Israel G. L., 1997, in ‘Applications
of Time Series Analysis in Astronomy and Meteorology’, T. Subba
Rao, M. B. Priestley, O. Lessi (eds.), Chapman & Hall (London)
(astro-ph/9411050 )
Timmer J., K onig M. 1995, A&A, 300, 707
Turner T. J., George I. M., Nandra K., Turcan D., 1999, ApJ, 524, 667
Uttley P., McHardy I. M., 2001, MNRAS, 323, 26
Uttley P., McHardy I. M., Papadakis, I. 2002, MNRAS, 332, 231
van der Klis M., 1989, in Timing Neutron Stars, H. Ogelman, E. P. J. van
den Heuvel (eds.), Kluwer (Dordrecht), NATO ASI Series C 262, p27
van der Klis M., 1995, in X-ray Binaries, W. H. G. Lewin, J. van Paradijs, E.
P. J. van den Heuvel (eds.), Cambridge Univ. Press (Cambridge), p252
van der Klis M., 1997, in Statistical Challenges in Modern Astronomy II,
ed. G.J. Babu, E.D. Feigelson, Springer-Verlag (New York), p321
Vaughan S., Fabian A. C., Nandra K. 2003, 339, 1237
Vaughan S., Fabian A. C., 2003, MNRAS, 341, 496
Welsh W. F., 1999, PASP, 111, 1347
APPENDIX A: PERIODOGRAM NORMALISATION
(A1)
The periodogram is calculated by normalising the modulus-squared
of the DFT (see equation 1):
P (fj ) = ADF T (fj )2 .
There are a variety of options for the normalisation A used
in the literature, each has desirable properties. In the mathematical
literature on time series analysis a normalisation of the form A =
2/N is standard (e.g. Priestley 1981; Bloomfield 2000). However ,
this normalisation is generally not used for time series analysis in
astronomy because the periodogram then depends on flux of the
source and the binning of the time series. Below are listed three
of the most commonly used normalisations, which only differ by
factors of ¯x, the mean count rate in cts/s (Aabs = ¯xALeahy =
¯x2Arms2 ). The factor of two is present in all these normalisations
to make the periodogram ‘one sided,’ meaning that integrating over
positive frequencies only yields the correct variance.
(i) Arms2 = 2∆Tsamp / ¯x2N – defined by van der Klis (1997)
(see also Miyamoto et al. 1991). This is the normalisation most
often used in analysis of AGN and X-ray binaries because the inte-
grated periodogram yields the fractional variance of the data. The
units for the periodogram ordinate are (rms/mean)2 Hz−1 (where
rms/mean is the dimensionless quantity Fvar ), or simply Hz−1 .
If a light curve consists of a binned photon counting signal (and
in the absence of other effects such as detector dead-time) the ex-
pected Poisson noise ‘background’ level in its periodogram is given
by
,
(A2)
Pnoise =
2( ¯x + B )
∆Tsamp
¯x2
∆Tbin
where ¯x is the mean source count rate, B is the mean background
count rate, ∆Tsamp is the sampling interval and ∆Tbin is the time
bin width. The factor of ∆Tsamp /∆Tbin accounts for aliasing of
the Poisson noise level if the original photon counting signal con-
tained gaps. If the light curve is a series of contiguous time bins
(i.e. ∆Tbin = ∆Tsamp ) and has zero background (which is approx-
imately true for many XMM-Newton light curves of AGN) then this
reduces to Pnoise = 2/ ¯x.
For a light curve with Gaussian errors σerr,i the noise level in the
periodogram is
.
(A3)
Pnoise =
2∆Tbinσ 2
∆Tsamp
err
¯x2
∆Tbin
(ii) ALeahy = 2∆Tsamp / ¯xN – originally due to Leahy et al.
(1983). This has the property that the expected Poisson noise level
is simply 2 (for continuous, binned photon counting data). If the
light curve consists only of Poisson fluctuations then the pe ri-
odogram should be distributed exactly as χ2
2 . It is this property that
makes this normalisation the standard for searching for periodic
signals in the presence of Poisson noise (see Leahy et al. 1983).
If the input light curve is in units of ct s−1 then the periodogram
ordinate is in units of ct s−1 Hz−1 .
(iii) Aabs = 2∆Tsamp /N – this is the normalisation used in
equation 2. This gives the periodogram in absolute units [e.g. (ct
s−1 )2 Hz−1 ] and so the integrated periodogram gives the total vari-
ance in absolute units [e.g. (ct s−1 )2 ] For a contiguously binned
light curve with Poisson errors the noise level is Pnoise = 2 ¯x, and
for Gaussian errors the noise level is Pnoise = 2∆Tbinσ 2
err .
APPENDIX B: MONTE CARLO DEMONSTRATION OF
POISSON NOISE INDUCED UNCERTAINTY ON EXCESS
VARIANCE
To estimate the effect on σ 2
NXS due only to Poisson noise the basic
strategy was as follows.
(i) Generate a random red noise light curve. This acts as the
‘true’ light curve of the source.
(ii) Add Poisson noise, i.e. draw fluxes from the light curve a c-
cording to the Poisson distribution. This simulates ‘observing’ the
true light curve. Error bars were assigned based on the ‘observed’
counts in each bin (√counts).
(iii) Measure the normalised excess variance σ 2
NXS of the ob-
served light curve. This will be different from the variance of the
true light curve because of the Poisson noise.
Steps 2 and 3 were repeated, using the same true light curve, to
obtain the distribution of σ 2
NXS . Fig. B1 shows some results. In this
example the ‘true’ light curve was generated with a f −2 PSD and
Properties of X-ray light curves from AGN
13
normalised to a pre-defined mean and variance, e.g. S 2 / ¯x2 = 0.04
(Fvar = 20%). This light curve was then observed (i.e. steps 2
and 3 were repeated) 104 times11 . The three panels correspond to
different mean count rates for the true light curve (i.e. different S/N
of the observation). The (1σ ) widths of the σ 2
NXS distributions are
Monte-Carlo estimates of the size of the error bars on σ 2
NXS due to
Poisson noise.
As is clear from Fig. B1 the distribution of σ 2
NXS becomes
narrower, i.e. the error on σ 2
NXS gets smaller, as the S/N of the
data increases. Obviously in the limit of very high S/N data the
measured value of σ 2
NXS will tend to the ‘true’ value (in this case
0.04), i.e. err(σ 2
NXS) → 0 as counts → ∞. It should also be
noted that the distributions are quite symmetrically centred on the
correct value, indicating that σ 2
NXS is an unbiased estimator of the
intrinsic variance in the light curve, even in relatively low S/N data.
In order to assess how the error on σ 2
NXS changes with S/N,
the width of its distribution was measured from simulated data at
various different settings of S/N and intrinsic variance (i.e. S 2 / ¯x2 ).
Width of the distribution at each setting was calculated from only
500 ‘observations’ of each light curve. In order that no particular
realisation adversely affect the outcome, and to increase the statis-
tics, this was repeated for 20 different random light curves (of the
same fractional variance) and the width of the σ 2
NXS distributions
were averaged (i.e. the whole cycle of steps 1–3 was repeated 20
times). Thus for each speci fied value of S/N and fractional va riance,
the error on σ 2
NXS is estimated from 104 simulated ‘observations.’
These Monte Carlo estimated errors on the normalised excess vari-
ance are shown in Fig. B2.
The solid lines show the functions defined by equation 11
(which was obtained by fitting various trial functions to the Monte
Carlo results). Clearly this equation gives a very good match to the
Monte Carlo results.
If the variability is not well detected, either because the S/N
is low or the intrinsic amplitude is weak, then S 2 ≈ σ 2
err . It is the
first term on the right hand side of equation 11 that dominates . If
the variability is well detected, i.e. S 2 ≫ σ 2
err , then it is the second
term that dominates.
NXS) ≈
N · σ2
p 2
: S 2 ≈ σ 2
err
err
¯x2
err(σ 2
q σ2
N · 2Fvar
: S 2 ≫ σ 2
err .
err
¯x
In the former case the deviations from the mean are dominated
by the errors and the fluxes are approximately normally distr ibuted.
In this regime the error equation becomes the same as that given
in equation A9 of Edelson et al. (2002). In the latter case the de-
viations in the light curve are enhanced by the intrinsic variance.
The second term is similar to the first except multiplied by a f actor
q2σ 2
err to account for this.
XS /σ 2
Equation 11 can be used to give the uncertainty on Fvar thusly
1
err(σ 2
err(Fvar ) =
NXS) =
2Fvar
vuut(r 1
¯x2Fvar )2
+ (r σ 2
σ 2
err
err
N ·
2N ·
11 As this measures only the effect due to Poisson noise, the results are
largely independent of the details of the light curve, including the PSD, as
long as the flux is non-zero throughout the light curve. This w as confirmed
by repeating the above experiment using data produced from PSD slopes in
the range α = 0 − 2.
¯x )2
1
(B1)
,
(B2)
Figure B2. Width of the distribution of σ2
NXS (resulting from Poisson
noise) as a function of the number of counts per bin. Compare with Fig. B1.
The solid curve shows the function described in the text (equation 11).
errors on the fluxes. It does not account for the intrinsic sca tter in
the fluxes inherent in any red noise process.
This paper has been typeset from a TEX/ LATEX file prepared by the
author.
14
Vaughan et al.
Figure B1. Distribution of measured σ2
NXS from 10,000 ‘observations’ of
the same light curve. In each case the ‘true’ σ2
NXS is 0.04 (dotted line). The
top panel used the lowest S/N data, the bottom panel used the highest S/N
data. The mean number of counts per bin in the simulated light curves was
15 (top), 30 (middle) and 100 (bottom). As the S/N increases (count rate
increases) the distribution of σ2
NXS becomes narrower. (Note: this is differ-
ent from Fig. 3, which shows how the variance changes between different
realisations of the same stochastic process.)
and this is the equation used to derive the errors shown in Fig. 10.
In the two regimes this becomes:
err(Fvar ) ≈
2N · σ2
p 1
: S 2 ≈ σ 2
err
err
¯x2 Fvar
q σ2
: S 2 ≫ σ 2
N · 1
err .
err
¯x
In the first instance, when the variability is not well detect ed, σ 2
NXS
should be preferred over Fvar as negative values of σ 2
NXS are possi-
ble. Additional Monte Carlo simulations confirmed the above equa-
tions are valid for both Gaussian and Poisson distributed flu x errors.
It is worth reiterating that this error accounts only for measurement
(B3)
|
astro-ph/0404556 | 1 | 0404 | 2004-04-28T13:32:53 | The structure of the ICM from High Resolution SPH simulations | [
"astro-ph"
] | We present results from a set of high (512^3 effective resolution), and ultra-high (1024^3) SPH adiabatic cosmological simulations of cluster formation aimed at studying the internal structure of the intracluster medium (ICM). We derive a self-consistent analytical model of the structure of the intracluster medium (ICM). We discuss the radial structure and scaling relations expected from purely gravitational collapse, and show that the choice of a particular halo model can have important consequences on the interpretation of observational data. The validity of the approximations of hydrostatic equilibrium and a polytropic equation of state are checked against results of our simulations. The properties of the ICM are fully specified when a 'universal' profile is assumed for either the dark or the baryonic component. We also show the first results from an unprecedented large-scale simulation of 500 Mpc/h and 2 times 512^3 gas and dark matter particles. This experiment will make possible a detailed study of the large-scale distribution of clusters as a function of their X-ray properties. | astro-ph | astro-ph |
Outskirts of Galaxy Clusters: intense life in the suburbs
Proceedings IAU Colloquium No. 195, 2004
A. Diaferio, ed.
c(cid:13) 2004 International Astronomical Union
DOI: 00.0000/X000000000000000X
The structure of the ICM from High
Resolution SPH simulations
G. Yepes1, Y. Ascasibar2, R. Sevilla1, S. Gottlober3 and V. Muller3
1Grupo de Astrof´ısica, Universidad Aut´onoma de Madrid, Madrid E-280049, Spain
2Theoretical Physics, University of Oxford, 1 Keble Road, Oxford OX1 3NP, United Kingdom
3Astrophysikalisches Institut Potsdam, An der Sternwarte 16, Potsdam D-14482, Germany
Abstract. We present results from a set of high (5123 effective resolution), and ultra-high
(10243) SPH adiabatic cosmological simulations of cluster formation aimed at studying the in-
ternal structure of the intracluster medium (ICM). We discuss the radial structure and scaling
relations expected from purely gravitational collapse, and show that the choice of a particular
halo model can have important consequences on the interpretation of observational data. The
validity of the approximations of hydrostatic equilibrium and a polytropic equation of state are
checked against results of our simulations. We also show the first results from an unprecedented
large-scale simulation of 500 h−1 Mpc and 2 × 5123 gas and dark matter particles. This experi-
ment will make possible a detailed study of the large-scale distribution of clusters as a function
of their X-ray properties.
1. Introduction
Galaxy clusters are a unique laboratory to test the hierarchical paradigm of structure
formation. They are the best probes of the large scale structure of the Universe and have
often been used as a diagnostic of the cosmological parameters. The intrinsic non-linear
nature of gravitational collapse and gas dynamics makes numerical simulations the most
useful tool to study in detail the process of cluster formation and evolution.
Simple analytical models for the structure of the ICM can be derived from the hy-
potheses of hydrostatic equilibrium and polytropic equation of state, P ∝ ργ
g . But real
clusters might not be well described by these two hypothesis. For instance, kinetic energy
makes a significant contribution to the energy budget of merging systems, and therefore
thermally-supported hydrostatic equilibrium ceases to be a valid approximation. Even in
relaxed systems, this assumption is not very accurate in the outermost parts, where gas
motions become more important. Departures from spherical symmetry can also play a
role in the final structure of the ICM (Lee & Suto 2003) and, last but not least, there is
no obvious physical reason for the gas to follow a polytropic relation. From our numeri-
cal experiments, we showed (see Ascasibar et al. (2003)) that hydrostatic equilibrium is
fulfilled within ∼ 20 per cent accuracy by all simulated clusters, as long as they are not
heavily disturbed. A polytropic equation of state seems to be a good approximation as
well, although its reliability near the centre is still a matter of debate (see e.g. Rasia et al.
2003). From our data we derive a polytropic index of γ ∼ 1.18.
We compared four different analytical halo models with our simulations. The first two
models assume that haloes are well described by Navarro et al. (1997) and Moore et al.
(1999) fitting formulae, while the other two assume that the gas follows a β-model
(Cavaliere & Fusco-Femiano 1976). We consider a 'canonical' version of the β-model, in
which the gas is isothermal (γ = 1) and β = 2/3, and a polytropic version with γ = 1.18
and β = 1. The same value of the polytropic index has been used for the first two models
1
2
G. Yepes et al.
as well. Hereafter we will use the abbreviations NFW, MQGSL, BM and PBM to refer to
these models. For a detailed description, the reader is referred to Ascasibar et al. (2003).
2. Numerical experiments
We have carried out a series of high-resolution gasdynamical simulations of cluster for-
mation in a flat ΛCDM universe (Ωm = 0.3; ΩΛ = 0.7; h = 0.7; σ8 = 0.9; Ωb = 0.02 h−2).
Simulations have been done with the parallel GADGET code (Springel et al. 2001), with
a novel version of SPH in which the entropy is explicitly conserved (Springel & Hernquist
2002). In a cubic volume of 80 h−1 Mpc on a side, an unconstrained realization of the
power spectrum of density fluctuations corresponding to the ΛCDM model was gener-
ated for a total of 10243 Fourier modes. The density field was then resampled to a grid
of 1283 particles, which were displaced from their Lagrangian positions according to the
Zeldovich approximation up to z = 49. Their evolution until the present epoch is traced
by means of a pure N-body simulation with 1283 dark matter particles. A sample of
clusters selected from this preliminary low-resolution experiment were re-simulated with
higher resolution by means of the multiple mass technique (see Klypin et al. 2001, for
details). Mass resolution is then increased by using smaller masses in the Lagrangian
volume depicted by these particles, including the additional small-scale waves from the
ΛCDM power spectrum in the new initial conditions. We use 3 levels of mass refinement,
reaching an effective resolution of 5123 CDM particles (2.96 × 108 h−1 M⊙). Gas was
added in the highest resolved area only. The total number of particles (dark+SPH) in
this area is greater than 1 − 2 × 106 for all clusters. The gravitational softening length
was set to ǫ = 2 − 5 h−1 kpc, depending on number of particles within the virial radius
(Power et al. 2003). The minimum smoothing length for SPH was fixed to the same value
as ǫ.
In order to study effects of resolution in the determination of X-ray properties of our
clusters, we have resimulated one of the objects with 8 times more mass resolution,
reaching an effective resolution of 10243 particles (i.e. mdark ∼ 3 × 107 h−1M⊙; msph ∼
5 × 106M⊙). The total number of particles within the virial radius was 11, 106, 465.
The list of our simulated clusters extracted from the 80 h−1 Mpc volume span a
relative small range in X-ray emission temperature (from 0.6 to 3 keV). In order to extend
our numerical sample of clusters to wider temperature (mass) range, we have simulated
a considerable much bigger volume (500 h−1 Mpc) in which a random realization of
the ΛCDM power spectrum was generated with 20483 particles. In this way, we will
have a similar resolution for our clusters than in the previous experiments. We have
resimulated the whole 500 h−1 Mpc box with different mass resolutions: 2×1283, 2×2563,
and 2 × 5123 dark and SPH particles. We identified all halos in the lowest resolution
run (1283). Using the same technique as before, we selected clusters in this run and
resimulated them at full resolution (20483), using 5 different species of dark particles
and SPH with msph ∼ 1.6 × 108h−1M⊙. The clusters selected cover a range of masses
from 2 × 1014 − 2 × 1015 M⊙. In this regard, we could extend the temperature range of
our cluster sample up to 11 keV. The total number of particles within virial radius in
these clusters is comparable to the number of particles of previous simulations (∼ 106).
We have recently finished the run with 2 × 5123 gas and dark particles that, to our
knowledge, is one of the largest adiabatic SPH simulations of large scale structure done
so far. The mass resolution is mdark ∼ 6 × 1010h−1 M⊙, which means that we can resolve
from galactic halos (100+100 particles) to the biggest galaxy clusters (4 × 106 particles).
We identified a total of 4 × 105 dark matter halos, with 10 or more particles, in this
simulation. Well resolved halos (7000 dark particles or more) correspond to clusters with
Outskirts of Galaxy Clusters: intense life in the suburbs
3
Figure 1. Results from 5003 simulated volume. Left, the X-ray Temperature function and its
observational estimates (Ikebe et al. 2002). Middle, the Mvir − Tx relation. Right, the LX − TX
relation. Stars represent the clusters resimulated with high resolution. The dotted line is a fit
to the hottest halos ( > 6 keV). The slope is ∼ 2, as in the self-similar scaling behaviour
emission weighted temperatures > 3 keV. If we go down to 1000 particles, the halos
resolved have temperatures > 1 keV, and temperatures go down to ∼ 0.6 keV for halos
with 500 or more particles. Due to the large simulated volume, we have a statistical
significant sample of clusters and groups. The total number of objects with TX > 0.6
keV, excluding substructure, is ∼ 30, 000. A total of 126 hot clusters have been found
with TX > 5 keV. The X-ray temperature function for clusters in this simulation is
shown in Figure 1, together with recent observational estimates. As can be seen, the
number density of clusters for the highest temperatures (> 6 keV) found in simulations
is compatible with observations (see Borgani et al. (2004) for results from non-adiabatic
simulations).
The resulting M − TX and LX − TX relations are depicted in Figure 1, in which
we also show 6 of the hotter clusters resimulated with high resolution. They show a
convergence of results at least for clusters with TX > 3 keV. The fit to the M − TX
relation for clusters with TX > 1 keV is (Mvir/M0) = (Tx/keV)α, with α = 1.56 ± 0.05
and M0 = 4.4 × 1013h−1M⊙. The slope is compatible with simulations which include
non adiabatic effects, although the zero point, M0, is a factor of 2 higher. Note however
that we are using virial mass instead of mass for overdensity 500. This implies that the
M-T relation is rather insensitive to the energy transfers due to non adiabatic processes
associated to star-gas interactions.
3. Radial structure of the ICM
We compare the universal halo models described earlier, with results from our simu-
lations. For NFW and MQGSL, we obtain the characteristic density and radius of each
cluster by fitting the dark matter distribution. Gas density and temperature profiles are
genuine predictions of these models. For BM and PBM, we fit the gas distribution and
predict the gas temperature and the dark matter density. In Figure 2 we plot the average
dark matter and gas density profiles. Not surprisingly, both NFW and MQGSL provide
good fits to the dark matter density. These models are able to accurately predict the
gas density, but they are too steep at large radii due to a systematic departure from
hydrostatic equilibrium.
On the other hand, β-models (BM and PBM) show a core in the dark matter density
that is not seen in the numerical data. Moreover, they do not give an accurate description
4
G. Yepes et al.
Figure 2. Universal radial profiles: Upper panel dark matter and gas density. Lower panel:
projected temperature and gas entropy. Black squares represent the numerical data, averaged
over all clusters except major mergers. Error bars denote one-sigma scatter. Lines are used
to plot the analytical models, and top panels quantify deviations from the simulated profile.
Dashed lines in the projected emission-weighted temperature profile, shows the 'universal' profile
proposed by Loken et al. (2002), while solid line represents the best fit to our data. Observations
by De Grandi & Molendi (2002, circles) and Markevitch et al. (1998, boxed region) are ploted
for comparison.
of the gas density profile. The inner regions are better described with low values of β,
while the outer parts require higher values for this parameter.
One of the most striking results from our simulations is that they favour the hy-
pothesis of a 'universal' temperature profile. As can be seen in Figure 2, the ICM is
not isothermal, but the temperature decreases by a factor of three or four from the
centre to the virial radius. The projected X-ray emission-weighted temperature profile
is plotted on the right panel of Figure 2. We compare our simulations with previous
work by Loken et al. (2002) based on Eulerian simulations. These authors propose a
'universal' form Tp(r) = T0(1 + r/ax)−δ Our results are well described by this rela-
tion. Although real clusters seem to be consistent with a polytropic equation of state
(Markevitch et al. 1998), recent observations indicate the presence of a large isothermal
core (De Grandi & Molendi 2002). Apart from this feature, which is not observed in our
objects, adiabatic simulations are in good agreement with observational data beyond
0.2R200.
Entropy profiles are plotted on the lower panel of Figure 2. Contrary to the com-
mon view, neither the analytical models nor the simulation data yield a pure power-
law profile, despite the fact that purely adiabatic gasdynamics has been considered. As
shown in Ascasibar et al. (2003), the standard implementation of the SPH algorithm can
lead to misleading results in the inner regions due to artificial entropy losses (see e.g.
Springel & Hernquist 2002). We find that the shape of the entropy profile does not depend
on the cluster mass or temperature, in agreement with recent observations (Ponman et al.
2003). NFW or MQGSL models provide a better estimate of the entropy profile than the
β-model, but the low gas densities predicted at large radii yield a very steep slope at
r ∼ R200.
Outskirts of Galaxy Clusters: intense life in the suburbs
5
Figure 3. Radial profiles for a cluster run with two different resolutions.
4. Convergence of results
To check for convergence of results in terms of numerical resolution, we compare, in
Figure 3 radial profiles for one of our clusters from the ultra-high resolution simulation
described in previous section. As it can be appreciated, the structure of the ICM is quite
similar, at least from the virial radius down to 1% of that. The features at the most
internal parts of the cluster are related with the different dynamical stage of the cluster.
The positive results of this comparison is that the overall properties of the halos are well
described with the resolution adopted in our simulations in which 500,000-1,000,000 SPH
particles are used.
Acknowledgements
This work has been partially supported by the MCyT (Spain, AYA-0973), by the Ac-
ciones Integradas Hispano-Alemanas HA2000-0026 and by DAAD (Germany). We thank
the Forschungszentrum Julich for allowing us to use the IBM p690+ supercomputer.
References
Ascasibar, Y., Yepes, G., Muller, V., & Gottlober, S. 2003, MNRAS, 346, 731
Borgani, S., Murante, G., Springel, V., et al. 2004, MNRAS, 348, 1078
Cavaliere, A. & Fusco-Femiano, R. 1976, A&A, 49, 137
De Grandi, S. & Molendi, S. 2002, ApJ, 567, 163
Ikebe, Y., Reiprich, T. H., Bohringer, H., Tanaka, Y., & Kitayama, T. 2002, A&A, 383, 773
Klypin, A., Kravtsov, A. V., Bullock, J. S., & Primack, J. R. 2001, ApJ, 554, 903
Lee, J. & Suto, Y. 2003, ApJ, 585, 151
Loken, C., Norman, M. L., Nelson, E., et al. 2002, ApJ, 579, 571
Markevitch, M., Forman, W. R., Sarazin, C. L., & Vikhlinin, A. 1998, ApJ, 503, 77
Moore, B., Quinn, T., Governato, F., Stadel, J., & Lake, G. 1999, MNRAS, 310, 1147
Navarro, J. F., Frenk, C. S., & White, S. D. M. 1997, ApJ, 490, 493
Ponman, T. J., Sanderson, A. J. R., & Finoguenov, A. 2003, MNRAS, 343, 331
Power, C., Navarro, J. F., Jenkins, A., et al. 2003, MNRAS, 338, 14
Rasia, E., Tormen, G., & Moscardini, L. 2003, astroph/0309405
Springel, V. & Hernquist, L. 2002, MNRAS, 333, 649
Springel, V., Yoshida, N., & White, S. D. M. 2001, New Astronomy, 6, 79
|
astro-ph/9810357 | 1 | 9810 | 1998-10-22T09:21:41 | The eclipsing Cataclysmic Variable GS Pavonis: Evidence for disk radius changes | [
"astro-ph"
] | We have obtained differential time series photometry of the cataclysmic variable GS Pavonis over a timespan of 2 years. These show that this system is deeply eclipsing (~2-3.5 mag) with an orbital period of 3.72 hr. The eclipse depth and out-of-eclipse light levels are correlated. From this correlation we deduce that the disk radius is changing and that the eclipses in the low state are total. The derived distance to GS Pav is 790+/-90 pc, with a height above the galactic plane of 420+/-60 pc. We classify GS Pav as a novalike system. | astro-ph | astro-ph |
A&A manuscript no.
(will be inserted by hand later)
Your thesaurus codes are:
06(08.02.2;08.09.2; 08.14.2)
ASTRONOMY
AND
ASTROPHYSICS
The eclipsing Cataclysmic Variable GS Pavonis: Evidence
for disk radius changes ⋆
P.J. Groot1, T. Augusteijn2, O. Barziv1,3, and J. van Paradijs1,4
1 Astronomical Institute 'Anton Pannekoek'/ CHEAF, Kruislaan 403, 1098 SJ, Amsterdam, The Netherlands
2 European Southern Observatory, Casilla 19001, Santiago 19, Chile
3 European Southern Observatory, Karl-Schwarzschildstr. 2, D-85748, Garching-bei-Munchen, Germany
4 Physics Department, University of Alabama in Huntsville, Huntsville, USA
Received date, accepted date
Abstract. We have obtained differential time series pho-
tometry of the cataclysmic variable GS Pavonis over a
timespan of 2 years. These show that this system is deeply
eclipsing (∼2 -- 3.5 mag) with an orbital period of 3.72 hr.
The eclipse depth and out-of-eclipse light levels are corre-
lated. From this correlation we deduce that the disk radius
is changing and that the eclipses in the low state are to-
tal. The derived distance to GS Pav is 790±90 pc, with a
height above the galactic plane of 420±60 pc. We classify
GS Pav as a novalike system.
Key words: Binaries:eclipsing -- stars: GS Pav -- cata-
clysmic variables
Table 1. Log of V-band observations of GS Pav.
Date
Start UT Integr. Time (s) No. Obs.
Sept 5, 1993
Sept 6, 1993
Sept 7, 1993
Sept 8, 1993
March 22, 1995
March 23, 1995
June 13, 1995
July 11, 1995
July 25, 1995
Aug 16, 1995
Sept 15, 1995
Sept 19, 1995
01h50m
23h21m
23h11m
03h57m
08h06m
07h15m
03h20m
05h40m
04h23m
04h52m
03h23m
02h50m
120
120
120
120
240
240
240
240
120
120
240
240
82
66
48
25
25
30
61
65
24
54
21
20
1. Introduction
Eclipsing non-magnetic Cataclysmic Variables (CVs) (for
a review see Warner 1995, hereafter W95) are of particular
interest not only because the masses of both stars can be
determined, but especially because studying their eclipses,
e.g., by the eclipse mapping method (Horne 1985), gives
the opportunity to learn more about the physics of ac-
cretion disks. In this Letter we report that GS Pav is an
eclipsing CV, that shows substantial disk radius changes.
GS Pav was first discovered by Hoffmeister (1963)
who denoted it as star S7040 and gave the comment
'raschwechselnd' (rapidly varying). It was classified as a
dwarf nova type CV in the GCVS and in the catalogue
of Downes, Webbink and Shara (1997), who also give a
finding chart for the object. It was selected for our ob-
servations as a possible member of the halo population
(Augusteijn, 1994). Zwitter and Munari (1995) show that
it has a normal CV spectrum. We determined its location
at RA= 20h08m07.s58, Dec= -- 69◦48′58.′′1 (J2000), almost
identical to that of Downes, Webbink and Shara (1997).
Send offprint requests to: Paul Groot ([email protected])
⋆ Based on observations with the 0.9m Dutch telescope at
ESO, Chile
2. Observations
Photometric observations were obtained with the Dutch
0.9m telescope at ESO La Silla, Chile. A log of the ob-
servations is given in Table 1. All observations were made
with a 512x512 TEK CCD detector, using a Bessel V fil-
ter. Standard flatfielding and debiasing were applied to all
observations. A photometric calibration was obtained on
September 6, 1993 using the standard star EG 21 (Lan-
dolt 1992). Table 2 gives the coordinates and magnitudes
of the reference stars we have used.
3. Photometric ephemeris
Arrival times of mid-eclipse were determined by fitting a
Gaussian profile to the eclipses, and by determining the
mid-point between the points of steepest ascent and de-
scent. The final arrival times listed in Table 3 were taken
as the average of the results from these two methods. We
estimate the accuracy of these arrival times to be 5·10−4
days, which corresponds to the typical difference between
the results from the two methods.
2
P.J. Groot et al.: The Eclipsing Cataclysmic Variable GS Pavonis
Table 2. Reference stars used for the differential photom-
etry of GS Pav.
No. Name
RA (J2000) Dec (J2000) Va
1
2
3
4
5
6
7
8
GAB J200816 -- 6949
GAB J200810 -- 6949
GAB J200804 -- 6949
GAB J200755 -- 6949
GAB J200802 -- 6948
GAB J200800 -- 6948
GAB J200757 -- 6948
GAB J200810 -- 6947
20h08m16.
20h08m10.
20h08m04.
20h07m55.
20h08m02.
20h08m00.
20h07m57.
20h08m10.
s47
s49
s83
s45
s49
s64
s63
s41
-- 69◦49.
-- 69◦49.
-- 69◦49.
-- 69◦49.
-- 69◦48.
-- 69◦48.
-- 69◦48.
-- 69◦47.
′36.
′34.
′40.
′14.
′57.
′49.
′19.
′55.
′′0
′′0
′′4
′′4
′′6
′′5
′′6
′′4
17.31(4)
17.41(4)
17.42(4)
17.09(4)
18.54(7)
16.33(3)
17.87(5)
15.45(2)
Figure 1 shows the phase folded eclipse light curves for
the 12 epochs listed in Table 1. The light curves of June
13 and July 11, 1995 contain two eclipses each.
The width of the eclipse (∆ϕ) can be estimated by
the phases of steepest descent and ascent. For GS Pav we
measure a mean ∆ϕ = 0.064 ± 0.005, where the error
is the scatter on the average of all measurements. From
this value and the range in mass ratio's we deduce (Horne
1985) a range in orbital inclination of 74◦< i < 83◦.
5. Correlation between eclipse depth and
out-of-eclipse light
a The quoted errors reflect the internal errors in the bright-
ness measurements of the stars which do not include a 0.1 mag
uncertainty in the transformation to standard magnitudes.
Table 3. Times of arrival, deduced cycle numbers and the
observed minus computed (O -- C) residuals for the obser-
vations of GS Pav
Cycle No. HJDmin -- 244 0000 O -- C (days)
-- 3062
-- 3050
-- 3044
-- 3043
565
571
1098
1099
1279
1280
1369
1511
1704
1729
9235.7370
9237.6014
9238.5330
9238.6876
9798.9010
9799.8340
9881.6603
9881.8147
9909.7641
9909.9191
9923.7376
9945.7857
9975.7542
9979.6362
-- 0.70·10−3
0.46·10−3
0.44·10−3
-- 0.23·10−3
-- 0.32·10−3
1.05·10−3
0.16·10−3
-- 0.71·10−3
0.13·10−3
-- 0.14·10−3
-- 0.66·10−3
-- 0.87·10−3
0.56·10−3
0.82·10−3
A linear fit to the arrival times listed in Table 3 yields
the following ephemeris:
HJDmin = 244 9711.17388(17) + 0.155269817(87) · N, (1)
with N the cycle number. The error estimates for the pa-
rameters are scaled to give a χ2
red = 1.0. The rms value of
the arrival times around the fit is 6.4·10−4 days, which is
in reasonable agreement with our error estimate.
4. Mass ratio, width of the eclipse and inclination
If the secondary follows the lower main-sequence standard
mass-period relation (W95):
M2 = 0.065 P
5/4
orb (h)
1.3 ≤ Porb(h) ≤ 9,
(2)
with M2 the mass of the secondary, M2 is ∼ 0.34 M⊙.
Since the mass ratio, q=M2/M1, has to be smaller than
2/3 for stable mass transfer (W95) and the mass of the pri-
mary can at most be the Chandrasekhar mass (1.4 M⊙),
the mass ratio is limited to the range 0.67 ≥ q ≥ 0.24.
Figure 1 shows that GS Pav does not have a constant
out-of-eclipse magnitude. In our observations the source
varied between V∼14.9 and V∼ 17.1, being mostly at the
bright end. Also the eclipse depth is not constant. We have
investigated if these two variations are correlated. Since
the eclipses are well represented by Gaussian functions,
we have taken the zero-level and depth of the Gaussians,
used to determine the times of mid-eclipse, as estimates
of the depth in magnitudes of mid-eclipse (∆m) and the
out-of-eclipse light level (mV ) (Fig. 2). The numbers in
Fig. 2 refer to the eclipses as shown in Fig. 1.
In the following we will make a distinction between
the observed radius of the accretion disk, which is the size
we infer from our observations, and the physical radius
of the accretion disk. The difference between these two is
determined by the fractions of the optically thick and thin
parts of the accretion disk and the brightness distribution
across the disk which determines what parts are visible
in the chosen passband (here in the V-band). A changing
brightness distribution can mimic a change in the physical
disk radius, when observed in only one band.
In Fig. 2 the straight line labeled 'Line of Totality'
shows what the correlation between ∆m and mV looks
like for a system in which the eclipse is total. A total
eclipse means that the observed size of the accretion disk
is smaller than the size of the secondary and that at mid-
eclipse the amount of observed light from the disk is neg-
ligible. In a total eclipse the mid-eclipse light level will
be constant and equal to the brightness of the secondary.
Every magnitude of brightening of the disk will cause a
magnitude of deepening of the eclipse: the system will fol-
low a straight line, with an angle of 45◦, in the ∆m-mV
diagram. The position of this line in the diagram will be
different for each individual system, but can be fixed by
determining the brightness of the secondary. If at any time
during our observation GS Pas was totally eclipsing, then
its mid-eclipse light level will be the brightness of the sec-
ondary. The minimum level occured on June 13 and July
11, 1995: V =19.9±0.1. We use this point to fix the position
of the 'Line of Totality' in Fig. 2. We see that points '7'
and '8' (which are from June 13 and July 11, 1995) lie on
this line. From the fact that the eclipse depths in point '7'
and '8' are different, but the brightness at mid-eclipse, we
P.J. Groot et al.: The Eclipsing Cataclysmic Variable GS Pavonis
3
Fig. 1. The phasefolded V-band eclipse light curves for the 12 epochs. The heliocentric corrected time of mid-eclipse
of the observations is given in UT.
conclude that at these epochs the eclipse is indeed total. It
follows that the observed minimum of V =19.9±0.1 is the
brightness of the secondary. The lack of a flat-bottom in
the light curves of point '7' and '8' shows that the eclipse is
only just total, although the integration time of 4 minutes
may be too long to resolve a flat bottom.
If we now look at the other points in Fig. 2 we see
that they do not fall on the 'Line of Totality'. With in-
creasing out-of-eclipse light levels, the depth of the eclipse
does not increase anymore (as it would have on the Line
of Totality), but decreases. Apparently, between point '7'
and '1' on the track the observed size of the disk increases
to the extent that the eclipse is no longer total, but that
part of the accretion disk remains visible in mid-eclipse.
With a further increase of the observed accretion disk ra-
dius, more and more of the disk is visible at mid-eclipse
and the eclipse becomes less and less deep. This behaviour
was first found by Walker (1963) in his study of RW Tri,
which shows no total eclipses, but does move back and
forth on this part of the track. We have therefore labeled
this the 'Walker Branch'. Another transition, which to our
knowledge has never been noted before, is the one that
happens near point '10' in Fig. 2. The out-of-eclipse light
level reaches a maximum, after which it declines again
(the curve bends to the right), but the eclipse depth con-
tinues to decrease. We have labeled this part the 'Shallow
Branch' because of its decreasing eclipse depth.
If the change in the observed radius is caused by a
change in the physical size of the disk (rather than a
change in the brightness distribution), then the manifesta-
tion of the Shallow Branch may be explained by the effect
of self-eclipses. Because the height of the disk is corre-
lated with its physical size (Frank, King and Raine 1985),
self-eclipses of the hot, and therefore luminous inner parts
of the concave disk by its outer parts, will occur when
the physical size of the disk exceeds a critical value and
therefore a critical height. The out-of-eclipse light level
decreases because the more luminous parts of the accre-
tion disk are self-eclipsed, and at the same time the eclipse
depth can continue to decrease if the disk radius continues
to increase. To eclipse the inner parts the disk flaring an-
gle must be (90-i)◦ or higher. In our case this would mean
a flaring angle of 7◦ -- 16◦, similar to what has been found
in other CVs (e.g. Robinson et al, 1995).
The variations appear to trace out a unique track over
a substantial period of time. Changes from one part of the
track to another can occur quite rapidly, e.g. the transition
from point '1' to '4' took place within 4 days. We are
P.J. Groot et al.: The Eclipsing Cataclysmic Variable GS Pavonis
4
∆ m
5
4
3
2
1
Line of Totality
9
6
1
7
8
Walker Branch
10
5
2
11
4
12
3
Shallow Branch
systems having low states in their long term light curves,
and showing no DN outbursts during these low states, we
should also classify it as a VY Scl star. This is supported
by its orbital period, since almost all known VY Scl stars
have periods between 3 and 4 hrs. However, the physical
interpretation as outlined in W95 may not apply to GS
Pav. In this description a VY Scl system in its low state
M , that is lower than the crit-
has a mass-transfer rate
Mcrit, below which DN outbursts are expected
ical rate,
to occur, but it does not show these outbursts. This dis-
tinguishes VY Scl systems from Z Cam systems, that do
show these outbursts in their low state. In our observations
GS Pav at all times seems to have an absolute magnitude
which was brighter or equal than that of Z Cam systems
M is thought to be larger than
during standstill, where
Mcrit. We therefore cannot conclude if GS Pav is a VY
Scl system or not. However, it could be that all NL sys-
tems with periods between 3 and 4 hours turn out to have
low and high states if sufficiently long observed (W95).
An interesting example of a system that may be anal-
ogous to GS Pav is VZ Scl for which O'Donoghue, Fairall
and Warner (1987) concluded that the size of the accre-
tion disk has changed from one observation to the other.
Unfortunately they only observed two eclipses. Since this
system, in the low state, is also totally eclipsing, a com-
parison with GS Pav would be of interest.
15
16
17
m V
18
19
20
Fig. 2. The depth of the eclipse in GS Pav as function
of the out-of-eclipse light. The numbers correspond to the
eclipses in Fig. 1.
currently modelling these changes in detail to constrain
the geometry of the system and the changing disk (Groot
et al., 1999).
6. Distance to the system
8. Conclusions
With the orbital period of 3.72 hr we can use the MV−Porb
relation of W95 to derive a MV = 10.4 with an esti-
mated error of 0.2 magnitudes for the secondary (from Fig.
2.46 in W95). Combined with the apparent magnitude of
V =19.9±0.1, this gives a distance to GS Pav of 790±90pc.
Given its position on the sky, the implied height above the
galactic plane is 420±60 pc. From the distance we can de-
duce the absolute magnitude of the system out of eclipse,
which is dominated by the accretion disk. To obtain an
absolute magnitude of the accretion disk, we correct for
the inclination of the disk, according to Eq. 2.63 of W95.
This correction varies from 1.0 mag (for i=74◦) to 2.1 mag
(for i=83◦). The absolute magnitude (MV) lies therefore
between 6.6 ≥ MV ≥ 4.4 (for i=74◦) and 5.5 ≥ MV ≥ 3.3
(for i=83◦). Comparison with the mean absolute magni-
tudes for NLs and DNe (Fig. 4.16 and 3.9 from W95)
shows that the derived range is in the normal regime for
NLs, but too bright for DN in quiescence.
7. Classification as a novalike system
From the shape of the light curve and its absolute magni-
tude, we conclude that GS Pav is a novalike system, and
considering its emission line spectrum (Zwitter and Mu-
nari, 1995), that it is of the RW Tri subclass, which is
defined as having emission line spectra (W95). According
to the definition in W95 of the VY Scl subclass, as NL
We have shown that GS Pav is a deeply eclipsing cata-
clysmic variable with a 3.72 hr period. Based on its photo-
metric behaviour, orbital period and absolute magnitude
we classify the system as a novalike variable. The depth of
the eclipse and the brightness of the system out-of-eclipse
are correlated. From this relation we infer that the disk
radius changes and that in two of our observations the
eclipse is total. Therefore, the apparent visual magnitude
of the secondary is V =19.9±0.1, giving a distance to the
system of 790±90 pc.
References
Augusteijn, T., 1994, PhD thesis, University of Amsterdam
Downes, R, Webbink, R.F and Shara, M.M., 1997, PASP 109,
345
Frank, J., King, A.R. and Raine, D.J., 1985, Accretion Power
in Astrophysics, Cambridge Astrophysics Series Vol. 21,
CUP
Groot, P.J., Heemskerk, M.H.M., Augusteijn, T. and Van
Paradijs, J., 1999, in preparation
Hoffmeister, C., 1963, Veroff. Sternwarte Sonneberg, 6, 1
Horne, K., 1985, MNRAS 213, 129
de Jong, J.A., Van Paradijs, J. and Augusteijn, T., A&A 314,
484
Landolt, A., 1992, AJ 104, 372
O'Donoghue, D., A.P. Fairall and Warner, B., 1987, MNRAS
225, 43
Ritter, H. and Kolb, U., 1998, A&AS 129, 83
P.J. Groot et al.: The Eclipsing Cataclysmic Variable GS Pavonis
5
Robinson, E.L., Wood, J.H, Bless, R.C. et al., 1995, ApJ 443,
295
Warner, B., 1995, Cataclysmic Variables, Cambridge Astro-
physics Series Vol. 28, CUP.
Walker, M.F., 1963, ApJ 138, 146
Zwitter, T. and Munari, U., 1995, A&AS 114, 575
|
astro-ph/0402422 | 1 | 0402 | 2004-02-18T03:38:36 | RR Lyrae Stars in the Andromeda Halo from Deep Imaging with the Advanced Camera for Surveys | [
"astro-ph"
] | We present a complete census of RR Lyrae stars in a halo field of the Andromeda galaxy. These deep observations, taken as part of a program to measure the star formation history in the halo, spanned a period of 41 days with sampling on a variety of time scales, enabling the identification of short and long period variables. Although the long period variables cannot be fully characterized within the time span of this program, the enormous advance in sensitivity provided by the Advanced Camera for Surveys on the Hubble Space Telescope allows accurate characterization of the RR Lyrae population in this field. We find 29 RRab stars with a mean period of 0.594 days, 25 RRc stars with a mean period of 0.316 days, and 1 RRd star with a fundamental period of 0.473 days and a first overtone period of 0.353 days. These 55 RR Lyrae stars imply a specific frequency S_RR=5.6, which is large given the high mean metallicity of the halo, but not surprising given that these stars arise from the old, metal-poor tail of the distribution. This old population in the Andromeda halo cannot be clearly placed into one of the Oosterhoff types: the ratio of RRc/RRabc stars is within the range seen in Oosterhoff II globular clusters, the mean RRab period is in the gap between Oosterhoff types, and the mean RRc period is in the range seen in Oosterhoff I globular clusters. The periods of these RR Lyraes suggest a mean metallicity of [Fe/H]=-1.6, while their brightness implies a distance modulus to Andromeda of 24.5+/-0.1, in good agreement with the Cepheid distance. | astro-ph | astro-ph | TO APPEAR IN THE ASTRONOMICAL JOURNAL
Preprint typeset using LATEX style emulateapj v. 04/03/99
4
0
0
2
b
e
F
8
1
1
v
2
2
4
2
0
4
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
RR LYRAE STARS IN THE ANDROMEDA HALO FROM DEEP IMAGING WITH THE ADVANCED
CAMERA FOR SURVEYS1
Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218; [email protected], [email protected],
[email protected]
THOMAS M. BROWN, HENRY C. FERGUSON, ED SMITH
RANDY A. KIMBLE, ALLEN V. SWEIGART
Code 681, NASA Goddard Space Flight Center, Greenbelt, MD 20771; [email protected], [email protected]
European Southern Observatory, Karl-Schwarzschild-Strasse 2, Garching bei Munchen, Germany; [email protected]
Division of Astronomy, Dpt. of Physics & Astronomy, UCLA, Los Angeles, CA 90095; [email protected]
R. MICHAEL RICH
ALVIO RENZINI
To appear in The Astronomical Journal
ABSTRACT
We present a complete census of RR Lyrae stars in a halo field of the Andromeda galaxy. These deep observa-
tions, taken as part of a program to measure the star formation history in the halo, spanned a period of 41 days with
sampling on a variety of time scales, enabling the identification of short and long period variables. Although the
long period variables cannot be fully characterized within the time span of this program, the enormous advance in
sensitivity provided by the Advanced Camera for Surveys on the Hubble Space Telescope allows accurate char-
acterization of the RR Lyrae population in this field. We find 29 RRab stars with a mean period of 0.594 days, 25
RRc stars with a mean period of 0.316 days, and 1 RRd star with a fundamental period of 0.473 days and a first
overtone period of 0.353 days. These 55 RR Lyrae stars imply a specific frequency SRR ≈ 5.6, which is large given
the high mean metallicity of the halo, but not surprising given that these stars arise from the old, metal-poor tail
of the distribution. This old population in the Andromeda halo cannot be clearly placed into one of the Oosterhoff
types: the ratio of RRc/RRabc stars is within the range seen in Oosterhoff II globular clusters, the mean RRab
period is in the gap between Oosterhoff types, and the mean RRc period is in the range seen in Oosterhoff I glob-
ular clusters. The periods of these RR Lyraes suggest a mean metallicity of [Fe/H]≈ −1.6, while their brightness
implies a distance modulus to Andromeda of 24.5±0.1, in good agreement with the Cepheid distance.
Subject headings: galaxies: evolution -- galaxies: stellar content -- galaxies: halos -- galaxies: individual (M31) --
stars: variables
1. INTRODUCTION
The textbook picture of a spiral galaxy halo comes from that
of our own Milky Way, which is old and metal-poor (Vanden-
Berg 2000; Ryan & Norris 1991). However, the stellar popu-
lation of the Andromeda (M31; NGC224) halo offers a strik-
ing contrast to this picture, with its wide range in metallicity
(Durrell, Harris, & Pritchet 2001) and age (Brown et al. 2003;
Brown 2003). These spreads in metallicity and age can sig-
nificantly affect the variable star population. For example, the
characteristics of RR Lyraes in Galactic globular clusters place
these clusters into two distinct Oosterhoff types (Oosterhoff
1939), while the RR Lyraes in Local Group dwarf spheroidals
(dSphs) place these galaxies in the gap between the two Ooster-
hoff types (Siegel & Majewski 2000; Dall'Ora et al. 2003; Pritzl
et al. 2002, 2004). Compared to the M31 halo, dSphs have an
even broader age range (van den Bergh 1999) and are generally
more metal-poor (Mateo 1998), but the old (> 10 Gyr) compo-
nent capable of producing RR Lyrae stars might be similar in
each case.
The specific frequency of RR Lyraes in the M31 halo has
been the subject of some debate. In a field 40 arcmin from the
1Based on observations made with the NASA/ESA Hubble Space Tele-
scope, obtained at the Space Telescope Science Institute, which is op-
erated by AURA, Inc., under NASA contract NAS 5-26555. These
observations are associated with proposal 9453.
nucleus on the southeast minor axis, Pritchet & van den Bergh
(1987) found 30 RR Lyraes, and with an estimated complete-
ness of 25%, determined that the frequency per unit luminosity
was very high (about half of that in variable-rich M3). More
recently, Dolphin et al. (2003) found only 24 RR Lyraes in a
larger field that included the Pritchet & van den Bergh (1987)
field, and with their estimated completeness of 24%, claimed
that the frequency of RR Lyrae was ∼15 times smaller. To sup-
port their claim, they estimated that the deep color magnitude
diagram (CMD) of Brown et al. (2003) contained only 10 RR
Lyraes, but as we shall show here, this was a severe underesti-
mate.
We have observed a field along the southeast minor axis of
the M31 halo, 51 arcmin from the nucleus, using the Advanced
Camera for Surveys (ACS; Ford et al. 1998) on the Hubble
Space Telescope (HST). The primary goal of this program was
to investigate the halo star formation history, by constructing
a deep CMD in the F606W (broad V ) and F814W (I) band-
passes, reaching V ≈ 30.7 mag on the main sequence (Brown
et al. 2003). However, the 250 individual exposures are scat-
tered with variable time sampling over a 41 day period, and
thus provide excellent time series photometry for the variable
star population in the M31 halo; in particular, the completeness
for RR Lyraes in our field is approximately 100%. In this pa-
per, we present a survey of the RR Lyraes and the other bright
variables in our field.
1
2
2. OBSERVATIONS AND DATA REDUCTION
Using the Wide Field Camera on the ACS, we obtained deep
optical images of a 3.5′ × 3.7′ field along the southeast minor
axis of the M31 halo, at α2000 = 00h46m07s, δ2000 = 40o42′34′′.
The surface brightness in this region is µV ≈ 26.3 mag arcsec−2
(Brown et al. 2003). The area is not associated with the tidal
streams and substructure found by Ferguson et al. (2002), and
lies just outside the "flattened inner halo" in their maps. We
placed a metal-rich M31 globular cluster, GC312 (Sargent et
al. 1977), near the edge of the field. From 2 Dec 2002 to 11
Jan 2003, we obtained 39.1 hours of images in the F606W fil-
ter (broad V ) and 45.4 hours in the F814W filter (I), with each
of the 250 exposures dithered to allow for hot pixel removal,
optimal point spread function sampling, smoothing of spatial
variations in detector response, and filling in the gap between
the two halves of the 4096 × 4096 pixel detector. Of the 250
exposures, 16 were short (< 600 s), to allow the correction of
bright saturated objects, leaving 234 long exposures (> 1200 s)
suitable for deep time-series photometry.
In order to create a deep catalog of the objects in this field,
we first co-added these images using the IRAF DRIZZLE pack-
age. This coaddition included masks for the cosmic rays and
hot pixels, and produced geometrically-correct images with
a plate scale of 0.03′′ pixel−1 and an area of approximately
210′′ × 220′′. The mask for each exposure was created in an
iterative process, comparing the value in every pixel to the dis-
tribution through the entire stack at that location on the sky;
pixels are masked on the bright end of the distribution (cosmic
rays and hot pixels) and the faint end of the distribution (dead
pixels), but this technique also masks bright, large-amplitude,
variable stars when they are near maximum or minimum. The
final coadded image in each bandpass thus approximates the av-
erage flux observed for the variable stars, but these masks can-
not be used in the time series photometry (discussed below).
We then performed both aperture and PSF-fitting photom-
etry using the DAOPHOT-II package (Stetson 1987), assum-
ing a variable PSF constructed from the most isolated stars.
The aperture photometry on isolated stars was corrected to true
apparent magnitudes using TinyTim models of the HST PSF
(Krist 1995) and observations of the standard star EGGR 102
(a V = 12.8 mag DA white dwarf) in the same filters, with
agreement at the 1% level. The PSF-fitting photometry was
then compared to the corrected aperture photometry, in order to
derive the offset between the PSF-fitting photometry and true
apparent magnitudes. Our photometry is in the STMAG sys-
tem: m = −2.5× log10 fλ − 21.1. For readers more familiar
with the Johnson V and Cousins I bandpasses, a star in the
middle of the RR Lyrae strip has V − mF606W = −0.17 mag
and I − mF814W = −1.29 mag. Light curve amplitudes are typ-
ically 6 -- 10% smaller in mF606W than in V , and 1 -- 2% smaller
in mF814W than in I. The transformation between the ACS and
ground-based bandpasses is still being characterized indepen-
dently by several groups, so for most of this paper we will refer
to magnitudes in the unambiguous STMAG system.
The CMD for the co-added images was shown by Brown et
al. (2003). In that analysis, we discarded ≈20% of the exposed
area (around bright foreground stars, near GC312, and in re-
gions with less than the full exposure time due to the dither
pattern). In the current analysis of the brighter stars, we include
the entire image area. We will only consider stars brighter than
mF814W = 28.25 mag, which have a signal-to-noise ratio of ∼ 5
in individual exposures. Extensive artificial star tests demon-
strate that our catalog, created from the deep co-added data, is
≈100% complete above this limit.
To obtain time-series photometry, we re-drizzled the entire
dataset into a stack of individual registered exposures, with
masked pixels set to an invalid data value. Because our orig-
inal masks sometimes discard data points near the maxima and
minima of variable stars, we unmasked those pixels that were
masked more than once in a two-hour window; this correction
will occasionally restore a true cosmic ray or dead pixel when
it should have been masked, but these occasional events will be
significant outliers in the time series photometry, which can be
discarded after the fact. We then performed aperture photome-
try on each individual frame, with positions fixed by the catalog
of the coadded exposures.
3. VARIABLE DETECTION AND CHARACTERIZATION
As explained above, we restricted our time-series analysis
to stars brighter than mF814W = 28.25 mag; out of the nearly
300,000 stars in the full image catalog, these 19,450 stars have
a signal-to-noise of at least 5 in individual exposures. Stars in
the RR Lyrae gap, near mF814W ≈ 26 mag, have typical photo-
metric errors of 0.03 mag in F606W exposures and 0.04 mag
in F814W exposures. Most of the stars in our search list have
∼ 100 photometric measurements in each bandpass over our 41
day observing program, but a small fraction have significantly
less because they fell near the edges of the two halves of the
detector; 424 stars, representing only 2% of the stars in the rel-
evant brightness range, were discarded because they had less
than 30 valid measurements in each bandpass. For the remain-
ing stars, we looked for variability using two methods, chosen
to suit the high signal-to-noise and large number of time sam-
ples in these data.
The first variability search was tuned to provide a complete
sample of RR Lyraes, but it also recovered a large number of
brighter, long period variables and fainter, short period vari-
ables. We used a fast algorithm (Press & Rybicki 1989) of
the Lomb-Scargle periodogram (Lomb 1976; Scargle 1982),
which looks for weak periodic signals in irregularly sampled
data. Besides providing a good initial estimate of the period,
this algorithm also quantifies the statistical significance of the
periodic signal. Our threshold was a non-random signal at 0.01
significance, independently found in each bandpass; a score
of 0.01 implies that the chance this signal arose from random
fluctuations is less than 1 percent. This stringent threshold
should only produce one or two false detections in the entire
search list. However, this threshold is not so stringent that
it misses RR Lyraes; all of the RR Lyraes recovered by this
method, regardless of amplitude or period, were detected at a
significance orders of magnitude beyond this threshold (if our
choice of threshold were in fact discarding RR Lyraes, then the
distribution of RR Lyrae scores would approach the threshold).
The initial period estimate returned by this method was then
refined by a search of the Lafler & Kinman (1965) statistic, for
periods within 5% of the initial period. Our method recovered
169 variables, two of which were false detections (each a faint
star sitting in the wings of a bright variable), and 55 of which
are clearly RR Lyraes. Of the remaining variables, 17 are short
period variables below the horizontal branch (HB), 3 have peri-
ods of 0.6 -- 7 days and lie above the HB, 1 is an eclipsing binary,
82 are long period variables (LPVs) and semiregulars near the
tip of the red giant branch (RGB), 5 are LPVs fainter than the
HB, and 4 are variables with periods of ∼1 week that lie on
the RGB. Thirteen of the detected variables, all with periods
V130 P=0.263
V89 P=0.267
V76 P=0.274
V100 P=0.275
3
mF606W
25.0
25.5
26.0
mF814W
26.5
mF606W
25.0
25.5
26.0
mF814W
26.5
mF606W
25.0
25.5
26.0
mF814W
26.5
mF606W
25.0
25.5
26.0
mF814W
26.5
mF606W
25.0
25.5
26.0
mF814W
26.5
mF606W
25.0
25.5
26.0
mF814W
26.5
mF606W
25.0
25.5
26.0
mF814W
26.5
V137 P=0.280
V37 P=0.301
V80 P=0.313
V11 P=0.331
V83 P=0.351
V54 P=0.366
V120 P=0.283
V102 P=0.301
V131 P=0.327
V43 P=0.338
V90 P=0.353
V1 P=0.382
V27 P=0.287
V8 P=0.306
V157 P=0.329
V59 P=0.339
V95 P=0.361
V147 P=0.442
V40 P=0.289
V163 P=0.313
V161 P=0.330
V5 P=0.339
V50 P=0.366
V44 P=0.464
0.5
1.0
0.5
1.0
Phase
0.5
1.0
0.5
1.0
FIG. 1 -- The light curves for RR Lyraes in our field, arranged in order of increasing period (labeled), shown as photometric error
bars (not data points). The first 26 stars are RRc and RRd stars (V90 being the RRd, shown phased with its first overtone period);
the remaining 29 are RRab stars.
longer than 20 days, fall within the tidal radius of GC312 (10′′;
Holland et al. 1997); because the area within the tidal radius of
GC312 comprises less than 1 percent of our total field, these 13
LPVs and semiregulars are clearly associated with GC312.
The second variability search simply looked for photome-
try that exceeded the expected scatter (given the photometric
errors) by 50%. This method was motivated by a search for
halo-on-halo microlensing events. We found no microlensing
events or new pulsating variables, but we did find 7 additional
eclipsing binaries. As might be expected, the Lomb-Scargle
periodogram is better suited to finding variables with roughly
sinusoidal light curves. We show the light curves for the RR
4
mF606W
25.0
25.5
26.0
mF814W
26.5
mF606W
25.0
25.5
26.0
mF814W
26.5
mF606W
25.0
25.5
26.0
mF814W
26.5
mF606W
25.0
25.5
26.0
mF814W
26.5
mF606W
25.0
25.5
26.0
mF814W
26.5
mF606W
25.0
25.5
26.0
mF814W
26.5
mF606W
25.0
25.5
26.0
mF814W
26.5
V47 P=0.496
V88 P=0.506
V79 P=0.529
V28 P=0.532
V162 P=0.533
V142 P=0.554
V164 P=0.580
V167 P=0.621
V136 P=0.634
V114 P=0.726
V126 P=0.534
V140 P=0.559
V166 P=0.583
V124 P=0.627
V77 P=0.678
V82 P=0.735
V42 P=0.552
V133 P=0.572
V122 P=0.589
V36 P=0.627
V10 P=0.687
V57 P=0.554
V112 P=0.573
V160 P=0.611
V123 P=0.631
V66 P=0.714
V78 P=0.774
0.5
1.0
0.5
1.0
Phase
0.5
1.0
FIG. 1 -- Continued.
Lyraes in Figure 1, light curves for the variables above and be-
low the HB in Figure 2, light curves for the eclipsing binaries in
Figure 3, and light curves for a subset of the LPVs and semireg-
ulars in Figure 4 (only those that varied by more than ∼ 0.1 mag
within the 41 days of observations). Note that only photometric
error bars are visible at the plotted scale of these light curves.
Five of the RR Lyrae light curves show significant scatter
(V1, V90, V95, V122, and V163). We reevaluated the peri-
odicity searches on these five stars, and found that one of the
stars, V90, showed clear evidence for double-mode pulsations,
making it a definite RRd star, with a fundamental period (P0)
of 0.4735 days and a first overtone period (P1) of 0.3534 days.
V163 shows a hint of an additional period at 0.4243 days,
but the signal is weak, and we found no additional periods
22.0
V84 P=0.611
21.0
V118 P=2.010
22.0
V60 P=7.135
5
mF606W
mF814W
mF606W
mF814W
mF606W
mF814W
mF606W
mF814W
mF606W
mF814W
mF606W
mF814W
mF606W
mF814W
24.5
25.0
27.5
26.5
29.0
26.0
28.5
26.0
28.5
26.0
28.5
24.5
27.0
23.5
25.5
28.0
26.5
29.0
26.0
28.5
26.0
28.5
25.0
27.5
26.5
29.0
V48 P=0.112
V105 P=0.115
V119 P=0.226
24.5
26.5
29.0
24.5
27.0
26.0
28.5
25.5
28.0
26.0
28.5
V113 P=0.230
V94 P=0.252
V6 P=0.267
V65 P=0.309
V12 P=0.317
V93 P=0.338
V63 P=0.351
V128 P=0.353
V91 P=0.367
V99 P=0.390
V85 P=0.407
V74 P=0.439
V29 P=0.573
V117 P=0.680
0.5
1.0
0.5
1.0
Phase
FIG. 2 -- The light curves for other variables bluer than the RGB (excluding eclipsing binaries), shown as error bars. The stars in
the top 3 panels are brighter than the HB; they are arranged in order of increasing period and increasing mF606W − mF814W color,
making them easy to identify in the CMD (Figure 5). The stars in the lower 17 panels are likely dwarf Cepheids fainter than the
HB, but V6 and V29, which lie immediately below the HB, might be RR Lyraes.
in the light curves of V1 and V95. Although V1, V95, and
V163 might be RRd stars, we have classified them as RRc. The
scatter in the light curve of V122 might be due to the Blazhko
effect -- a secondary modulation in the variability of 20 -- 30% of
RRab stars, with a period of ∼10 -- 500 days (Smith et al. 2003).
To determine the mean magnitudes and amplitudes of the
variables with periods less than ∼ 10 days, we fit a Fourier se-
ries to the magnitudes in each bandpass as a function of phase,
with an order of 1 or 2 for the sinusoidal light curves (e.g.,
RRcd stars) and an order of 8 for the sawtooth light curves (e.g.,
6
26.5
V168 P=0.811
26
V172 P=0.913
26.5
V170 P=0.976
25
V173 P=1.245
mF606W
mF814W
29.5
25.5
mF606W
mF814W
28.5
V169 P=1.550
V171 P=1.806
29
25.5
28.5
0.5 1.0
0.5 1.0
29.5
25.5
28.5
Phase
V174 P=2.254
V64 P=3.037
28
25
28
0.5 1.0
0.5 1.0
FIG. 3 -- The light curves for candidate eclipsing binaries in our field. The most secure (V64) was found through the Lomb-Scargle
periodogram search, while the others were found because their photometric errors were larger than expected. Note that a secondary
minimum is sometimes visible at a phase opposite to the primary minimum.
RRab stars). Table 1 lists the properties of the RR Lyraes in
our field, and Table 2 gives their time-series photometry (with
the full dataset available only as a machine-readable table in the
electronic version of this paper). Table 3 gives the positions of
the other variables in our field (Figures 2 -- 4). Figure 5 shows
the CMD locations of the stars with light curves (Figures 1 -- 4).
The variables are shown at their mean mF606W − mF814W color
and mF814W magnitude, as determined from the fitting above,
instead of the average observed value in the catalog from the
deep co-added images, which can be systematically brighter or
fainter in a given bandpass by a few hundredths of a magnitude,
depending upon the random sampling of each light curve. Fig-
ure 6 shows the distribution of mF606W amplitude vs. period,
mean mF606W vs. period, and the period distribution for the RR
Lyrae stars. The RR Lyraes are cleanly separated from the re-
maining variables by their position in the mF606W vs. period
diagram, although two stars lying immediately below the HB
might also be RR Lyraes (V6 and V29 in Figure 2). The RR
Lyraes can be classified as either RRab or RRcd by the gap in
amplitude vs. period distribution; these classes do not overlap
in the CMD (Figure 5).
4. RR LYRAE PROPERTIES
4.1. Oosterhoff Type
The RR Lyrae population of the M31 halo cannot be classi-
fied into either of the Oosterhoff (1939) types, but this is not
because the various characteristics appear as an average of the
two types (Figure 7). Looking at the ensemble variable popu-
lation in Galactic globular clusters, the ratio of RRc to RRabc
stars is 0.22 in Oosterhoff I clusters and 0.48 in Oosterhoff II
clusters (Clement et al. 2001), while in the M31 halo it is 0.46
-- a value in the Oosterhoff II regime. In contrast, looking at
the mean period of the RRc stars in Galactic globular clusters,
it is 0.326 days in Oosterhoff I clusters and 0.368 days in Oost-
erhoff II clusters (Clement et al. 2001), while in the M31 halo
it is 0.316 days, within the Oosterhoff I regime. The mean pe-
riod for the known RRab stars in Oosterhoff I clusters is 0.559
days, and in Oosterhoff II clusters it is 0.659 days (Clement
et al. 2001), while in the M31 halo it is 0.594 days -- midway
between the two types. The RRab period distribution is more
sharply peaked at this intermediate value than one might ex-
pect from an equal mix of Oosterhoff types (see Figure 8), but
a Kolmogorov-Smirnov test shows the difference is not statis-
tically significant. The period-amplitude diagram for the RRab
stars is also suggestive of an intermediate Oosterhoff type (Fig-
ure 9). If we convert our mF606W amplitudes to V amplitudes,
and compare against the period-amplitude relations of Ooster-
hoff I and Oosterhoff II clusters (Clement 2000), the RRab stars
in M31 fall between the two types, although there is a tendency
toward Oosterhoff I at the larger amplitudes.
Although the Galactic halo field is more difficult to charac-
terize than the Galactic clusters, it is interesting to note that the
field RR Lyraes also show the Oosterhoff dichotomy, with a real
gap in the distribution of period shifts at fixed amplitude with
respect to the M3 RR Lyraes (see, e.g., Figure 8 of Suntzeff,
Kinman, & Kraft 1991). However, these same data show that
the Galactic field population is predominantly of Oosterhoff
type I, as has been known for some time. In their study of the
Palomar-Groningen Survey, Cacciari & Renzini (1976) found
<Pab>= 0.529 days, <Pc>= 0.329 days, and Nc/Nabc = 0.09.
Selection effects in a field survey might favor RR Lyraes with
higher amplitudes (thus reducing the relative number of RRc
stars and long-period RRab stars), but these characteristics are
all well within the Oosterhoff I regime. In contrast, the cor-
responding values for the M31 halo RR Lyraes are quite dif-
ferent. In particular, the mean RRab period is much longer in
M31 (<Pab>= 0.594 days), and the frequency of RRc variables
is much higher (Nc/Nabc = 0.46). Most importantly, the distri-
bution of period shifts of the M31 RR Lyraes with respect to
the M3 RR Lyraes (i.e., the Oosterhoff I line in Figure 9) shows
no such gap. All of these characteristics clearly distinguish the
M31 halo field from the Milky Way halo field.
The characteristics of the M31 halo variables do not track
those of the Local Group dSphs, either, except in the broad
sense that the M31 halo cannot be put cleanly into one of
the Oosterhoff types (Figure 7). For dSphs, the mean period
of RRab stars tends to fall in the gap between Oosterhoff types
7
FIG. 4 -- The light curves for a subset of the bright LPVs and semiregulars in our field (those that varied by more than 0.1 mag in
each bandpass), arranged roughly in order of increasing period (the 41 day span of our observations limits our ability to characterize
the periods).
(Siegel & Majewski 2000; Pritzl et al. 2002, 2004; Dall'Ora et
al. 2003 and references therein), as found in M31. However, the
fraction of RR Lyrae stars that are RRc is higher in M31 (typi-
cal of Oosterhoff II) than that in any of the dSphs (which tend
toward the Oosterhoff I values), while the mean period of RRc
stars is much lower in M31 (typical of Oosterhoff I) than in any
of the dSphs (which tend to fall in the gap between Oosterhoff
types). These differences might suggest that the M31 halo is
not comprised of dissolved globular clusters like those in the
Milky Way, or dissolved Local Group dSphs. Note that globular
clusters in the Large Magellanic Cloud are also predominantly
of intermediate Oosterhoff type, in contrast to the clusters of the
8
FIG. 4 -- Continued.
Milky Way (Bono, Caputo, & Stellingwerf 1994).
It has been argued that the Oosterhoff dichotomy among
Galactic globular clusters is due to a real gap in metallicity
between the two Oosterhoff types, with clusters at metallici-
ties between the two Oosterhoff types having very blue HBs
that do not populate the RR Lyrae instability strip (Renzini
1983; Sandage 1993). Local Group dSphs and LMC globular
clusters fill the gap in a plot of [Fe/H] vs. <Pab>, as illustrated
by Siegel & Majewski (2000) and Pritzl et al. (2004) (see Fig-
ure 6 in each paper). This indicates that the dSphs and Galactic
globular clusters occupy separate regions of the age vs. chem-
ical composition parameter space (Renzini 1980), which in the
simplest option reduces to the [Fe/H], age, and [α/Fe] three-
dimensional space. The M31 halo appears to bridge the gap
in a manner similar to the dSphs, signaling that the early star
formation history in the M31 halo proceeded at a different pace
Name
V1
V5
V8
V10
V11
V27
V28
V36
V37
V40
V42
V43
V44
V47
V50
V54
V57
V59
V66
V76
V77
V78
V79
V80
V82
V83
V88
V89
V90
V95
V100
V102
V112
V114
V120
V122
V123
V124
V126
V130
V131
V133
V136
V137
V140
V142
V147
V157
V160
V161
V162
V163
V164
V166
V167
R.A.
(J2000)
0h45m59.76s
0h46m 0.61s
0h46m 1.02s
0h46m 0.68s
0h45m56.88s
0h45m57.64s
0h46m 0.08s
0h46m 2.04s
0h46m 3.26s
0h46m 1.20s
0h46m 0.77s
0h46m 1.74s
0h45m59.93s
0h46m 1.55s
0h46m 2.68s
0h46m 4.33s
0h46m 2.60s
0h46m 3.80s
0h46m 4.71s
0h46m 5.86s
0h46m 4.22s
0h46m 8.34s
0h46m 4.60s
0h46m 5.78s
0h46m 5.29s
0h46m 8.84s
0h46m 9.86s
0h46m 6.68s
0h46m 7.71s
0h46m 9.60s
0h46m 6.94s
0h46m 8.57s
0h46m11.88s
0h46m10.11s
0h46m12.78s
0h46m13.10s
0h46m12.46s
0h46m 9.88s
0h46m10.38s
0h46m11.91s
0h46m13.72s
0h46m13.43s
0h46m10.23s
0h46m11.13s
0h46m14.51s
0h46m11.47s
0h46m11.66s
0h46m16.21s
0h46m14.30s
0h46m13.32s
0h46m16.08s
0h46m15.92s
0h46m14.21s
0h46m13.71s
0h46m10.38s
Dec.
(J2000)
40o41′18.2′′
40o40′52.4′′
40o40′44.1′′
40o41′ 2.9′′
40o43′44.1′′
40o43′33.5′′
40o42′ 2.9′′
40o41′29.0′′
40o40′39.9′′
40o42′19.8′′
40o43′ 0.9′′
40o42′20.6′′
40o43′38.9′′
40o42′44.8′′
40o42′31.0′′
40o41′35.2′′
40o43′10.6′′
40o42′36.9′′
40o42′32.6′′
40o42′52.9′′
40o44′ 5.8′′
40o41′24.6′′
40o44′ 9.8′′
40o43′28.8′′
40o44′ 2.4′′
40o41′35.5′′
40o41′ 3.5′′
40o43′21.9′′
40o42′51.1′′
40o41′39.8′′
40o43′58.7′′
40o43′10.5′′
40o41′21.5′′
40o42′54.5′′
40o41′24.6′′
40o41′14.2′′
40o41′42.2′′
40o43′34.7′′
40o43′32.5′′
40o42′45.1′′
40o41′30.6′′
40o41′54.5′′
40o44′23.2′′
40o43′48.8′′
40o41′37.0′′
40o44′ 4.1′′
40o44′23.3′′
40o41′47.6′′
40o43′22.7′′
40o44′18.7′′
40o42′25.4′′
40o42′37.9′′
40o43′55.8′′
40o44′24.4′′
40o43′44.0′′
TABLE 1
RR LYRAE PROPERTIES
Period <mF606W > <mF814W > AF606W
(mag)
(days)
0.42
0.382
0.339
0.40
0.37
0.306
0.97
0.687
0.40
0.331
0.287
0.49
0.85
0.532
1.06
0.627
0.46
0.301
0.289
0.15
1.05
0.552
0.56
0.338
1.07
0.464
1.09
0.496
0.366
0.45
0.38
0.366
1.03
0.554
0.43
0.339
0.714
0.51
0.15
0.274
0.63
0.678
0.49
0.774
0.529
1.12
0.42
0.313
0.47
0.735
0.52
0.351
0.506
1.07
0.41
0.267
0.50
0.353
0.36
0.361
0.48
0.275
0.301
0.46
0.95
0.573
0.33
0.726
0.52
0.283
0.589
1.09
0.67
0.631
0.75
0.627
0.96
0.534
0.263
0.48
0.38
0.327
0.88
0.572
0.47
0.634
0.34
0.280
0.559
1.12
0.91
0.554
1.12
0.442
0.40
0.329
0.611
0.68
0.37
0.330
1.01
0.533
0.37
0.313
0.580
0.85
0.65
0.583
0.621
0.67
(mag)
25.38
25.55
25.61
25.17
25.64
25.65
25.59
25.32
25.48
25.40
25.35
25.39
25.67
25.57
25.47
25.42
25.42
25.46
25.37
25.42
25.52
25.38
25.51
25.52
25.48
25.36
25.56
25.53
25.49
25.42
25.59
25.52
25.41
25.30
25.39
25.36
25.27
25.35
25.48
25.66
25.50
25.37
25.56
25.60
25.38
25.41
25.52
25.47
25.26
25.56
25.57
25.35
25.45
25.46
25.43
9
class
c
c
c
ab
c
c
ab
ab
c
c
ab
c
ab
ab
c
c
ab
c
ab
c
ab
ab
ab
c
ab
c
ab
c
d
c
c
c
ab
ab
c
ab
ab
ab
ab
c
c
ab
ab
c
ab
ab
ab
c
ab
c
ab
c
ab
ab
ab
AF814W
(mag)
0.30
0.29
0.29
0.66
0.28
0.32
0.70
0.73
0.33
0.10
0.71
0.36
0.79
0.80
0.31
0.26
0.68
0.30
0.39
0.11
0.43
0.42
0.77
0.28
0.37
0.35
0.80
0.28
0.26
0.26
0.35
0.31
0.62
0.24
0.35
0.52
0.50
0.52
0.68
0.33
0.22
0.64
0.32
0.22
0.84
0.65
0.68
0.29
0.53
0.27
0.78
0.25
0.64
0.59
0.47
(mag)
25.99
26.17
26.27
25.72
26.25
26.32
26.15
25.91
26.17
26.14
25.91
26.04
26.28
26.15
26.13
26.05
25.99
26.12
25.87
26.14
26.04
25.92
26.07
26.21
25.99
26.03
26.15
26.26
26.11
26.05
26.29
26.21
26.00
25.80
26.11
25.96
25.82
25.89
26.05
26.37
26.15
25.95
26.06
26.27
25.96
25.97
26.10
26.16
25.81
26.18
26.15
26.00
26.01
25.99
25.95
10
MJD
(days)
TABLE 2
RR LYRAE PHOTOMETRYa
Photometry (mag)b
Band
V1
V8
25.87
F606W 25.58
F606W 25.27
25.57
25.48
F606W 99.99
25.51
F606W 25.22
26.39
F814W 26.09
F814W 25.98
99.99
26.47
F814W 25.96
26.37
F814W 25.75
25.42
F606W 25.29
F606W 25.37
25.47
V5
25.75
25.57
25.42
25.44
26.14
26.07
26.09
26.19
25.74
25.83
52610.06377
52610.12537
52610.19249
52610.25919
52611.20557
52611.25942
52611.27610
52611.32653
52611.39623
52611.41250
aTable 2 is published in its entirety in the electronic edition of The
Astronomical Journal. A portion is shown here for guidance re-
garding its form and content.
bA magnitude of 99.99 signifies missing data.
with respect to the Galactic halo, therefore resulting in different
age vs. chemical composition patterns.
4.2. Specific Frequency
The specific frequency of RR Lyraes in the M31 halo has
been the subject of debate, due perhaps in part to the diffi-
culty of observing these stars from the ground. Pritchet & van
den Bergh (1987) observed a 2.2′ × 3.3′ field 40′ from the nu-
cleus on the southeast minor axis, and found 30 RR Lyraes
with amplitudes ∼>0.7 mag. Assuming a completeness of 25%,
they determined there should be 120 RR Lyraes with ampli-
tudes ∼>0.7 mag. They took this to be a very strict lower limit,
given the lack of low-amplitude RR Lyraes, and claimed their
value was 55% of the frequency for M3 (NGC5272), a Galac-
tic globular cluster with a high frequency of RR Lyraes. The
specific frequency of RR Lyraes in M3, as given by Harris
(1996) and normalized to a total cluster luminosity (MV t) of
−7.5 mag, is SRR = NRR10(MVt +7.5)/2.5 = 49. Thus, Pritchet
& van den Bergh (1987) determined SRR ≈ 27, which was a
surprisingly high number, given the mean metallicity [Fe/H]=
−0.6 of Mould & Kristian (1986).
Recently, Dolphin et al. (2003) observed a 9.6′ × 9.6′ field
that included the Pritchet & van den Bergh (1987) field, and
found a much lower frequency of RR Lyraes. After finding 24
RR Lyraes and estimating a completeness of 24%, they esti-
mated that there are 100 RR Lyraes in their field. Given the
difference in field size, the Dolphin et al. (2003) frequency is
15 times smaller than that of Pritchet & van den Bergh (1987),
i.e., SRR ≈ 1.8. Dolphin et al. (2003) claimed that their lower
frequency was supported by the CMD of Brown et al. (2003)
-- the same data we are using in this paper -- because they es-
timated that there were only 10 RR Lyraes in the Brown et al.
(2003) data. However, the CMD shown by Brown et al. (2003)
was displayed as a greyscale Hess diagram, in order to clearly
show the characteristics of the main sequence turnoff in a cat-
alog of ∼ 300,000 stars; individual stars were not shown, nor
was variability indicated in any way. In reality, 41 of the 55 RR
Lyraes reported in our current work also appeared in the CMD
of Brown et al. (2003); the remaining 14 were in the ∼ 20%
of the image area masked for that earlier analysis of the star
formation history (see §2).
These earlier estimates of the RR Lyrae frequency were
limited by several factors. The first was uncertainty in the
completeness. RR Lyraes were at the edge of detection in
both the Dolphin et al. (2003) and Pritchet & van den Bergh
(1987) studies, and neither group quantified the completeness
at that depth through artificial star tests. Furthermore, the total
luminosity in these fields could be estimated only from photo-
graphic plates or counts of the bright RGB stars. In contrast,
our artificial star tests indicate that the completeness in our data
POSITIONS OF OTHER VARIABLES
Name
V6
V7
V12
V14
V18
V19
V20
V21
V25
V29
V31
V38
V39
V41
V48
V51
V52
V60
V61
V63
R.A.
(J2000)
0h45m57.91s
0h45m58.14s
0h45m58.06s
0h45m58.78s
0h45m58.87s
0h45m58.90s
0h46m 1.02s
0h45m58.90s
0h45m58.97s
0h46m 1.76s
0h46m 2.46s
0h46m 1.79s
0h46m 0.08s
0h46m 1.07s
0h46m 2.27s
0h46m 1.59s
0h46m 2.51s
0h46m 4.91s
0h46m 5.72s
0h46m 5.43s
Dec.
(J2000)
Name
40o42′48.1′′
V64
40o42′38.6′′
V65
40o42′55.2′′
V67
40o42′29.7′′
V70
40o42′32.3′′
V72
40o42′30.9′′
V74
40o41′ 2.1′′
V84
40o42′32.7′′
V85
40o42′32.0′′
V86
40o40′53.1′′
V91
40o40′41.3′′
V93
40o41′44.0′′
V94
40o42′57.0′′
V99
40o42′44.9′′ V103
40o42′16.3′′ V105
40o43′18.1′′ V106
40o42′39.2′′ V111
40o41′55.9′′ V113
40o41′24.2′′ V116
40o41′55.9′′ V117
TABLE 3
R.A.
(J2000)
0h46m 3.27s
0h46m 4.06s
0h46m 4.42s
0h46m 3.88s
0h46m 6.25s
0h46m 4.44s
0h46m 5.30s
0h46m 9.12s
0h46m 9.55s
0h46m 7.74s
0h46m 9.40s
0h46m 9.77s
0h46m 6.54s
0h46m 8.70s
0h46m 9.32s
0h46m10.84s
0h46m11.28s
0h46m 8.95s
0h46m12.44s
0h46m11.67s
Dec.
(J2000)
Name
40o43′29.4′′ V118
40o42′59.2′′ V119
40o42′46.9′′ V127
40o43′28.5′′ V128
40o41′53.8′′ V129
40o43′25.7′′ V134
40o44′ 5.8′′ V139
40o41′24.7′′ V143
40o41′ 8.5′′ V145
40o42′50.6′′ V150
40o41′45.2′′ V151
40o41′29.9′′ V158
40o44′ 8.0′′ V159
40o43′10.2′′ V168
40o42′53.2′′ V169
40o41′51.5′′ V170
40o41′42.7′′ V171
40o43′29.5′′ V172
40o41′20.8′′ V173
40o41′54.6′′ V174
R.A.
(J2000)
0h46m10.05s
0h46m11.23s
0h46m11.02s
0h46m12.17s
0h46m 9.88s
0h46m13.50s
0h46m11.90s
0h46m14.94s
0h46m14.38s
0h46m12.63s
0h46m13.16s
0h46m13.67s
0h46m14.91s
0h46m 0.63s
0h46m 2.52s
0h46m 8.72s
0h46m 7.35s
0h46m 9.93s
0h46m10.45s
0h46m 8.98s
Dec.
(J2000)
40o43′ 6.0′′
40o42′19.0′′
40o43′ 5.9′′
40o42′18.4′′
40o43′57.4′′
40o41′53.2′′
40o43′17.9′′
40o41′42.8′′
40o42′13.8′′
40o43′48.2′′
40o43′36.0′′
40o43′37.6′′
40o42′51.6′′
40o41′13.8′′
40o40′37.7′′
40o41′53.0′′
40o44′17.1′′
40o43′18.3′′
40o43′ 1.6′′
40o43′ 9.2′′
11
0
1
2
3
4
22
24
26
)
G
A
M
T
S
(
W
4
1
8
F
m
28
-1
25.5
26.0
26.5
27.0
27.5
28.0
)
G
A
M
T
S
(
W
4
1
8
F
m
-1.0
-0.8
-0.6
mF606W-mF814W (STMAG)
-0.4
-0.2
FIG. 5 -- Top panel: Subset of the M31 halo CMD of Brown et al. (2003), highlighting the bright variables. From Figure 2, V84
is marked by a closed circle, while V118 and V60 are marked by closed diamonds. The variables of Figure 4 are shown by open
diamonds. Bottom panel: Expanded view of the faint variables: RRab stars (crosses), RRc stars (open triangles), an RRd star
(closed triangle), dwarf Cepheids (open circles), and eclipsing binaries (closed squares).
is ≈100% at the horizontal branch, our time sampling enables
a search for periodicity with high statistical significance, and
our deep catalog allows an accurate determination of the to-
tal luminosity in our field. Brown et al. (2003) determined
µV ≈ 26.3 mag arcsec−2 in this field. Assuming an extinc-
tion of E(B −V ) = 0.08 ± 0.03 mag (Schlegel et al. 1998) and
a distance modulus (m − M)0 = 24.44 ± 0.1 mag (Freedman
& Madore 1990), we estimate a total brightness in our field
of MV ≈ −10.0 mag, and SRR ≈ 5.6 (∼ 1 RR Lyrae star per
1.5 × 104L⊙). Our frequency is ∼3 times higher than that
found by Dolphin et al. (2003), with the number of RR Lyraes
compared either to the number of RGB stars (the metric of
Dolphin et al. 2003) or the total luminosity of the field (the
metric of Harris 1996). The specific frequency we find is much
higher than one would expect for a population with metallic-
ity [Fe/H]= −0.8 (Figure 10), but not surprising given that the
M31 halo has a wide metallicity distribution, with ∼40% of
the stars at −2.5 <[Fe/H]< −1 (Durrell et al. 2001). Because
about half of the M31 halo is metal-rich and of intermediate age
(Brown et al. 2003), and thus incapable of producing RR Lyrae
stars, one could reasonably shift the location of the M31 data
point in Figure 10 to a metallicity near [Fe/H]= −1.6 and to a
frequency twice as large as that found at [Fe/H] = −0.8.
4.3. Metallicity
Given a larger number of RR Lyraes than expected for a
mean metallicity of [Fe/H]= −0.8, it is reasonable to assume
that the RR Lyraes come from the old, metal-poor component
of the M31 halo. As done in studies of dSph populations (e.g.,
Pritzl et al. 2002; Siegel & Majewski 2000), we can use one
of several empirical relations to estimate the metallicity of the
RR Lyraes in our field. Looking at the mean properties of the
0.6
0.5
0.4
0.3
0.2
0.1
c
b
a
N
/
c
N
0.0
-0.40
-0.42
-0.44
-0.46
-0.48
-0.50
-0.52
-0.54
c
>
P
<
g
o
l
-0.28
-0.24
-0.20
log <Pab>
-0.16
FIG. 7 -- Top panel: The fractions of RRab+RRc stars that are RRc,
vs. the mean RRab periods, for different populations. Galactic glob-
ular clusters (Clement et al. 2001) of Oosterhoff I (open circles) and
Oosterhoff II (closed circles) types are well separated, while Local
Group dSphs (diamonds; Pritzl et al. 2002, 2004; Dall'Ora et al.
2003) and the M31 halo (asterisk) bridge the gap; the Milky Way
halo field is, on average, Oosterhoff I (cross; Cacciari & Renzini
1976). The fraction of RRc stars in the M31 halo is like that in Oost-
erhoff II clusters, while the fraction of RRc stars in dSphs is like that
in Oosterhoff I clusters. Bottom panel: The mean RRc period vs.
the mean RRab period. Again, the dSphs and the M31 halo bridge
the gap between Oosterhoff types, but the mean RRc period in M31
is like that in Oosterhoff I clusters, while the mean RRc period in
dSphs is like that in Oosterhoff II clusters.
12
)
G
A
M
T
S
(
W
6
0
6
F
A
)
G
A
M
T
S
(
W
6
0
6
F
m
s
r
a
t
S
e
a
r
y
L
R
R
1.2
1.0
0.8
0.6
0.4
0.2
0.0
23
24
25
26
27
28
10
8
6
4
2
Ncd / Nabcd = 0.47
Nc / Nabc = 0.46
6
1
3
.
0
=
>
P
<
c
4
9
5
.
0
=
>
b
a
P
<
0
-1.0
-0.8
-0.6
-0.4
-0.2
Log Period (days)
-2
-1
)
g
0
a
m
(
V
M
1
2
3
0.0
FIG. 6 -- Top panel: The amplitude vs. period diagram for short
period variables (excluding the eclipsing binaries). The RRab stars
(crosses) are well-separated from the RRc stars (open triangles), but
the dwarf Cepheids (open circles) are mixed in with the RR Lyraes.
The RRd star (closed triangle) is mixed in the main RRc clump. The
bright blue variable above the HB is also marked (closed circle).
Middle panel: The luminosity vs. period diagram, with the same
symbols. The RR Lyraes are well-separated from the other variables,
although two "dwarf Cepheids" might actually be RR Lyraes. The
left axis provides the observed F606W magnitudes, while the right
gives the transformation to MV appropriate for the color of the center
of the RR Lyrae gap, assuming the Cepheid distance to M31 (see §4.6
for details). Bottom panel: The period histogram for the RR Lyraes.
The parameters used to distinguish Oosterhoff type are labeled.
7 M31
6
5
4
3
2
1
0
7 M3+M15 (normalized)
6
5
4
3
2
1
0
40
M3 - OoI
(a)
(b)
(c)
)
g
a
m
(
V
A
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
-0.4
13
0.0
OoII
OoI
-0.3
-0.2
-0.1
Log Period (days)
FIG. 9 -- The period-amplitude diagram for the RRab stars of the M31
halo (crosses), compared to the period-amplitude relations for Ooster-
hoff I and Oosterhoff II clusters (lines; Clement 2000). Note that we
converted amplitudes in the mF606W bandpass to amplitudes in John-
son V using Lejeune synthetic spectra, but this correction is small (a
6 -- 10% increase in the F606W amplitude). The M31 halo stars appear
to be an intermediate case between Oosterhoff types, although there is
a tendency toward the Oosterhoff I type at large amplitudes.
4.4. Double-Mode Pulsators
As mentioned in §2, five RR Lyrae stars in our sample
show significant scatter in their light curves, as expected for
double-mode pulsators (RRd stars) and RRab stars exhibiting
the Blazhko effect. One of these stars, V90, shows clear pe-
riodicity at two different periods, with P0 = 0.4735 days and
P1 = 0.3534 days. The ratio P1/P0 is sensitive to the mass of the
RRd star, with RRd stars from the two Oosterhoff types clearly
separated in a P1/P0 vs P0 diagram (see Bono et al. 1996 and
references therein). Using these periods to place our confirmed
RRd star in the diagrams of Bono et al. (1996), we find that the
P0 for this star is similar to that of RRd stars found in Oosterhoff
I clusters, but the P1/P0 ratio is typical for RRd stars found in
Oosterhoff II clusters, with a mass of ≈ 0.75 M⊙. It is curious
that an individual star in the M31 halo cannot be placed into
either Oosterhoff type in such a diagram; again, this suggests
that the M31 halo population is not a simple mix of Oosterhoff
types.
4.5. Distance
With the metallicities derived above, we can estimate the
distance to the M31 halo population using the relation of Car-
retta et al. (2000): MV = (0.18 ± 0.09)([Fe/H]+ 1.5) + (0.57 ±
0.07). Note that this relation is consistent with the average of
the methods discussed by Cacciari & Clementini (2003), who
give an excellent review of distance determination using RR
Lyrae stars. Using the synthetic spectra of Lejeune et al. (1997),
an assumed extinction of E(B − V ) = 0.08 ± 0.03 (Schlegel
et al. 1998), and the extinction curve of Fitzpatrick (1999),
we can calculate the offset between mF606W and V as a func-
tion of mF606W − mF814W . The RRab stars have a color range
−0.61 ≤ (mF606W − mF814W ) ≤ −0.5 mag, with correspond-
ing corrections of −0.17 ≤ (V − mF606W ) ≤ −0.13 mag; the
b
a
R
R
f
o
r
e
b
m
u
N
30
20
10
0
10 M15 - OoII
8
6
4
2
0
-0.40 -0.35 -0.30 -0.25 -0.20 -0.15 -0.10
(d)
Log Period (Days)
FIG. 8 -- Panel (a): The period distribution of RRab stars in the M31
halo. Panel (b): The period distribution of RRab stars when the stars
are drawn equally from M3 and M15 -- two clusters rich in RR Lyraes
with distinct Oosterhoff types. Panel (c): The period distribution in
M3 (Clement et al. 2001). Panel (d): The period distribution in M15
(Clement et al. 2001).
RRab and RRc stars, the period-metallicity relations of Sandage
(1993) give [Fe/H] = ( -- log<Pab> −0.389)/0.092 = −1.77,
and [Fe/H] = ( -- log<Pc> −0.670)/0.119 = −1.43, on the Zinn
& West (1984) scale. Given the ensemble population in the
M31 halo and the scatter for these relations shown by Sandage
(1993), the difference in metallicity for the RRab and RRc stars
is not significant, but we can say that the RR Lyraes lie in the
metal-poor tail of the halo metallicity distribution. This finding
is further supported by the period-amplitude-metallicity rela-
tion of Alcock et al. (2000), which provides metallicities on the
Zinn & West (1984) scale for individual RRab stars: [Fe/H]= --
8.85(logPab + 0.15AV) -- 2.60. Converting our amplitudes in the
mF606W bandpass to those in Johnson V , we find a mean [Fe/H]
of -- 1.79, with a standard deviation of 0.32; the accuracy of this
method is σ[Fe/H] = 0.31 per star (Alcock et al. 2000), so the
dispersion in our individual metallicities is not statistically sig-
nificant.
14
y
c
n
e
u
q
e
r
F
c
i
f
i
c
e
p
S
e
a
r
y
L
R
R
140
120
100
80
60
40
20
0
-2.0
-1.5
-1.0
[Fe/H]
-0.5
0.0
FIG. 10 -- The specific frequency of RR Lyraes, normalized to a pop-
ulation at MVt = −7.5 mag, as a function of metallicity, for Galactic
globular clusters (points; Harris 1996) and the M31 halo (asterisk).
Although the frequency of RR Lyraes in the M31 halo is higher than
expected for the mean halo metallicity, the RR Lyraes have a metal-
licity near [Fe/H]= −1.6. Considered as part of the old metal-poor
population, the frequency in the M31 halo is not surprising.
RRc stars have a color range −0.74 ≤ (mF606W − mF814W ) ≤
−0.61 mag, with corresponding corrections of −0.21 ≤ (V −
mF606W ) ≤ −0.17 mag. After correcting the individual stars
with the offsets appropriate to their colors, the RRab stars
lie at <mF606W >= 25.43 mag and <V >= 25.28 ± 0.01 mag,
while the RRc stars lie at <mF606W >= 25.49 mag and <V >=
25.31 ± 0.01 mag (the statistical errors on <V > are much
smaller than 0.01 mag, but the throughputs of the ACS filters
are only known to ∼1%). Assuming [Fe/H]= −1.77, the Car-
retta et al. (2000) relation yields MV = 0.52 ± 0.07 mag for
the RRab stars; assuming [Fe/H]= −1.43 gives MV = 0.58 ±
0.07 mag for the RRc stars. With an extinction of AV = 0.25 ±
0.09, we have distance moduli of (m − M)0 = 24.51 ± 0.11 mag
for the RRab stars, and (m − M)0 = 24.48 ± 0.11 mag for the
RRc stars. Averaging these results gives an RR Lyrae distance
modulus of (m − M)0 = 24.5 ± 0.1 mag, which is in very good
agreement with the Cepheid distance modulus of (m − M)0 =
24.44 ± 0.1 mag (Freedman & Madore 1990), given the uncer-
tainties.
Dolphin et al. (2003) made similar estimates of the RR Lyrae
luminosity in their field, but found <V0>= 24.81 ± 0.11 mag
-- about 0.24 mag brighter than our own estimate. Given their
relatively low completeness, it is plausible that their RR Lyrae
sample is biased toward brighter stars (although they argue oth-
erwise). Dolphin inspected the Brown et al. (2003) CMD and
found support for their bright RR Lyrae estimate, but we stress
again that our previous presentation of the M31 halo data only
showed a greyscale Hess diagram, making such estimates from
the published figure difficult.
5. OTHER VARIABLES
Although the focus of this paper is the RR Lyrae population
in the Andromeda halo, we briefly note the other classes of vari-
ables suitable for followup studies. The short period variables
below the HB (see Figures 2 and 5) fall in the broad category
of dwarf Cepheids; given the wide age range in the M31 halo
(Brown et al. 2003), these are probably a mix of pulsating and
eclipsing stars on the main sequence (e.g., δ Scuti stars) and in
the blue straggler population (e.g., SX Phoenicis stars). Above
the HB, the bluest star (V84) is outside of the Cepheid instabil-
ity strip, but might be a pulsating post-asymptotic branch star.
The other two stars above the HB (V118 and V60) have periods
that are longer than typically found for Anomalous Cepheids,
but their luminosities and periods put them close to the Popula-
tion I and Population II Cepheids. In any case, if they are both
pulsating variables, they cannot belong to the same class, given
that the star with the longer period is fainter. The semiregulars
and long period variables (see Figure 4) include RV Tauri stars
and Mira stars. In globular clusters, Miras are found only at
[Fe/H] > −1 (Frogel & Whitelock 1998), where they have pe-
riods ∼< 310 days. Given the wide age spread in the M31 halo,
one might expect some of the Miras to have periods longer than
310 days, but characterization of these stars would require a
longer baseline than that currently available.
6. SUMMARY
We have presented a complete survey of the RR Lyrae stars
in an M31 halo field, 51 arcmin from the nucleus. We find 29
RRab stars with a mean period of 0.594 days, 25 RRc stars
with a mean period of 0.316 days, and 1 RRd star with a fun-
damental period of 0.473 days and a first overtone period of
0.353 days. The RR Lyrae population of the M31 halo can-
not be clearly placed into either Oosterhoff type, and is dis-
tinct from the Milky Way cluster and halo field populations. In
a broad sense, the Local Group dSphs share the intermediate
Oosterhoff status found the M31 halo, but the characteristics
of the M31 RR Lyraes (<Pab>, <Pc>, Nc/Nabc) are distinct
from the dSphs, suggesting that the M31 halo is not comprised
of dissolved globular clusters like those in the Milky Way or
dissolved Local Group dSphs. The specific frequency of RR
Lyraes (SRR = 5.6) is very high for a mean halo metallicity of
[Fe/H]= −0.8, but within the normal range when considered
as a component of the old, metal-poor halo population. The
mean metallicity of the RR Lyrae population is indeed much
lower than that of the halo, with a mean [Fe/H]= −1.77 for
the RRab stars and a mean [Fe/H]= −1.43 for the RRc stars.
The distance to M31 determined from the RR Lyrae luminos-
ity is (m − M)o = 24.5 ± 0.1 mag, in good agreement with the
Cepheid distance.
Support for proposal 9453 was provided by NASA through
a grant from STScI, which is operated by AURA, Inc., under
NASA contract NAS 5-26555. We are grateful to P. Stetson for
providing the latest version of DAOPHOT and assistance with
its use, and to R. Gilliland for useful discussions. H. Smith
kindly reviewed a draft of this manuscript and offered helpful
comments. We thank the members of the scheduling and op-
erations teams at STScI (especially P. Royle, D. Taylor, and D.
Soderblom) for their efforts in executing a large program during
a busy HST cycle. Our paper was improved by the suggestions
of an anonymous referee.
REFERENCES
15
Krist, J. 1995, ASP Conference Series 77, Astronomical Data Analysis
Software and Systems IV, ed. R.A. Shaw, H.E. Payne, & J.J.E. Hayes, 349
Lafler, J., & Kinman, T.D. 1965, ApJS, 11, 216
Lejeune, T., Cuisinier, F., & Buser, R. 1997, A&AS, 125, 229
Lomb, N.R. 1976, Ap&SS
Mateo, M. 1998, ARA&A, 36, 435
Mould, J., & Kristian, J. 1986, ApJ, 305, 591
Oosterhoff, P. Th. 1939, Observatory, 62, 104
Press, W.H., & Rybicki, G.B. 1989, ApJ, 338, 277
Pritchet, C.J., & van den Bergh, S. 1987, ApJ, 316, 517
Pritzl, B.J., Armandroff, T.E., Jacoby, G.H., & Da Costa, G.S. 2002, AJ, 124,
Pritzl, B.J., Armandroff, T.E., Jacoby, G.H., & Da Costa, G.S. 2002, AJ, 127,
Alcock, C., et al. 2000, AJ, 119, 2194
Bono, G., Caputo, F., Castellani, V., & Marconi, M. 1996, ApJ, 471, 33
Bono, G., Caputo, F., & Stellingwerf, R.F. 1994, ApJ, 423, 294
Brown, T.M. 2003, in "The Local Group as an Astrophysical Laboratory," ed.
M. Livio (Cambridge University Press), in press, astro-ph/0308298.
Brown, T.M., Ferguson, H.C., Smith, E., Kimble, R.A., Sweigart, A.V.,
Renzini, A., Rich, R.M., & VandenBerg, D.A. 2003, ApJ, 592, L17.
Cacciari, C., & Clementini, G. 2003, in Lecture Notes in Physics Volume 635,
Stellar Candles for the Extragalactic Distance Scale, ed. D.M. Alloin & W.
Gieren (New York: Springer), 635, 105
Cacciari, C., & Renzini, A. 1976, A&AS, 25, 303
Carretta, E., Gratton, R.G., Clementini, G., Fusi Pecci, F. 2000, 533, 215
Clement, C.M. 2000, in ASP Conf. Ser. 203, The Impact of Large-Scale
Surveys on Pulsating Star Research, ed. L. Szabados & D.W. Kurtz (IAU
Colloq. 176) (San Francisco: ASP), 266
Clement, C.M., et al. 2001, AJ, 122, 2587
Dall'Ora, M., et al. 2003, AJ, 126, 197
Dolphin, A.E., Saha, A., Olszewski, E.W., Thim, F. Skillman, E.D., Gallagher,
J.S., & Hoessel, J. 2003, AJ, in press, astro-ph/0311300
Durrell, P.R., Harris, W.E., & Pritchet, C.J. 2001, AJ, 121, 2557
Ferguson, A.M.N., Irwin, M.J., Ibata, R.A., Lewis, G.F., & Tanvir, N.R. 2002,
AJ, 124, 1452
Fitzpatrick, E.L. 1999, PASP, 111, 63
Ford, H.C., et al. 1998, Proc. SPIE, 3356, 234.
Freedman, W.L., & Madore, B.F. 1990, ApJ, 365, 186
Frogel, J.A., & Whitelock, P.A. 1998, 116, 754
Harris, W.E. 1996, AJ, 112, 1487
Holland, S., Fahlman, G.G., & Richer, H.B. 1997, AJ, 114, 1488
1464
318
82, 947
Renzini, A. 1980, Mem. S. A. It., 51, 749
Renzini, A. 1983, Mem. S. A. It., 54, 335
Ryan, S.G., & Norris, J.E. 1991, AJ, 101, 1865
Sandage, A. 1993, AJ, 106, 687
Sargent, W.L.W., Kowal, C.T., Hartwick, F.D.A., van den Bergh, S. 1977, AJ,
Scargle, J.D. 1982, ApJ, 263, 835
Schlegel, D.J., Finkbeiner, D.P., & Davis, M. 1998, ApJ, 500, 525
Siegel, M.H., & Majewski, S.R. 2000, AJ, 120, 284
Smith, H.A., et al. 2003, PASP, 115, 43
Stetson, P. 1987, PASP, 99, 191
Suntzeff, N.B., Kinman, T.D., & Kraft, R.B. 1991, ApJ, 367, 528
van den Bergh, S. 1999, A&A Rev., 9, 273
VandenBerg, D.A., 2000, ApJS, 129, 315
Zinn, R., & West, M.J. 1984, ApJS, 55, 45
|
astro-ph/0011094 | 1 | 0011 | 2000-11-03T20:36:28 | Stellar population synthesis models at low and high redshift | [
"astro-ph"
] | The basic assumptions behind Population Synthesis and Spectral Evolution models are reviewed. The numerical problems encountered by the standard population synthesis technique when applied to models with truncated star formation rates are described. The Isochrone Synthesis algorithm is introduced as a means to circumvent these problems. A summary of results from the application of this algorithm to model galaxy spectra by Bruzual and Charlot (1993, 2000) follows. I present a comparison of these population synthesis model predictions with observed spectra and color magnitude diagrams for stellar systems of various ages and metallicities. It is argued that models built using different ingredients differ in the resulting values of some basic quantities (e.g. $M/L_V$), without need to invoking violations of physical principles. The range of allowed colors in the observer frame is explored for several galaxy redshifts. | astro-ph | astro-ph |
Stellar population synthesis models at low
and high redshift†
By Gustavo Bruzual A.
Centro de Investigaciones de Astronom´ıa (CIDA), A.P. 264, M´erida, Venezuela
The basic assumptions behind Population Synthesis and Spectral Evolution models are reviewed.
The numerical problems encountered by the standard population synthesis technique when
applied to models with truncated star formation rates are described. The Isochrone Synthesis
algorithm is introduced as a means to circumvent these problems. A summary of results from
the application of this algorithm to model galaxy spectra by Bruzual and Charlot (1993, 2000)
follows. I present a comparison of these population synthesis model predictions with observed
spectra and color magnitude diagrams for stellar systems of various ages and metallicities. It is
argued that models built using different ingredients differ in the resulting values of some basic
quantities (e.g. M/LV ), without need to invoking violations of physical principles. The range
of allowed colors in the observer frame is explored for several galaxy redshifts. †
1. Introduction
The number distribution of the stellar populations present in a galaxy is a function of
time. Thus, the number of stars of a given spectral type, luminosity class, and metallicity
content changes as the galaxy ages.
In early-type galaxies (E/S0) most of the stars
were formed during, or very early after the initial collapse of the galaxy and the stellar
population ages as times goes by. The chemical abundance in these systems must have
reached the value measured in the stars very quickly during the formation process since
most E/S0 galaxies show little evidence of recent major events of star formation. In late-
type galaxies the stellar population also ages, but there is a significant number of new
stars being formed. Depending on the star formation rate, Ψ(t), the mean age of the stars
in a galaxy may even decrease as the galaxy gets older. In general, in late-type systems
the metal content of the stars and the interstellar medium is an increasing function of
time. Ψ(t) can also increase above its typical value due to interactions between two or
more galaxies or with the environment.
As a consequence of the aging of the stellar population, or its renewal in the case of
galaxies with high recent Ψ(t), the rest-frame spectral distribution of the light emitted by
a galaxy is a function of its proper time. Observational properties such as photometric
magnitude and colors, line strength indices, metal content of gas and young stars, depend
on the epoch at which we observe a galaxy on its reference frame. This intrinsic evolution
should not be mistaken with the apparent evolution produced by the cosmological redshift
z. For distant galaxies both effects may be equally significant.
In order to search for spectral evolution in galaxy samples we must (a) quantify the
amount of evolution expected between cosmological epoch t1 and t2, and (b) design
observational tests that will reveal this amount of evolution, if present. Evolutionary
population synthesis models predict the amount of evolution expected under different
scenarios and allow us to judge the feasibility of measuring it. In the ideal case (Arag´on-
Salamanca et al. 1993; Stanford et al. 1995, 1998; Bender et al. 1996) spectral evolution
is measured simply by comparing spectra of galaxies obtained in such a way that the same
† To appear in Proceedings of the XI Canary Islands Winter School of Astrophysics on Galax-
ies at High Redshift, eds. I. P´erez-Fournon, M. Balcells and F. S´anchez
1
2
Gustavo Bruzual A.: Population Synthesis at low and high z
rest frame wavelength region is sampled in all galaxies, irrespective of z. In this case there
is no need to apply uncertain K-corrections to transform all the spectra to a common
wavelength scale. Alternatively, if this approach cannot be applied, e.g. when studying
large samples of faint galaxies, one can use models which allow for different degrees of
evolution (including none) to derive indirectly the amount of evolution consistent with
the data (Pozzetti et al. 1996; Metcalfe et al. 1996). Clearly, the first approach is to be
preferred whenever possible. In all cases we must rule out possible deviations from the
natural or passive evolution of the stellar population in some of the galaxies under study
induced, for example, by interactions with other galaxies or cluster environment, etc.
The large amount of astrophysical data that has become available in the last few years
has made possible to build several complete sets of stellar population synthesis models.
The predictions of these models have been used to study many types of stellar systems,
from local normal galaxies to the most distant galaxies discovered so far (approaching
z of 4 to 5), from globular clusters in our galaxy to proto-globular clusters forming in
different environments in distant, interacting galaxies. In this paper I present an overview
of results from population synthesis models directly applicable to the interpretation of
galaxy spectra.
2. The Population Synthesis Problem
Stellar evolution theory provides us with the functions Tef f (m, Z, t) and L(m, Z, t)
which describe the behavior in time t of the effective temperature Tef f and luminosity
L of a star of mass m and metal abundance Z. For fixed m and Z, L(t) and Tef f (t)
describe parametrically in the H-R diagram the evolutionary track for stars of this mass
and metallicity. The initial mass function (IMF), φ(m), indicates the number of stars of
mass m born per unit mass when a stellar population is formed. The star formation rate
(SFR), Ψ(t), gives the amount of mass transformed into stars per unit time according
to φ(m). The metal enrichment function (MEF), Z(t), also follows from the theory of
stellar evolution. The population synthesis problem can then be stated as follows. Given
a complete set of evolutionary tracks and the functions φ(m) and Ψ(t), compute the
number of stars present at each evolutionary stage in the H-R diagram as a function
of time. To solve this problem exactly we need additional knowledge about the MEF,
Z(t), which gives the time evolution of the chemical abundance of the gas from which
the successive generations of stars are formed. For simplicity, it is commonly assumed
that φ(m) and Ψ(t) are decoupled from Z(t), even though it is recognized that in real
stellar systems these three quantities are most likely closely interrelated. The spectral
evolution problem can be solved trivially once the population synthesis problem is solved,
provided that we know the spectral energy distribution (SED) at each point in the H-R
diagram representing an evolutionary stage in our set of tracks. The discussion that
follows is based in the work of Charlot and Bruzual (1991, hereafter CB91) and Bruzual
and Charlot (1993, 2000, hereafter BC93 and BC2000). See also Bruzual (1998, 1999,
2000). Unless otherwise indicated, I will ignore chemical evolution and assume that at
all epochs stars form with a single metallicity, Z(t) = constant, and the same φ(m).
Let N o
i be the number of stars of mass Mi born when an instantaneous burst of star
formation occurs at t = 0. This kind of burst population has been called simple stellar
population (SSP, Renzini 1981). When we look at this population at later times, we will
see the stars traveling along the corresponding evolutionary track. If the stars live in
their kth evolutionary stage from time ti,k−1 to time ti,k, then at time t the number of
Gustavo Bruzual A.: Population Synthesis at low and high z
stars of this mass populating the kth stage is simply
Ni,k(t) = (cid:26) N o
0,
i ,
if ti,k−1 ≤ t < ti,k;
otherwise.
3
(1)
For an arbitrary SFR, Ψ(t), we compute the number of stars ηi,k(t) of mass Mi at the
kth evolutionary stage from the following convolution integral
ηi,k(t) = Z t
0
Ψ(t − t′)Ni,k(t′)dt′,
which in view of (1) can be written as
ηi,k(t) = N o
i Z min(t,ti,k)
ti,k−1
Ψ(t − t′)dt′.
(2)
(3)
From (3) we see that the commonly heard statement that the number of stars expected
in a stellar population at a given position in the H-R diagram is proportional to the time
spent by the stars at this position, i.e.
ηi,k ∝ N o
i (ti,k − ti,k−1),
(4)
is accurate only for a constant Ψ(t). For non-constant SFRs, the integral in (3) assigns
more weight to the epochs of higher star formation. For instance, for an exponentially
decaying SFR with e-folding time τ , Ψ(t) = exp(−t/τ ), we have
ηi,k(t) ∝ N o
i {exp[−(t − ti,k)/τ ] − exp[−(t − ti,k−1)/τ ]}.
(5)
Ψ(t) was stronger at time (t − ti,k) than at time (t − ti,k−1), which is clearly taken into
account in (5).
A prerequisite for building trustworthy population synthesis and spectral evolution
models is an adequate algorithm to follow the evolution of consecutive generations of
stars in the H-R diagram. This goal is accomplished by the standard technique described
above provided that the function Ψ(t) extends from t = 0 to t = ∞. Special caution is
required if Ψ(t) becomes 0 at a finite age. As an illustration, let us consider the case of
a burst of star formation which lasts for a finite length of time τ ,
or equivalently,
Ψ(t) = (cid:26) Ψo,
0,
if 0 ≤ t ≤ τ ;
otherwise,
Ψ(t − t′) = (cid:26) Ψo,
0,
if t − τ ≤ t′ ≤ t;
otherwise.
In this case, equation (2) reduces to
ηi,k(t) = (cid:26) > 0,
0,
if [ti,k−1, ti,k] ∩ [t − τ, t] 6= ∅;
otherwise.
(6)
(7)
(8)
From (8) we see that ηi,k(t) can be = 0 depending on the value of τ and on the value of t
chosen to sample the stellar population. The most extreme case is that of the SSP (τ = 0),
for which Ψ(t) = δ(t), and ηi,k(t) in (8) is identical to Ni,k(t) in (1). For a typical set of
evolutionary tracks, there is no grid of values tj of the time variable t for which all the
stellar evolutionary stages included in the tracks can be adequately sampled for arbitrarily
chosen values of τ . This is obviously an undesirable property of population synthesis
models. The models should be capable of representing galaxy properties in a continuous
and well behaved form, independent of the sampling time scale t. Consequently, standard
population synthesis models for truncated Ψ(t) as given by (6) will reflect the coarseness
of the set of evolutionary tracks, and may miss stellar evolutionary phases depending on
4
Gustavo Bruzual A.: Population Synthesis at low and high z
our choice of t. This results in unwanted and unrealistic numerical noise in the predicted
properties of the stellar systems studied with the synthesis code (see example in CB91).
A solution to this problem is to build a set of evolutionary tracks with a resolution in
mass which is high enough to guarantee that all evolutionary stages, i.e. all values of k in
(2), can be populated for any choice of t or τ in (6). In other words, there must be tracks
for so many stellar masses in this ideal library that for any model age, we have at hand
the position in the H-R diagram of the star which at that age is in the kth evolutionary
stage. The realization of this library is not possible with present day computers. This
limitation can be circumvented by careful interpolation in a relatively complete set of
evolutionary tracks.
3. The Isochrone Synthesis Algorithm
The isochrone synthesis algorithm described in this section allows us to compute con-
tinuous isochrones of any age from a carefully selected set of evolutionary tracks. The
isochrones are then used to build population synthesis models for arbitrary SFRs Ψ(t),
without encountering any of the problems mentioned above. This algorithm has been
used by CB91, BC93, and BC2000. The details of the algorithm follow.
A relationship is built between the main sequence (MS) mass of a star M and the age
t of the star during the kth evolutionary stage (CB91). Ignoring mass loss is justified
because M is used only to label the tracks. For the set of tracks used by BC93 (described
in the next section) 311 different relationships log M vs. log t are built, which can be
visualized as 311 different curves in the (log t,
log M ) plane. We derive by linear
interpolation the MS mass mk(t′) of the star which will be at the kth evolutionary stage
at age t′, given by
log mk(t′) = Ak,i log(Mi) + (1 − Ak,i) log(Mi+1),
where
Ak,i =
log ti+1,k − log t′
log ti+1,k − log ti,k
.
(9)
(10)
ti,k represents the age of the star of mass Mi at the kth evolutionary stage, and
and
ti,k ≤ t′ < ti+1,k,
Mi+1 ≤ mk(t′) < Mi.
The procedure is performed for all the curves that intersect the log t = log t′ line. We
thus obtain a series of values of mk which must now be assigned values of log L and
log Tef f in order to define the isochrone corresponding to age t′.
To compute the integrated properties of the stellar population, we must specify the
number of stars of mass mk. This number is determined from the IMF, which we can
write generically as
n(mk) = φ(m−
k , m+
k , 1 + x).
(11)
Of this number of stars,
(12)
stars are assigned the observational properties of the star of mass Mi at the kth evolu-
tionary stage, and
Ni,k = Ak,in(mk)
(13)
stars, the observational properties of the star of mass Mi+1 at the same kth stage. This
Ni+1,k = (1 − Ak,i)n(mk)
Gustavo Bruzual A.: Population Synthesis at low and high z
5
procedure is equivalent to interpolating the tracks for the stars of mass Mi and Mi+1 to
derive the values of log L and log Tef f to be assigned to the star of mass mk, but this
intermediate step is unnecessary. In (11) m−
k are computed from
k and m+
m−
k = (mk−1mk)1/2,
m+
k = (mkmk+1)1/2.
(14)
In this case mk−1, mk, and mk+1 represent the masses obtained by interpolation at the
given age in the segments representing the (k−1)th, kth, and (k+1)th stages, respectively.
The IMF has been assumed to be a power law of the form (Salpeter 1955)
φ(m1, m2, 1 + x) = cZ m2
m1
m−(1+x)dm.
(15)
For a given Ψ(t) equation (2) gives the number ηi,k(t) of stars of mass Mi at the kth
evolutionary stage. If fi,k(λ) represents the SED corresponding to the star of mass Mi
during the kth stage, then the contribution of the ηi,k(t) stars to the integrated evolving
SED of the stellar population is simply given by
The resulting SED for the stellar population is given by
Fi,k(λ, t) = ηi,k(t)fi,k(λ).
F (λ, t) = Xi,k
Fi,k(λ, t).
(16)
(17)
4. Evolutionary Population Synthesis Models
A number of groups has developed in recent years different population synthesis mod-
els which provide a sound framework to investigate the problem of spectral evolution
of galaxies. Some of the most commonly used models are Arimoto & Yoshii (1987),
Guiderdoni & Rocca-Volmerange (1987), Buzzoni (1989), Bressan, Chiosi & Fagotto
(1994), Fritze-v.Alvensleben & Gerhard (1993), Worthey (1994), Bruzual & Charlot
(1993, 2000). The basic astrophysical ingredients used in these models are: (1) Stellar
evolutionary tracks of one or more metallicities; (2) Spectral libraries, either empirical
or theoretical model atmospheres; (3) Sets of rules, or calibration tables, to transform
the theoretical HR diagram to observational quantities (e.g. B − V vs. Tef f , V −
K vs. Tef f , B.C. vs. Tef f , etc.). These rules are not necessary when theoretical model
atmosphere libraries are used which are already parameterized according to Tef f , log g,
and [Fe/H]; (4) Additional information, such as analytical fitting functions, required to
compute various line strength indices (Worthey et al. 1994). Regardless of the specific
computational algorithm used, all evolutionary synthesis models depend on three ad-
justable parametric functions: (1) the stellar initial mass function, f (m), or IMF; (2)
the star formation rate, Ψ(t); and (3) the chemical enrichment law, Z(t). For a given
choice of f (m), Ψ(t), and Z(t), a particular set of evolutionary synthesis models provides:
(1) Galaxy spectral energy distribution vs. time, Fλ(λ, Z(t), t); (2) Galaxy colors and
magnitude vs. time; (3) Line strength and other spectral indices vs. time. Some authors
(e.g. Bressan et al. 1994; Fritze-v.Alvensleben & Gerhard 1994) consider that Z(t) can
be derived self-consistently from their models. In other instances Z(t) is introduced as
an external piece of information. I discuss below some results from work still in progress
in collaboration with S. Charlot.
6
Gustavo Bruzual A.: Population Synthesis at low and high z
Figure 1. Evolving spectral energy distributions. (a) Evolution in time of the SED of a SSP
computed for the Salpeter IMF (mL = 0.1, mU = 125 M⊙). The age in Gyr is indicated next
to each spectrum. (b) Same as (a) but for a composite population in which stars form according
to Ψ(t) = exp(−t/τ ) for τ = 3 Gyr. The total mass of each model galaxy is 1 M⊙. Fλ in frame
(b) has been multiplied by 100 to use a common vertical scale.
5. Stellar Ingredients
BC2000 have extended the BC93 evolutionary population synthesis models to provide
the evolution in time of the spectrophotometric properties of SSPs for a wide range of
stellar metallicity. In an SSP all the stars form at t = 0 and evolve passively afterward.
The BC2000 models are based on the stellar evolutionary tracks computed by Alongi et
al. (1993), Bressan et al. (1993), Fagotto et al. (1994a, b, c), and Girardi et al. (1996),
which use the radiative opacities of Iglesias et al. (1992). This library includes tracks for
stars with initial chemical composition Z = 0.0001, 0.0004, 0.004, 0.008, 0.02, 0.05, and
0.10 (Table 1), with Y = 2.5Z + 0.23, and initial mass 0.6 ≤ m/M⊙ ≤ 120 for all
metallicities, except Z = 0.0001 (0.6 ≤ m/M⊙ ≤ 100) and Z = 0.1 (0.6 ≤ m/M⊙ ≤ 9).
This set of tracks will be referred to as the Padova or P tracks hereafter. A similar set
of tracks for slightly different values of Z has been published by Girardi et al. (2000).
A comparison of the predictions of models built with both sets of Padova tracks will be
shown elsewhere (but see Fig. 7 and Bruzual 2000).
The published tracks go through all phases of stellar evolution from the zero-age main
sequence to the beginning of the thermally pulsing regime of the asymptotic giant branch
(AGB, for low- and intermediate-mass stars) and core-carbon ignition (for massive stars),
and include mild overshooting in the convective core of stars more massive than 1 M⊙.
Gustavo Bruzual A.: Population Synthesis at low and high z
7
Z
X
Y
[Fe/H]
0.0001
0.0004
0.0040
0.0080
0.0200
0.0500
0.1000
0.7696
0.7686
0.7560
0.7420
0.7000
0.5980
0.4250
0.2303
0.2310
0.2400
0.2500
0.2800
0.3520
0.4750
-2.2490
-1.6464
-0.6392
-0.3300
0.0932
0.5595
1.0089
Table 1. Model chemical composition
The Post-AGB evolutionary phases for low- and intermediate-mass stars were added to
the tracks by BC2000 from different sources (see BC2000 for details).
BC2000 use as well a parallel set of tracks for solar metallicity computed by the Geneva
group (Geneva or G tracks hereafter), which provides a framework for comparing models
computed with two different sets of tracks.
The BC2000 models use the library of synthetic stellar spectra compiled by Lejeune
et al. (1997, 1998, LCB97 and LCB98 hereafter) for all the metallicities in Table 1. This
library consists of Kurucz (1995) spectra for the hotter stars (O-K), Bessell et al. (1989,
1991) and Fluks et al. (1994) spectra for M giants, and Allard & Hauschildt (1995)
spectra for M dwarfs. For Z = Z⊙, BC2000 also use the Pickles (1998) stellar atlas,
assembled from empirical stellar data.
6. Spectral evolution at fixed metallicity
Fig. 1a shows the evolution in time of the SED for the SSP model. In an SSP all
the stars form at t = 0 and evolve passively afterward. In all the examples shown in
this paper I assume that stars form according to the Salpeter (1955) IMF in the range
from mL = 0.1 to mU = 125 M⊙. The total mass of the model galaxy is 1 M⊙. The
evolution is fast and is dominated by massive stars during the first Gyr in the life of
the SSP (6 top SEDs). The flux seen around 2000 A at 4 and 7 Gyr is produced by
the turn-off stars. The UV-rising branch (Burstein et al. 1988, Greggio and Renzini
1990) seen after 10 Gyr is produced by the PAGB stars. These stars are also responsible
of the decrease in the amplitude of the 912 A discontinuity observed after 4 Gyr. The
SSP model is the basic ingredient which, together with the convolution integral (2), is
used to compute models with arbitrary SFRs and equal IMF. For illustration I show in
Fig. 1b the evolution of a model with Ψ(t) = exp(−t/τ ) for τ = 3 Gyr. The UV to
optical spectrum remains roughly constant during the main episode of star formation
because of the continuous input of young massive stars, but the near-infrared light rises
as evolved stars accumulate. When star formation drops, the spectral characteristics at
various wavelengths are determined by stars in advanced stages of stellar evolution.
7. Dependence of galaxy properties on stellar metallicity
Fig. 2 shows the predicted SEDs at t = 12 Gyr for chemically homogeneous SSPs of
the indicated metallicity. The SEDs shown in Fig. 2 have been normalized at λ = 5500A
to make the comparison more clear. Fig. 3. shows the evolution in time of the B − V
8
Gustavo Bruzual A.: Population Synthesis at low and high z
Figure 2. Chemically homogeneous BC2000 SSP model galaxy SEDs at age = 12 Gyr. Each
line pattern represents a different metallicity, as indicated inside the frame. All the models
shown were computed for the Salpeter IMF (mL = 0.1, mU = 125 M⊙). The total mass of each
model galaxy is 1 M⊙. The SEDs have been normalized at λ = 5500A.
and V − K colors, and the M/LV ratio predicted by BC2000 for the same SSPs shown
in Fig. 2.
From Figs. 2 and 3 it is apparent that there is a uniform tendency for galaxies to
become redder in B − V as the metallicity increases from Z = 0.0001 ( 1
200 Z⊙) to Z =
0.05 (2.5 × Z⊙). The V − K color and the M/LV ratio show the expected tendency with
metallicity, i.e. V − K becomes redder and M/LV becomes higher with increasing Z.
Fig. 4 shows the evolution in time of the Mgb, Hβ, and Ca spectral indices as defined
by Worthey (1994) for the same BC2000 SSP models shown in Figs. 2 and 3. Again, the
models show the expected tendency with Z and match the values computed by Worthey
(1994).
It should be remarked that the time behavior of the line strength indices at
constant Z is due to the change in the number of stars at different positions in the
HR diagram produced by stellar evolution and is not related to chemical evolution. The
indices change also in chemically homogeneous populations. The Hβ index is less sensitive
to the stellar metallicity than the Mgb and Ca index. Instead, the Hβ index is high when
there is a large fraction of MS A-type stars (t < 1) Gyr.
In Fig. 5 I compare the behavior of the SSP models in the (U − B) vs. (B − V ) color
plane with the LMC cluster data from Bica et al. (1996b), discriminated by SWB class
and model age. In each panel the models (lines) are shown in the range of age for which
the predicted colors overlap the observed colors for the class. It is apparent from the
figure that the models reproduce quite well the observed colors.
Gustavo Bruzual A.: Population Synthesis at low and high z
9
Figure 3. Evolution in time of the B − V , and V − K colors, and the M/LV ratio for the
BC2000 SSP models shown in Fig. 2. Each line represents a different metallicity, as indicated
in the bottom panel.
8. Calibration of the models in the C-M diagram
Population synthesis models, frequently used to study composite stellar populations in
distant galaxies, are rarely confronted with observations of local simple stellar popula-
tions, such as globular star clusters, whose age and metal content have been constrained
considerably in later years. If the models cannot reproduce the CMDs and SEDs of these
objects for the correct choice of parameters, their predictive power becomes weaker and
their usage to study complex galaxies may not be justified.
It is important to compare the properties of the population synthesis models to obser-
vations of stellar systems whose age and metallicity is well determined. This is a means
to test to what extent the adopted relationships between stellar color and magnitude
and effective temperature and luminosity (or surface gravity) introduce systematic shifts
between the predicted and observed isochrones in the C-M diagram.
For each value of the metallicity Z listed in Table 1, and a particular choice of the IMF
φ(m), a BC2000 SSP model consists of a set of 221 evolving integrated SEDs spanning
from 0 to 20 Gyr. The isochrone synthesis algorithm used to build these models renders
it straightforward to compute the loci described by the stellar population in the CMD
at any time step and in any photometric band. We can extract the model SED that best
reproduces a given observed SED and assign an age to the program object, and then
examine how well the isochrone computed at this age fits the most significant features in
the CMD of this object, if available.
10
Gustavo Bruzual A.: Population Synthesis at low and high z
Figure 4. Evolution in time of the Mgb, Hβ, and Ca spectral indices as defined by Worthey
(1994) for the BC2000 models shown in Fig. 2. The different symbols represent the values of
the indices computed by Worthey for the same range in Z. Each line represents a different
metallicity, as indicated in the middle panel.
Fig. 6 compares the (B, B − V ) observations of the clusters M67 and the Hyades with
the isochrones obtained from the BC2000 models. These two clusters have nearly solar
metallicity: M67 ([Fe/H]≈ 0.01), the Hyades ([Fe/H]≈ 0.15). For M67 we adopted a
distance modulus of 9.5 mag and a color excess E(B − V ) = 0.06 mag (Janes 1985). For
the Hyades a distance modulus of 3.4 mag (Peterson and Solensky 1988) and E(B −V ) ≈
0. Estimates of the ages of these clusters vary from 4 to 4.3 Gyr for M67 and from 0.5 to
0.8 Gyr for the Hyades. The isochrones are shown at 4 Gyr (M67) and 0.6 Gyr (Hyades).
The data points for M67 are from Eggen and Sandage (1964, open circles), Racine (1971,
filled circles), Janes and Smith (1984, triangles), and Gilliland et al. (1991; crosses). For
the Hyades the observations are from Upgren (1974, triangles), Upgren and Weis (1977,
filled circles), and Micela et al. (1988, open circles).
Fig. 7 shows a comparison of the excellent HST CMD diagram of NGC 6397 assembled
from various sources by D'Antona (1999) with isochrones computed from the Padova-
1994 tracks for Z = 0.0004 and the Padova-2000 tracks for Z = 0.001 at ages 10 to 16
Gyr. The original version of the model atmospheres in the LCB atlas was used to derive
the colors in Fig. 7. The corrected version of these models produces considerably worse
agreement with the observations, mainly in the MS from the turn-off down. The redder
cluster RGB most likely reflects a slightly higher metallicity than Z = 0.0004, close to
Z = 0.001. Despite the discrepancies seen in Figs. 6 and 7 (mainly in the RGB of M67
and NGC 6397) the agreement between the predicted isochrones and the loci in the CMD
Gustavo Bruzual A.: Population Synthesis at low and high z
11
Figure 5. Points: LMC cluster (U − B) vs. (B − V ) colors from Bica et al. (1996b),
discriminated by SWB class. The lines represent SSP models at various age ranges.
of these clusters may be regarded as satisfactory, and is excellent in some parts of the
diagram.
9. Observed Color-Magnitude Diagrams and Integrated Spectra
The high quality and depth of the HST VI-photometry NGC 6528 (Z <∼ Z⊙), as well
as the high signal-to-noise ratio of the integrated SED of NGC 6528 currently available,
provide an excellent framework for testing the range of validity of population synthesis
models (see Bruzual et al. 1997 for details).
The integrated spectrum of NGC 6528 in the wavelength range λ = 3500 - 9800 A was
obtained by combining the visible, near-infrared, and near-ultraviolet spectra of Bica
& Alloin (1986, 1987) and Bica et al.
(1994), respectively. We applied a reddening
correction of E(B-V) = 0.66, adopted in Bica et al. (1994 and references therein). The
SED of NGC 6528 is typical of old stellar populations of high metal content (Santos et
al. 1995). It may happen that NGC 6528, similarly to NGC 6553, has [Fe/H] < 0.0
(Barbuy et al. 1997a), whereas the [α-elements/Fe] are enhanced, resulting in [Z/Z⊙] ≈
0.0, or possibly slightly below solar. Since we have evolutionary tracks for Z = 0.02 and
Z = 0.008, we adopt Z = 0.02.
The data available for this cluster provide a unique opportunity to examine the 6
different options for the Z = Z⊙ models considered by BC2000. Thus, we can study
objectively which of the basic building blocks used by BC2000 for Z = Z⊙: P or G tracks;
Pickles, LCB97-O (original version), or LCB97-C (corrected version) stellar libraries, is
12
Gustavo Bruzual A.: Population Synthesis at low and high z
Figure 6. CMD of M67 and the Hyades compared with isochrones derived from the Padova
tracks for solar metallicity. See text for details.
Model
1
2
3
4
5
6
Spectral
Library
Pickles
"
LCB97-C
"
LCB97-O
"
Stellar Best-fitting
Tracks
age (Gyr) Σ2
min
P
G
P
G
P
G
11.75
10.25
10.00
8.50
11.50
9.00
2.22
1.34
1.73
1.38
3.35
1.95
Table 2. Z⊙ model fits to SED of NGC 6528
most successful at reproducing the data. In Table 2 I show the age at which Σ2, defined
as the sum of squared residuals [log Fλ(observed) − log Fλ(model)]2, is minimum for
various Z = Z⊙ models. The values of Σ2
min given in Table 2 indicate the goodness-of-
fit. According to this criterion, models 2 and 4 provide the best fit to the integrated
SED of NGC 6528 in the wavelength range λ 3500-9800 A. This fit is shown in Fig.
8. Except for the differences in the best-fitting age, model 3 provides a comparable,
although somewhat poorer, fit to the SED of this cluster. The residuals for models 1, 5
Gustavo Bruzual A.: Population Synthesis at low and high z
13
Figure 7. CMD of NGC 6397 compared with isochrones derived from the Padova-1994 tracks
for Z = 0.0004 (top), and the Padova-2000 tracks for Z = 0.001 (bottom). The observations
were assembled by D'Antona (1999) from the sources indicated in the figure.
14
Gustavo Bruzual A.: Population Synthesis at low and high z
Figure 8. Best fit to the integrated spectrum of NGC 6528 (heavy line) in the range λλ 3500
- 9800 A for model 2 in Table 2 (thin line extending over the full wavelength range). The best
fit occurs at 10.25 Gyr. The residuals of the fit, log Fλ(observed) − log Fλ(model), are shown
as a function of wavelength.
Figure 9.
Intrinsic MV vs. (V − I)0 CMD of NGC 6528 shown together with theoretical
isochrones for the Z = Z⊙ models listed in Table 2.
Gustavo Bruzual A.: Population Synthesis at low and high z
15
and 6 are considerably larger. Spectral evolution is slow at these ages, and the minimum
in the function Σ2 vs. age is quite broad. Reducing or increasing the model age by 1
or 2 Gyr produces fits of comparable quality to the one at which Σ2 is minimum. For
instance, for model 2, Σ2(12 Gyr) = 1.88, and Σ2(8 Gyr) = 2, which are still better fits
according to the Σ2 criterion than the best fits provided by some models in Table 2.
Fig. 9 shows the intrinsic HST VI CMD of NGC 6528 together with the isochrones
corresponding to the models in Table 2.
It is apparent from this figure that all the
isochrones shown provide a good representation of the cluster population in this CMD,
especially the position of the turn-off and the base of the asymptotic giant branch (AGB).
We note that NGC 6528 shows a double turnoff, the upper one being due to contamination
from the field star main sequence. The appropriate TO location would be around that
indicated by the 10, 11 and 12 Gyr isochrones. Despite the fact that models 2 and 4
provide better fits to the SED of this cluster than models 1 and 3, the isochrones from
model 3 reproduce more closely the CMD diagram than models 1 and 2. The noisy
nature of the isochrones computed with models 1 and 2 is due to the lack of some stellar
types in the Pickles stellar library.
From these results we conclude that:
(1) The λ = 3500 − 9800 A SED for E(B-V)=0.66 and the VI CMD of NGC 6528 are
well reproduced by Z = Z⊙ models at an age from 9 to 12 Gyr.
(2) The age derived from the fit to the observed SED of NGC 6528 is extremely sensitive
to the assumed E(B-V). Using E(B-V) = 0.59 instead of 0.66, increases the best-fitting
ages by 2 to 3 Gyr, since the observed SED is then intrinsically redder. On the other
hand, if E(B-V) = 0.69, the observed SED is intrinsically bluer and the derived ages are
2 to 3 Gyr younger than for E(B-V) = 0.66. However, inspecting the isochrones in the
CMD, the ages derived for E(B-V) = 0.66 listed in Table 2, seem appropriate.
(3) The SED and CMD of this cluster are consistent with those expected for a Z = Z⊙
population at an age of ≈ 9-12 Gyr, if overshooting occurs in the convective core of stars
down to 1 M⊙ (P tracks, models 1, 3, and 5 in Table 2). If overshooting stops at 1.5 M⊙,
as in the G tracks, this age is reduced to ≈ 8-10 Gyr (models 2, 4, and 6 in Table 2).
(4) For the same spectral library, the ages derived from the P tracks are older than
the ones derived from the G tracks. This is due to the fact that the P tracks include
overshooting in the convective core of stars more massive than 1 M⊙ whereas the G
tracks stop overshooting at 1.5 M⊙. Thus, stars in this mass range require more time in
the P tracks to leave the main sequence than in the G tracks.
(5) For the same set of evolutionary tracks, the corrected LCB97 library seems to pro-
vide a better fit to the CMD than the Pickles atlas. Interpolation in the finer LCB97 grid
of models produces smoother isochrones than in the coarser Pickles atlas. We attribute
this to the fact that M stars are very sparse in the Pickles atlas. Furthermore, the tem-
perature scale becomes problematic for these stars in the Pickles atlas. In the LCB97
library, the temperature scale for giants relies on measurements of angular diameters
and fluxes, which enter directly in the definition of effective temperature. For dwarfs,
the temperature scale is more difficult to define, as discussed in LCB98.
(6) Noticeable differences exist in the isochrones computed for both sets of LCB97
libraries. The differences are more pronounced for the M giants of (V − I)0 > 1.6, and
(J − K)0 > 1, corresponding to a temperature of Te ≤ 4000K, and for the cool dwarfs
of (V − I)0 > 1, corresponding to a temperature of Te ≤ 4700K. There are very few of
these stars in the CMD of NGC 6528 to favor a particular choice of library. However,
the corrected library produces better fits to the observed SED. We attribute this fact to
the relative importance of the luminosity of M giants.
(7) In general the LCB97 corrections redden the stellar SEDs in the optical range,
16
Gustavo Bruzual A.: Population Synthesis at low and high z
producing redder SSP models at an earlier age. As a consequence, the ages derived in
Table 1 for the LCB97-C library are younger than the ones derived from the LCB97-O
library for the same set of tracks.
(8) We have adopted Z = Z⊙ for this cluster. However, for a slightly lower value of
Z, the derived age would be older.
10. Comparison of model and observed spectra
10.1. Solar metallicity
Fig. 10 shows a model fit to the average spectrum of an E galaxy (kindly provided by
M. Rieke). The model SED is the line extending over the complete wavelength range
shown in the figure. The observed SED covers the range from 3300 A to 2.75 µm. The
residuals (observed - model) are shown at the bottom of the figure in the same vertical
scale. The model corresponds to a 10 Gyr Z = Z⊙ SSP computed for the Salpeter
(1955) IMF (mL = 0.1, mU = 125 M⊙) using the P tracks and the Pickles (1998) stellar
atlas. The fit is excellent over most of the spectral range. A minor discrepancy remains
in the region from 1.1 to 1.7 µm. The source of this discrepancy is not understood at
the moment. In Fig. 11 I show the same model and E galaxy SED as in Fig. 10 but in
different units. In addition, in Fig. 11 I include the broad band fluxes representing the
average of many E galaxies in the Coma cluster (solid squares) from A. Stanford (private
communication). The observed SED is the one with the lowest spectral resolution. Fig.
12 shows a closer look at the same data in an enlarged scale. Again the agreement is
excellent for the 3 data sets. The discrepant line in Fig. 12 corresponds to the same
model shown in Figs 10 and 11 but I used the LCB97 synthetic stellar atlas instead of
the empirical stellar SEDs. Fig. 12 shows clearly that the models based on empirical
stellar SEDs are to be preferred over the ones based on theoretical model atmospheres.
Unfortunately, complete libraries of empirical stellar SEDs are available only for solar
metallicity.
10.2. Non-solar metallicity
Figs 13 and 14 show the results of a comparison of SSP models built for various metallic-
ities using the LCB97 atlas, all for the Salpeter IMF, with several of the average spectra
compiled by Bica et al. (1996a). The name and the metal content of the observed spectra
indicated in each panel is as given by Bica et al. The quoted age is derived from the
best fit of our model spectra to the corresponding observations. The residuals (observed
- model) are shown in the same vertical scale. See the description to Fig. 10 above for
more details. Even though, in detail, the fits for non-solar metallicity stellar populations
are not as good as the ones for solar metallicity, over all the models reproduce the ob-
servations quite well over a wide range of [Z/Z⊙], and provide a reliable tool to study
these stellar systems. The discrepancy can be due both to uncertainties in the synthetic
stellar atlas or the evolutionary tracks at these [Z/Z⊙]. I have used SSPs in all the fits,
neglecting possibly composite stellar populations, as well as any interstellar reddening.
11. Different sources of uncertainties in population synthesis models
11.1. Uncertainties in the astrophysics of stellar evolution
There are significant differences in the fractional contribution to the integrated light by
red giant branch (RGB) and asymptotic giant branch (AGB) stars in SSPs computed for
different sets of evolutionary tracks. Fig. 15 shows the contribution of stars in various
Gustavo Bruzual A.: Population Synthesis at low and high z
17
Figure 10. Best fit to an average Elliptical galaxy SED (heavy line) in the range λλ 3300
- 27500 A. The model is the thin line extending over the full wavelength range. The best fit
occurs at 10 Gyr for this model SED. The residuals of the fit, log Fλ(observed) − log Fλ(model),
are shown as a function of wavelength.
Best fit to an average Elliptical galaxy SED (same model and E galaxy SED
Figure 11.
as in Fig. 10 but in different units). The broad band fluxes representing the average of many
E galaxies in the Coma cluster (solid squares) from A. Stanford (private communication) are
shown. The observed SED is the one with the lowest spectral resolution.
18
Gustavo Bruzual A.: Population Synthesis at low and high z
Figure 12. Fits to an average Elliptical galaxy SED. This figure shows a closer look at the
same data of Fig. 11 in an enlarged scale. The discrepant line in corresponds to the same model
shown in Figs 10 and 11 but using the LCB97 synthetic stellar atlas instead of the empirical
stellar SEDs.
evolutionary stages to the bolometric light, and to the broad-band U BV RIKL fluxes for
a Z = Z⊙ model SSP computed for the Salpeter IMF (mL = 0.1, mU = 125 M⊙) using
the P tracks and the Pickles (1998) stellar atlas. The meaning of each line is indicated
in the top central frame. Fig. 16 shows the corresponding plot for an equivalent model
computed according to the G tracks. The contribution of the RGB stars is higher in the
P track model than in the G track model. Correspondingly, the AGB stars contribute
less in the P track model than in the G track model. For instance, for t > 1 Gyr, RGB
and AGB stars contribute 40% and 10%, respectively, to the bolometric light in the P
track model (Fig. 15). These fractions change to 30% and 20% in the G track model
(Fig. 16). These differences are seeing more clearly in Fig. 17 which shows the ratio
of the fractional contribution by different stellar groups in the G track model to that in
the P track model. According to the fuel consumption theorem (Renzini 1981), these
numbers reflect relatively large differences in the amount of fuel used up in the RGB and
AGB phases by stars of the same mass and initial chemical composition depending on
the stellar evolutionary code.
Fig. 20 shows the difference in B magnitude and B − V and V − K color between a G
track model SSP and a P track model SSP as seen both in the rest frame of the galaxy
(vs. galaxy age in the left hand side panels) and the observer frame (vs. redshift in the
right hand side panels). These differences reach quite substantial values. The observer
frame quantities include both the k and the evolutionary corrections. Here and elsewhere
in this paper I assume H0 = 65 km s−1 Mpc−1, Ω = 0.10, and the age of galaxies to be
tg = 12 Gyr.
Gustavo Bruzual A.: Population Synthesis at low and high z
19
Figure 13. Best fit to average optical SED (heavy line) of star clusters of various metallicities
compiled by Bica et al.
(1996a). The SSP model is the thin line extending over the full
wavelength range. The residuals of the fit, log Fλ(observed) − log Fλ(model), are shown as a
function of wavelength. The name and the metal content of the observed spectra indicated in
each panel is as given by Bica et al. The quoted age is derived from the best fit of our model
spectra for the indicated metallicity to the corresponding observations.
Figure 14. Same as Fig. 13, but for four different star clusters.
20
Gustavo Bruzual A.: Population Synthesis at low and high z
Figure 15. Contribution of stars in various evolutionary stages to the bolometric light, and
to the broad-band U BV RIKL fluxes for a Z = Z⊙ model SSP computed for the Salpeter IMF
(mL = 0.1, mU = 125 M⊙) using the P tracks and the Pickles (1998) stellar atlas.
Figure 16. Contribution of stars in various evolutionary stages to the bolometric light, and
to the broad-band U BV RIKL fluxes for a Z = Z⊙ model SSP computed for the Salpeter IMF
(mL = 0.1, mU = 125 M⊙) using the G tracks and the Pickles (1998) stellar atlas.
Gustavo Bruzual A.: Population Synthesis at low and high z
21
11.2. On the energetics of model stellar populations
Buzzoni (1999) has argued that most population synthesis models violate basic prescrip-
tions from the fuel consumption theorem (FCT). Fig. 18 should be compared to Fig. 2
of Buzzoni (1999). The line with square dots along it is reproduced from Buzzoni's Fig.
2. The other lines show the dependence of the ratio of the Post-MS to MS contribution
to the bolometric flux for different models. The heavy lines correspond to the Salpeter
IMF models. The thin lines to the Scalo IMF models. The solid lines correspond to
P track models, whereas the dashed lines correspond to the G track models. The G
track model for the Salpeter IMF (heavy dashed line) is in quite good agreement with
Buzzoni's model for t > 5 Gyr.
Fig. 19 (after Buzzoni's Fig. 1) plots the M/LV ratio vs. the Post-MS to MS contribu-
tion in the V band. The open dots correspond to the models shown in Buzzoni's Fig. 1.
The solid dot is Buzzoni's model marked B in his Fig. 1. The solid triangles correspond
to our Z⊙ SSP models for various stellar atlas using the P tracks and the Salpeter IMF.
The open triangles are for the same models but for the Scalo IMF. The solid pentagons
represent the G track models for the Salpeter IMF and the open pentagons the same
models but for the Scalo IMF. The three solid squares joined by a line represent sub-
solar metallicity models for the P tracks and the Salpeter IMF. The three open squares
joined by a line are for identical models using the Scalo IMF. Fig. 19 shows clearly that
the position of points representing various models in this diagram is a strong function
of the stellar IMF, the set of evolutionary tracks, and the chemical composition of the
stellar population. It may be too simplistic to attribute the dispersion of the points to a
violation of the FCT (Buzzoni 1999).
11.3. Uncertainties in the stellar IMF
It is constructive to compare models computed for identical ingredients except for the
stellar IMF. Fig. 21 shows the results of such a comparison. Brightness and color
differences with respect to the Salpeter IMF model SSP are shown vs. galaxy age in the
galaxy rest frame (LHS panels) and vs. redshift in the observer frame (RHS panels) for
SSP P track models computed for the following IMFs: Scalo (1986, solid line), Miller &
Scalo (1979, short dashed line), and Kroupa et al. (1993, long dashed line).
Fig. 22 compares in the same format as before the results of using different solar
metallicity stellar libraries for a P track SSP model. Brightness and color differences
with respect to the SSP model computed with the Pickles (1998) stellar atlas are shown
vs. galaxy age in the galaxy rest frame (LHS panels) and vs. redshift in the observer
frame (RHS panels) for SSP P track models computed for the following stellar libraries:
Extended Gunn & Stryker atlas used by BC93 (solid line), LCB97 uncorrected atlas
(short dashed line), and LCB97 corrected atlas (long dashed line).
11.4. Different chemical composition
In this section we explore the differences between SSP models for non-solar composition
and the solar case. Fig. 23 shows the brightness and color differences with respect to
the SSP model for Z = Z⊙. All models shown in this figure are for the P tracks and use
the LCB97 stellar atlas. The lines in this figure have the following meaning: Z = 0.0001
(solid line), Z = 0.0004 (short dashed line), Z = 0.004 (long dashed line), Z = 0.008 (dot
- short dashed line), Z = 0.05 (short dash - long dashed line), and Z = 0.1 (dot - long
dashed line).
22
Gustavo Bruzual A.: Population Synthesis at low and high z
Figure 17. Ratio of the fractional contribution by different stellar groups in the G track
model to that in the P track model.
Ratio of the Post-MS to MS contribution to the bolometric flux vs. age for
Figure 18.
different models. This figure should be compared to Fig. 2 of Buzzoni (1999). The line with
square dots along it is reproduced from Buzzoni's Fig. 2. The heavy lines correspond to the
Salpeter IMF models. The thin lines to the Scalo IMF models. The solid lines correspond to P
track models, whereas the dashed lines correspond to the G track models. The G track model
for the Salpeter IMF (heavy dashed line) is in quite good agreement with Buzzoni's model for
t > 5 Gyr.
Gustavo Bruzual A.: Population Synthesis at low and high z
23
Figure 19. M/LV ratio vs. the Post-MS to MS contribution in the V band. This figure should
be compared to Fig. 1 of Buzzoni (1999). The open dots correspond to the models shown in
Buzzoni's Fig. 1. The solid dot is Buzzoni's model marked B in his Fig. 1. The solid triangles
correspond to our Z⊙ SSP models for various stellar atlas using the P tracks and the Salpeter
IMF. The open triangles are for the same models but for the Scalo IMF. The solid pentagons
represent the G track models for the Salpeter IMF and the open pentagons the same models
but for the Scalo IMF. The three solid squares joined by a line represent sub-solar metallicity
models for the P tracks and the Salpeter IMF. The three open squares joined by a line are for
identical models using the Scalo IMF.
11.5. Different history of chemical evolution
Fig. 24 shows three possible chemical evolutionary histories, Z(t), quick (ZQ, short
dashed line), linear (ZL, solid line), and slow (ZS, long dashed line), that reach Z =
0.1 = 5 × Z⊙ at 15 Gyr. The dotted lines indicate Z⊙ and tg = 12 Gyr. I have computed
models for a SFR Ψ(t) ∝ exp(−t/τ ), with τ = 5 Gyr, which evolve chemically accordingly
to the lines shown in Fig. 24. The difference in brightness and color of these three models
with respect to a Z = Z⊙ model for the same SFR are shown in Fig. 25. The meaning
of the lines is as follows: Z(t) = ZQ (short dashed line), Z(t) = ZL (solid line), and
Z(t) = ZS (long dashed line).
11.6. Evolution in the observer frame at various cosmological epochs
In Figs. 26 to 31, I summarize the range of values expected in the measured (V − R)
and (V − K) colors in the observer frame at various redshifts z as a function of galaxy
age. In these figures, the panel marked T RACKS shows the range of colors obtained for
solar metallicity SSP models computed using the Pickles empirical stellar atlas with the
Salpeter IMF for the P and the G tracks. In the panel marked IM F I show Z = Z⊙ SSP
24
Gustavo Bruzual A.: Population Synthesis at low and high z
Figure 20. Difference in B magnitude and B − V and V − K color between a G track model
SSP and a P track model SSP, vs. age in the galaxy rest frame (LHS panels) and vs. redshift
in the observer frame (RHS panels).
Figure 21. Difference in B magnitude and B − V and V − K color for various IMF, P track
model SSP's, with respect to the Salpeter IMF model, vs. galaxy age in the galaxy rest frame
(LHS panels) and vs. redshift in the observer frame (RHS panels).
Gustavo Bruzual A.: Population Synthesis at low and high z
25
Figure 22. Difference in B magnitude and B − V and V − K color for P track model SSP's
computed for various Z = Z⊙ stellar libraries, with respect to the SSP model computed with
the Pickles (1998) stellar atlas, vs. galaxy age in the galaxy rest frame (LHS panels) and vs.
redshift in the observer frame (RHS panels).
Figure 23. Differences between SSP models for non-solar composition and the solar case, vs.
galaxy age in the galaxy rest frame (LHS panels) and vs. redshift in the observer frame (RHS
panels).
26
Gustavo Bruzual A.: Population Synthesis at low and high z
Figure 24. Three possible chemical evolutionary histories, Z(t), quick (ZQ, short dashed line),
linear (ZL, solid line), and slow (ZS, long dashed line), that reach Z = 0.1 = 5 × Z⊙ at 15 Gyr.
Figure 25. Difference in brightness and color of three models that evolve chemically according
to Fig. 24 with respect to a Z = Z⊙ model for the same SFR, vs. galaxy age in the galaxy rest
frame (LHS panels) and vs. redshift in the observer frame (RHS panels).
Gustavo Bruzual A.: Population Synthesis at low and high z
27
models computed for the P tracks, the Pickles stellar atlas, and the Salpeter, the Scalo,
and the Miller-Scalo IMFs. The panel marked SEDs shows the evolution of Z = Z⊙
SSP models computed with the P tracks and the Salpeter IMF, using the empirical
Gunn-Stryker and Pickles stellar libraries, as well as the original and corrected versions
of the LCB atlas for Z = Z⊙. The panel marked SF R shows the evolution of an SSP
model together with a model in which stars form at a constant rate during the first Gyr
in the life of the galaxy (1 Gyr model), both computed with the P tracks, Salpeter IMF,
and the Pickles stellar library. The panel marked Z shows the range of colors covered
by SSP models of metallicity Z = 0.004, 0.008 and 0.02 (solar), computed with the P
tracks and the Salpeter IMF. In the solar case, I repeat the models shown in the panel
marked SEDs. The panel marked ALL summarizes the results of the previous panels.
The reddest color obtained at any age in the previous 5 panels is shown as the top solid
line. The bluest color is shown as a dotted line. The average color is indicated by the
solid line between these two extremes. The 1 Gyr model is shown as a dashed line to
show the dominant effects of star formation in galaxy colors.
Figs. 26 and 27 show the range of values expected in (V − R) and (V − K) in the
observer frame at z = 0 as a function of galaxy age. The maximum age allowed is the
age of the universe, tu = 13.5 Gyr at z = 0 using H0 = 65 km s−1 Mpc−1, Ω = 0.10.
Figs. 28 and 29 show the same quantities but for galaxies seen at z = 1.552. The age of
the universe for this z in this cosmology is tu = 4.6 Gyr. Figs. 30 and 31 correspond to
z = 3, in this case tu = 2.7 Gyr. From Figs. 26 to 31 I conclude that metallicity Z and
the SFR are the most dominant factors determining the range of allowed colors.
The horizontal lines shown across the panels in Figs. 28 and 29 indicate the color ±σ
of the galaxy LBDS 53W091 observed by Spinrad et al. (1997). Our models reproduce
the colors of this galaxy at an age close to 1.5 Gyr.
Figs. 32 to 35 are based on the panel marked ALL of Figs. 26 to 29, and similar figures
for (V − U ), (V − B), (V − I), and (V − J) not shown in this work. To build these figures
I have subtracted from each line in the previous figures, the color of the Z = Z⊙ SSP
model computed with the P tracks, the Salpeter IMF, and the Pickles stellar library. We
conclude that the evolution of (V − R) is less model dependent than for any other color
shown in these figures.
Figs. 36 to 39 are also based on the panel marked ALL of Figs. 26 to 29 and similar
figures for other values of z not shown in this work. Figs. 36 to 39 show the evolution in
time of (V − U ), (V − B), (V − R), and (V − K) in the observer frame for several values
of the redshift z. The color of the Z = Z⊙ SSP model computed with the P tracks, the
Salpeter IMF, and the Pickles stellar library has been subtracted from the lines in the
previous figures. Again, (V − R) shows less variations with model than the other colors.
12. Summary and Conclusions
Present population synthesis models show reasonable agreement with the observed
spectrum of stellar populations of various ages and metal content. Differences in results
from different codes can be understood in terms of the different ingredients used to
build the models and do not necessarily represent violations of physical principles by
some of these models. However, inspection of Fig. 20 shows that two different sets
of evolutionary tracks for stars of the same metallicity produce models that at early
ages differ in brightness and color from 0.5 to 1 mag, depending on the specific bands.
The differences decrease at present ages in the rest frame, but are large in the observer
frame at z > 2. Thus any attempt to date distant galaxies, for instance, based on fitting
observed colors to these lines will produce ages that depend critically on the set of models
28
Gustavo Bruzual A.: Population Synthesis at low and high z
Figure 26.
(V − R) vs. time in the observer frame at z = 0. See §11.6 for details.
Figure 27.
(V − K) vs. time in the observer frame at z = 0. See §11.6 for details.
Gustavo Bruzual A.: Population Synthesis at low and high z
29
Figure 28.
(V − R) vs. time in the observer frame at z = 1.552. See §11.6 for details.
Figure 29.
(V − K) vs. time in the observer frame at z = 1.552. See §11.6 for details.
30
Gustavo Bruzual A.: Population Synthesis at low and high z
Figure 30.
(V − R) vs. time in the observer frame at z = 3. See §11.6 for details.
Figure 31.
(V − K) vs. time in the observer frame at z = 3. See §11.6 for details.
Gustavo Bruzual A.: Population Synthesis at low and high z
31
Figure 32. Color vs. time in the observer frame at z = 0. The color of the Z = Z⊙ SSP
model computed with the P tracks, the Salpeter IMF, and the Pickles stellar library have been
subtracted from each line. See §11.6 for details.
Figure 33. Color vs. time in the observer frame at z = 1. The color of the Z = Z⊙ SSP
model computed with the P tracks, the Salpeter IMF, and the Pickles stellar library have been
subtracted from each line. See §11.6 for details.
32
Gustavo Bruzual A.: Population Synthesis at low and high z
Figure 34. Color vs. time in the observer frame at z = 2. The color of the Z = Z⊙ SSP
model computed with the P tracks, the Salpeter IMF, and the Pickles stellar library have been
subtracted from each line. See §11.6 for details.
Figure 35. Color vs. time in the observer frame at z = 3. The color of the Z = Z⊙ SSP
model computed with the P tracks, the Salpeter IMF, and the Pickles stellar library have been
subtracted from each line. See §11.6 for details.
Gustavo Bruzual A.: Population Synthesis at low and high z
33
(V − U ) vs. time in the observer frame for various values of z. The color of
Figure 36.
the Z = Z⊙ SSP model computed with the P tracks, the Salpeter IMF, and the Pickles stellar
library have been subtracted from each line. See §11.6 for details.
(V − B) vs. time in the observer frame for various values of z. The color of
Figure 37.
the Z = Z⊙ SSP model computed with the P tracks, the Salpeter IMF, and the Pickles stellar
library have been subtracted from each line. See §11.6 for details.
34
Gustavo Bruzual A.: Population Synthesis at low and high z
(V − R) vs. time in the observer frame for various values of z. The color of
Figure 38.
the Z = Z⊙ SSP model computed with the P tracks, the Salpeter IMF, and the Pickles stellar
library have been subtracted from each line. See §11.6 for details.
(V − K) vs. time in the observer frame for various values of z. The color of
Figure 39.
the Z = Z⊙ SSP model computed with the P tracks, the Salpeter IMF, and the Pickles stellar
library have been subtracted from each line. See §11.6 for details.
Gustavo Bruzual A.: Population Synthesis at low and high z
35
which is used. Note that from z of 3 to 3.5 (V − K) in the two models differs by more
than 1 mag. This difference is produced by the corresponding difference between the
models seen in the rest frame at 10 Myr. From Figs. 15 and 16 these differences can be
understood in terms of the different contribution of the same stellar groups to the total
V and K flux in the two models.
Even though at the present age models built with different IMFs show reasonably
similar colors and brightness, the early evolution of these models is quite different at
early ages (Fig. 21), resulting in larger color differences in the observer frame at z >
2. Thus, the more we know about the IMF, the better the model predictions can be
constrained. The small color differences seen in the rest frame when different stellar
libraries of the same metallicity are used, are magnified in the observer frame (Fig. 22).
When the k correction brings opposing flux differences into each filter, the difference in
the resulting color is enhanced. Fig. 23 shows the danger of interpreting data for one
stellar system with models of the wrong metallicity. The color differences between these
models, especially in the observer frame, are so large as to make any conclusion thus
derived very uncertain.
It is common practice to use solar metallicity models when no information is available
about the chemical abundance of a given stellar system. Galaxies evolving according the
Z(t) laws of Fig. 24 show color differences with respect to the Z = Z⊙ model which are
not larger than the differences introduced by the other sources of uncertainties discussed
so far. Hence, the solar metallicity approximation may be justified in some instances.
The color differences between the chemically inhomogeneous composite population and
the purely solar case (Fig. 25), are much smaller than the ones shown in Fig. 23 for
chemically homogeneous SSPs.
Figs. 26 to 39 indicate that some colors, especially (V − R), when measured in the
observer frame are less sensitive to model predictions than other colors. From Figs. 26
to 31, metallicity Z and the SFR are the most dominant factors determining the range
of allowed colors.
I expect that through these simple examples the reader can get a feeling of the kind of
uncertainties introduced by the many ingredients entering the stellar population synthesis
problem, and that he or she will be motivated to try his or her own error estimates when
using these models.
REFERENCES
Allard, F., & Hauschildt, P.H. 1995, ApJ, 445, 433
Alongi, M., Bertelli, G., Bressan, A., Chiosi, C., Fagotto, F., Greggio, L., & Nasi, E. 1993,
A&AS, 97, 851
Arag´on-Salamanca, A., Ellis, R.S.E., Couch, W. J., Carter, D. 1993, MNRAS, 262, 764A
Arimoto, N., & Yoshii, Y. 1987, A&A, 173, 23
Barbuy, B., Ortolani, S., Bica, E., Renzini, A., & Guarnieri, M.D. 1997, in IAU Symp. 189,
Fundamental Stellar Parameters: Confrontation Between Observation and Theory, eds. J.
Davis, A. Booth & T. Bedding, Kluwer Acad. Pub., p. 203
Bender, R., Ziegler, B., & Bruzual A., G. 1996, ApJ Letters, 463, L51
Bessell, M.S., Brett, J., Scholtz, M., & Wood, P. 1989, A&AS, 77, 1
-- -- -- . 1991, A&AS, 89, 335
Bica, E., & Alloin, D. 1986, A&A, 162, 21
-- -- -- . 1987, A&A, 186, 49
Bica, E., Alloin, D., & Schmitt, H. 1994, A&A, 283, 805
36
Gustavo Bruzual A.: Population Synthesis at low and high z
Bica, E., Alloin, D., Bonatto, C., Pastoriza, M.G., Jablonka, P., Schmidt, A., & Schmitt, H.R.
1996a, in A Data Base for Galaxy Evolution Modeling, eds. C. Leitherer et al., PASP, 108,
996
Bica, E., Clari´a, J.J., Dottori, H., Santos Jr., J.F.C., Piatti, A. E. 1996b, ApJS, 102, 57
Bressan, A., Chiosi, C., & Fagotto, F. 1994, ApJS, 94, 63
Bressan, A., Fagotto, F., Bertelli, G., & Chiosi, C. 1993, A&AS, 100, 647
Bruzual A., G. 1998,
in The Evolution of Galaxies on Cosmological Time scales, eds. J.E.
Beckman and T.J. Mahoney, ASP Conference Series, Vol. 187, p. 245
-- -- -- . 1999, in The Hy-Redshift Universe: Galaxy Formation and Evolution at High Redshift,
eds. A. J. Bunker and W. J. M. van Breugel, ASP Conference Series, Vol. 193, p. 121
-- -- -- . 2000, in Euroconference on The Evolution of Galaxies, I- Observational Clues, eds. J.M.
V´ılchez, G. Stasinska, and E. P´erez, Kluwer Academic Publisher, in press
Bruzual A., G., Barbuy, B., Ortolani, S., Bica, E., Cuisinier, F., Lejeune, T., & Schiavon, R.
1997, AJ, 114, 1531
Bruzual A., G. & Charlot, S. 1993, ApJ, 405, 538 (BC93)
-- -- -- . 2000, ApJ, in preparation (BC2000)
Burstein, D., Bertola, F., Buson, L.M., Faber, S.M., and Lauer, T.R. 1988, ApJ, 328, 440
Buzzoni, A. 1989, ApJS, 71, 817
-- -- -- . 1999, in IAU Symposium No. 183 Cosmological Parameters and the Evolution of the
Universe, ed. K. Sato, Dordrecht: Kluwer, p. 134
Charlot, S., and Bruzual A., G. 1991, ApJ, 367, 126 (CB91)
Cool, A.M. 1997, in Advances in Stellar Evolution, eds. R. T. Rood and A. Renzini, Cambridge
University Press, p. 191
D'Antona, F. 1999, in The Galactic Halo: from Globular Clusters to Field Stars, 35th Liege Int.
Astroph. Colloquium, astro-ph/9910312
Eggen, O.J., and Sandage, A.R. 1964, ApJ, 140, 130
Fagotto, F., Bressan, A., Bertelli, G., & Chiosi, C. 1994a, A&AS, 100, 647
-- -- -- . 1994b, A&AS, 104, 365
-- -- -- . 1994c, A&AS, 105, 29
Fluks, M. et al. 1994, A&AS, 105, 311
Fritze-v.Alvensleben, U. & Gerhard, O.E. 1994, A&A, 285, 751
Gilliland, R.L., Brown, T.M., Duncan, D.K., Suntzeff, N.B., Wesley Lockwood, G., Thompson,
D.T., Schild, R.E., Jeffrey, W.A., and Penprase, B.E., 1991, AJ, 101, 541
Girardi, L., Bressan, A., Chiosi, C., Bertelli, G., & Nasi, E. 1996, A&AS, 117, 113
Girardi, L., Bressan, A., Bertelli, G., & Chiosi, C. 2000, A&AS, 141, 371
Greggio, L., and Renzini, A., 1990, ApJ, 364, 35
Guarnieri, M.D., Ortolani, S., Montegriffo, P., Renzini, A.,
Guiderdoni, B. & Rocca-Volmerange, B. 1987, A&A, 186, 1
Gunn, J.E., and Stryker, L.L., 1983, ApJS, 52, 121
Iglesias, C.A., Rogers, F.J., & Wilson, B.G. 1992, ApJ, 397, 717
Janes, K.A., 1985, in Calibration of Fundamental Stellar Quantities, IAU Symposium No. 111,
D.S. Hayes, L.E. Pasinetti, and A.G. Davis Philip, (Dordrecht: Reidel), 361
Janes, K.A., and Smith, G.H., 1984, AJ, 89, 487
Kaluzny, J. 1997, A&AS, 121, 455
King, I.R., Anderson, J., Cool, A.M., Piotto, G. 1998, ApJ, 492, L37
Kroupa, P., Tout, C.A., & Gilmore, G. 1993, MNRAS, 262, 545
Kurucz, R. 1995, private communication
Lejeune, T., Cuisinier, F., & Buser, R. 1997, A&AS, 125, 229 (LCB97)
Gustavo Bruzual A.: Population Synthesis at low and high z
37
-- -- -- . 1998, A&AS, 130, 65 (LCB98)
Metcalfe, N., Shanks, T., Fong, R., Gardner, J., Roche, N. 1996, IAU Symp. 171, p. 225
Micela, G., Sciortino, S., Vaiana, G.S., Schmitt, J.H.M.M., Stern, R.A., Harnden, F.R., Jr., and
Rosner, R., 1988, ApJ, 325, 798
Miller, G.E. & Scalo, J.M. 1979, 41, 513
Peterson, D.M., & Solensky, R. 1988, ApJ, 333, 256
Pickles, A.J. 1998, PASP, 110, 863
Pozzetti, L., Bruzual A., G., Zamorani, G. 1996, MNRAS, 281, 953
Racine, R., 1971, ApJ, 168, 393
Renzini, A., 1981, Ann. Phys. Fr., 6, 87
Salpeter, E.E. 1955, ApJ, 121, 161
Santos, J.F.C.Jr., Bica, E., Dottori, H., Ortolani, S., & Barbuy, B. 1995, A&A, 303, 753
Scalo, J.M. 1986, Fund. Cosmic Phys, 11, 1
Spinrad, H., Dey, A., Stern, D., Dunlop, J., Peacock, J., Jim´enez, R., Windhorst, R. 1997, ApJ,
484, 581
Stanford, S.A., Eisenhardt, P.R., & Dickinson, M. 1995, ApJ, 450, 512
-- -- -- . 1998, ApJ, 492, 461
Upgren, A.R., 1974, ApJ, 193, 359
Upgren, A.R., and Weis, E.W., 1977, AJ, 82, 978
Worthey, G. 1994, ApJS, 95, 107
Worthey, G., Faber, S.M., Gonz´alez, J.J., & Burstein, D. 1994, ApJS, 94, 687
|
astro-ph/0405138 | 1 | 0405 | 2004-05-07T09:33:27 | A Survey of z>5.7 Quasars in the Sloan Digital Sky Survey III: Discovery of Five Additional Quasars | [
"astro-ph"
] | We present the discovery of five new quasars at z>5.7, selected from the multicolor imaging data of the Sloan Digital Sky Survey (SDSS). Three of them, at redshifts 5.93, 6.07, and 6.22, were selected from ~1700 deg^2 of new SDSS Main Survey imaging in the Northern Galactic Cap. An additional quasar, at redshift 5.85, was discovered by coadding the data obtained in the Fall Equatorial Stripe in the SDSS Southern Survey Region. The fifth object, at redshift 5.80, is selected from a non-standard SDSS scan in the Southern Galactic Cap outside the Main Survey area. The spectrum of SDSS J162331.81+311200.5 (z=6.22) shows a complete Gunn-Peterson trough at z_abs > 5.95, similar to the troughs detected in other three z>6.2 quasars known. We present a composite spectrum of the z>5.7 quasars discovered in the SDSS to date. The average emission line and continuum properties of z~6 quasars exhibit no significant evolution compared to those at low redshift. Using a complete sample of nine z>5.7 quasars, we find that the density of quasars with M_1450 < -26.7 at z~6 is (6+/-2) x 10^-10 per Mpc^3 consistent with our previous estimates. The luminosity distribution of the sample is fit with a power law luminosity function Psi(L) ~ L^(-3.2+/-0.7), somewhat steeper than but consistent with our previous estimates. | astro-ph | astro-ph |
AJ, in press (Aug, 2004)
A Survey of z > 5.7 Quasars in the Sloan Digital Sky Survey III: Discovery of
Five Additional Quasars1
Xiaohui Fan2,3 , Joseph F. Hennawi3,4, , Gordon T. Richards4 , Michael A. Strauss4 , Donald P.
Schneider5 , Jennifer L. Donley2 , Jason E. Young2 , James Annis6 , Huan Lin6 , Hubert Lampeitl6 ,
Robert H. Lupton4 , James E. Gunn4 , Gillan R. Knapp4 , W. N. Brandt5 , Scott Anderson6 , Neta
A. Bahcall4 , Jon Brinkmann7 , Robert J. Brunner8 , Masataka Fukugita7 , Alexander S. Szalay9 ,
Gyula P. Szokoly10 , Donald G. York11
ABSTRACT
We present the discovery of five new quasars at z > 5.7, selected from the multicolor
imaging data of the Sloan Digital Sky Survey (SDSS). Three of them, at redshifts 5.93,
1Based on observations obtained with the Sloan Digital Sky Survey, and with the Apache Point Observatory
3.5-meter telescope, which is owned and operated by the Astrophysical Research Consortium; and with the MMT
Observatory, a joint facility of the University of Arizona and the Smithsonian Institution, with the Univeristy of
Arizona 2.3-meter Bok Telescope, with the Kitt Peak National Observatory 4-meter Mayall Telescope, with the
6.5-meter Landon Clay Telescope at the Las Campanas Observatory, a collaboration between the Observatories
of the Carnegie Institution of Washington, University of Arizona, Harvard University, University of Michigan,
and Massachusetts Institute of Technology , and with the Hobby-Eberly Telescope, which is a joint pro ject of
the University of Texas at Austin, the Pennsylvania State University, Stanford University, Ludwig-Maximillians-
Universitat Munchen, and Georg-August-Universitat Gottingen.
2Steward Observatory, The University of Arizona, Tucson, AZ 85721
3Visiting Astronomer, Kitt Peak National Observatory, National Optical Astronomy Observatory, which is
operated by the Association of Universities for Research in Astronomy, Inc. (AURA) under cooperative agreement
with the National Science Foundation.
4Princeton University Observatory, Princeton, NJ 08544
5Department of Astronomy and Astrophysics, The Pennsylvania State University, University Park, PA 16802
6Fermi National Accelerator Laboratory, P. O. Box 500, Batavia, IL 60510
6University of Washington, Department of Astronomy, Box 351580, Seattle, WA 98195
7Apache Point Observatory, P. O. Box 59, Sunspot, NM 88349-0059
7 Institute for Cosmic Ray Research, University of Tokyo, Midori, Tanashi, Tokyo 188-8502, Japan
8Dept. of Astronomy & NCSA, University of Illinois, 1002 W. Green Street, Urbana, IL 61801
9Department of Physics and Astronomy, The Johns Hopkins University, Baltimore, MD 21218, USA
10Max-Planck-Institut fur extraterrestrische Physik, Postfach 1312, 85741 Garching, Germany
11University of Chicago, Astronomy & Astrophysics Center, 5640 S. Ellis Ave., Chicago, IL 60637
– 2 –
6.07, and 6.22, were selected from ∼ 1700 deg2 of new SDSS Main Survey imaging
in the Northern Galactic Cap. An additional quasar, at redshift 5.85, was discovered
by coadding the data obtained in the Fall Equatorial Stripe in the SDSS Southern
Survey Region. The fifth ob ject, at redshift 5.80, is selected from a non-standard
SDSS scan in the Southern Galactic Cap outside the Main Survey area. The spectrum
of SDSS J162331.81+311200.5 (z = 6.22) shows a complete Gunn-Peterson trough at
zabs > 5.95, similar to the troughs detected in other three z ∼> 6.2 quasars known.
We present a composite spectrum of the z > 5.7 quasars discovered in the SDSS to
date. The average emission line and continuum properties of z ∼ 6 quasars exhibit no
significant evolution compared to those at low redshift. Using a complete sample of
nine z > 5.7 quasars, we find that the density of quasars with M1450 < −26.7 at z ∼ 6
is (6 ± 2) × 10−10 Mpc−3 (H0 = 65 km s−1Mpc−1 , Ω = 0.35 and Λ = 0.65), consistent
with our previous estimates. The luminosity distribution of the sample is fit with a
power law luminosity function Ψ(L) ∝ L−3.2±0.7 , somewhat steeper than but consistent
with our previous estimates.
Subject headings: quasars: general; quasars: emission line; quasars: absorption lines
1.
Introduction
This paper is the third in a series presenting i-dropout z ∼> 5.7 quasars selected from the
multicolor imaging data of the Sloan Digital Sky Survey (SDSS; York et al. 2000, Stoughton et
al. 2002, Abaza jian et al. 2003, 2004). In Fan et al. (2000) and in the first two papers of this
series (Fan et al. 2001c, Paper I, Fan et al. 2003, Paper II), we presented the discovery of seven
luminous quasars at z = 5.74 − 6.42, selected from ∼ 2900 deg2 of SDSS imaging in the Northern
Galactic Cap. In this paper, we describe the discovery of five quasars at z = 5.80, 5.85, 5.93, 6.07
and 6.22, respectively. The scientific ob jectives, photometric data reduction, target selection and
follow-up observation procedures are described in detail in Paper I. Three of the new quasars
were selected using the same color selection procedures described in Paper II, as outlined briefly
in §2.1. One ob ject was selected using the co-added catalog from multi-epoch imaging of the Fall
Equatorial Stripe in the SDSS Southern Survey region. The co-addition procedures are described
in §2.2. The final ob ject was selected in a non-standard SDSS scan in the Southern Galactic
Cap outside the main survey area. We present the spectroscopic follow-up observations of the
i-dropout candidates and the photometric and spectroscopic properties of the newly-discovered
quasars in §3. Combining the quasars in this paper and those presented in Fan et al. (2002) and
in Papers I and II, we construct the composite spectrum of z ∼ 6 quasars, and compare it with
the average quasar spectrum at low redshift (§4). Finally, we update our estimate of the evolution
of the high-redshift quasar luminosity function in §5.
Following the previous papers in this series, we use two cosmologies to present our results: (1)
– 3 –
H0 = 50 km s−1 Mpc−1 , ΩΛ = 0 and ΩM = 1 (Ω-model); (2) H0 = 65 km s−1 Mpc−1 , ΩΛ = 0.65
and ΩM = 0.35 (Λ-model). Coincidentally, the luminosity and number densities measured
under our Λ-model are within 1% of those in the WMAP (Spergel et al. 2003) cosmology with
H0 = 71 km s−1 Mpc−1 , ΩΛ = 0.73 and ΩM = 0.27.
2. Candidate Selection and Identification
The Sloan Digital Sky Survey is using a dedicated 2.5m telescope and a large format CCD
camera (Gunn et al. 1998) at the Apache Point Observatory in New Mexico to obtain images in
five broad bands (u, g , r , i and z , centered at 3551, 4686, 6166, 7480 and 8932 A, respectively;
Fukugita et al. 1996) of high Galactic latitude sky in the Northern Galactic Cap. About 6000 deg2
of sky have been imaged at the time of this writing (Mar 2004). The imaging data are processed
by the astrometric pipeline (Pier et al. 2003) and photometric pipeline (Lupton et al. 2001), and
are photometrically calibrated to a standard star network (Smith et al. 2002, see also Hogg et al.
2001). In addition to the Main Survey, the SDSS also obtains multi-epoch imaging in ∼ 270 deg2
along the Celestial Equator in the Southern Galactic Cap (the SDSS Southern Survey, York et al.
2000). At the time of this writing, the Southern Survey region has been imaged between 5 and 15
times, depending on the RA of the field. The multi-epoch data are used to study variable ob jects
and are co-added to reach fainter limiting magnitudes. We will use imaging from both the Main
Survey and the Southern Survey to search for quasar candidates.
2.1. Selection in the Main Survey Area
The SDSS quasar target selection pipeline (Richards et al. 2002) selects only quasar
candidates at z ∼< 5.5. The highest-redshift ob ject discovered in the SDSS spectroscopic survey is
at z = 5.4 (Anderson et al. 2001). At higher redshift, quasars become i-dropout ob jects; their
selections require additional follow-up observations. Papers I and II present the results from a
survey of i-dropout candidates selected from ∼ 2900 deg2 of SDSS Main Survey imaging carried
out in the Springs of 2000, 2001 and 2002. In Spring 2003, we searched for i-dropout quasar
candidates in 37 new SDSS imaging runs. These imaging data were taken between 10 October
2002 (Run 3358) and 29 April 2003 (Run 3919). We used the same criterion to decide which
photometric runs to include in the i-dropout survey as in Papers I and II: the z band image
quality, measured by the psfWidth parameter (= 1.06 FWHM for a Gaussian profile) in the fourth
column of the SDSS camera, should be better than 1.8′′ . The median seeing in the i and z bands
is ∼ 1.4′′ for the entire survey area (Abaza jian et al. 2003). There is overlap between adjacent
SDSS strips and stripes (York et al. 2000), meaning that these 37 new runs overlap somewhat
with the area covered in Papers I and II. Taking these overlaps into account, we find that the
total new area of the sky covered by these runs is 1708 deg2 , bringing the combined sky coverage
of Papers I, II and this paper to 4578 deg2 .
– 4 –
We applied the same color selection criteria as in Paper II (see Figure 1 and 2 in Paper II)
to the new SDSS imaging data to selection z > 5.7 quasar candidates. A total of 80 i-dropout
candidates were selected in the main survey area. The photometric and spectroscopic follow-up
observations were carried out over a number of nights between December 2002 and June 2003.
Independent z photometry was carried out using the Seaver Prototype Imaging camera (SPICAM)
in the SDSS z filter on the ARC 3.5m telescope at the Apache Point Observatory. J photometry
was carried out using the 256 × 256 NICMOS imager on Steward Observatory’s 2.3m Bok Telescope
at Kitt Peak, and using the GRIM II instrument (the near infrared GRIsm spectrometer and
IMager), also on the ARC 3.5m. The spectroscopic follow-up observations were obtained using
the Red-Channel Spectrograph on the MMT 6.5m telescope on Mt. Hopkins, the Double Imaging
Spectrograph (DIS) on the ARC 3.5m, the Multi-Aperture Red Spectrograph (MARS) on the
4-m telescope on Kitt Peak, and the Low Resolution Spectrograph (LRS, Hill et al. 1998) on the
Hobby-Eberly Telescope.
2.2. Selection in the SDSS Southern Survey Region
The SDSS Main Survey imaging consists of single epoch observations with exposure times
of 54.1 seconds. For the average seeing conditions (FWHM ∼ 1.4′′ ), the 5-σ limiting magnitudes
in the i and z bands are 22.5 and 20.5, respectively (Stoughton et al. 2002, Abaza jian et al.
2003). At this depth, the SDSS Main Survey only allows selection of i-dropout candidates at
z < 20.2, and therefore is only sensitive to the most luminous quasars at z ∼ 6 (MB ∼< −27).
The multi-epoch imaging obtained in the SDSS Southern Survey area will eventually provide
photometry ∼ 1.5 magnitudes deeper than the main survey over ∼ 270 deg2 , enabling selection of
much fainter quasar candidates. By the end of 2003, the SDSS Southern Survey Region had been
scanned 5 – 15 times, resulting in a dataset that goes more than one magnitude deeper than the
Main Survey.
In Fall 2002, we used a preliminary co-added catalog of ∼ 5 epoch imaging to select z ∼ 6
quasar candidates. The co-added catalog was generated by matching sources detected in more
than one SDSS run within a radius of 1′′ . The average flux is calculated from the SDSS asinh
magnitudes (Lupton, Gunn & Szalay 1999), weighted by the inverse of the flux variances. Note
that when co-adding at the catalog level, an ob ject is required to be detected in the individual
runs. Therefore, while the co-addition improves the S/N of faint sources, allowing us to select
candidates all the way down to the 6-σ detection limit, it does not allow measurement of ob jects
fainter than the detection limit of single SDSS exposures. Therefore, we can only select candidates
at z < 20.5 − 20.7. We searched for candidates from ∼ 100 deg2 of co-added catalogs. The
color selection criteria in the Southern Survey region are the same as those used in the Main
survey area. A number of faint i-dropout candidates were observed using Magellan II Clay
Telescope and the Boller and Chivens (B&C) spectrograph on 16 Oct 2002. One quasar, SDSS
– 5 –
J000552.34–000655.814 (hereafter SDSS J0005–0006, zAB = 20.5) was discovered at a redshift of
5.85.
2.3. Selection in the Constant-Longitude Scans
The SDSS has obtained some imaging data outside the survey boundary (York et al. 2000;
Finkbeiner et al. 2004), including a few strips in the Southern Galactic Cap that extend to low
Galactic latitude. We have also selected i-dropout candidates in these runs. One new quasar,
SDSS J000239.39+255034.8 (z = 5.80), was selected from Run 4152, a constant-longitude scan (at
l = 110◦ ), observed on 29 Sep 2003. The sample in these scans which go outside the main survey
area is not yet complete, and so we do not include this ob ject in our complete sample.
3. Discovery of Five New Quasars at z > 5.7
Among the 80 i-dropout candidates candidates in the Main Survey area, 15 are false z
band only detections which are most likely cosmic rays; 55 are M or L dwarfs (mostly classified
photometrically based on their red z ∗ − J colors); and 7 are likely T dwarfs. Several ob jects
still lack proper infrared spectroscopy, so the T dwarf classification is still preliminary. Three
of the candidates are identified as quasars at z > 5.7: SDSS J141111.29+121737.4 (z = 5.93,
SDSS J0002+2550 hereafter), SDSS J160254.18+422822.9 (z = 6.07, SDSS J1602+4228), and
SDSS J162331.81+311200.5 (z = 6.22, SDSS J1623+3112). The discovery spectra of the first two
ob jects were obtained using DIS on the ARC 3.5m in April 2003, and the discovery spectrum of
the last ob ject was obtained in June 2003 at MMT using the Red Channel spectrograph. We have
subsequently obtained longer exposures of these three quasars using the Red Channel on MMT
and MARS on the Kitt Peak 4-meter, with total exposure times of 2 – 4 hours each. These spectra
are shown in Figure 2. As discussed above, SDSS J000552.34–000655.8 (z = 5.85) was discovered
in the deep imaging of the SDSS Southern Survey region. Figure 2 shows the discovery spectrum,
a 40-min exposure obtained with the Clay telescope and B&C spectrograph in Oct 2002. The
discovery spectrum of SDSS J000239.39+255034.8 (z = 5.80, §2.3) was obtained at the MMT
using the Red Channel Spectrograph in Nov 2003. This one hour exposure is shown in Figure 2.
The finding charts of the five new quasars are presented in Figure 1. The spectra are
flux-calibrated to match the observed z band photometry. Table 1 presents the photometric
properties of the new quasars, and Table 2 presents the measurements of their continuum
properties. Following Papers I and II, the quantity AB1280 is defined as the AB magnitude of
the continuum at rest-frame 1280A, after correcting for interstellar extinction using the map
14The naming convention for SDSS sources is SDSS JHHMMSS.SS±DDMMSS.S, and the positions are expressed
in J2000.0 coordinates. The astrometry is accurate to better than 0.1′′ in each coordinate.
– 6 –
of Schlegel, Finkbeiner & Davis (1998). We extrapolate the continuum to rest-frame 1450A,
assuming a continuum shape fν ∝ ν −0.5 , to calculate AB1450 . None of the five quasars is detected
in the FIRST (Becker, White & Helfand 1995) or NVSS (Condon et al. 1998) radio surveys. The
discovery of these five new quasars brings the total of z > 5.7 quasars known to twelve, all selected
from SDSS imaging.
3.1. Notes on Individual Ob jects
SDSS J000239.39+255034.8 (z = 5.80). This ob ject is selected in a non-standard SDSS
scan in the Southern Galactic Cap. It is outside the main survey region, and the spectroscopic
follow-up of this region for i-dropout candidates is not yet complete. We therefore do not include
it in the luminosity function calculations below.
SDSS J0002+2550 has i = 21.47 and z = 18.99, but is undetected in 2MASS (Skrutskie
et al. 1997), indicating that J > 16.5 and z − J < 2.5. If it were an L dwarf, it would have a
z − J color of 2.5 or redder, so we targeted this ob ject as a high-redshift quasar candidate. SDSS
J1044–0125 (z = 5.74, Fan et al. 2000) was selected in a similar manner.
At zAB ∼ 19 and M1450 = −27.55 (Λ-model), SDSS J0002+2550 is very luminous. It is
the second brightest quasar known at z > 5.7 so far and provides an excellent target for high
resolution spectroscopic follow-up observations. The redshift of SDSS J0002+2550 is determined
by the locations of the peak of the Lyα emission line and of the OI 1300A emission line, and is
accurate to 0.02. The quasar has a Lyα+NV emission line rest-frame equivalent width of ∼ 60A,
comparable to that of most high-redshift quasars (Fan et al. 2001b).
SDSS J000552.34–000655.8 (z = 5.85). This ob ject is selected in the SDSS Southern Survey
region. At zAB = 20.54 ± 0.10, it is the faintest z > 5.7 quasar in our sample. Although the
discovery spectrum (Figure 2) has low S/N, the strong Lyα and NV emission lines and the strong
Lyman break are clearly visible. The redshift is determined by the locations of the Lyα and NV
emission line peaks, and is accurate to 0.02. The emission lines in this quasar appear to be quite
narrow: the Lyα and NV emission are clearly separated. We estimate a FWHM of about 1500 –
2500 km s−1 for individual lines, with large uncertainties due to the low S/N.
SDSS J141111.29+121737.4 (z = 5.93). The redshift is determined by the locations of the OI
1300A and NV 1240A lines and is accurate to 0.02. The ob ject has a moderately strong Lyα+NV
line (rest-frame equivalent width of ∼ 100A). This ob ject is detected in two overlapping SDSS
runs with consistent photometry (Table 1).
SDSS J160254.18+422822.9 (z = 6.07). The redshift of this ob ject is determined by the peaks
of the Lyα and NV 1240A lines and is accurate to 0.02. A weak OI 1300A line is also detected in
the spectrum. The quasar spectrum shows a number of dark patches in the Lyα and Lyβ forest,
although the S/N of the current spectrum is not sufficient to determine whether there is a short
– 7 –
Gunn-Peterson trough in the spectrum.
Due to its proximity on the sky to the high-redshift cluster candidate CL1603,
SDSS J1602+4228 serendipitously lies in a 28 ks pointed observation (sequence rp800239)
made with the ROSAT Position Sensitive Proportional Counter (PSPC). We do not find any
significant X-ray detection of SDSS J1602+4228 in these data. The 3σ upper limit on its
observed-frame, Galactic absorption-corrected, 0.5–2 keV flux is 5.9 × 10−14 erg cm−2 s−1 ,
adopting a power-law model with a photon index of Γ = 2 and the Galactic column density of
NH = 1.3 × 1020 cm−2 . Given the AB1450 magnitude of SDSS J1602+4228, comparison with
Figure 2 of Vignali et al. (2003) shows that the X-ray upper limit is consistent with X-ray
observations of other z > 4 quasars. The slope of a nominal power law between rest-frame 2500 A
and 2 keV is constrained to be αox > 1.2.
SDSS J162331.81+311200.5 (z = 6.22). This is the highest redshift quasar presented in this
paper, and is the third highest redshift quasar yet known. 15 This quasar has two striking features.
First, it has an extremely strong Lyα emission line. The total equivalent width of Lyα+NV is
∼> 150A in the rest-frame, without taking into account the strong absorption due to the Lyα forest
on the blue side of the Lyα emission. For comparison, Fan et al. (2001b) measured the mean and
standard deviation of the rest-frame Lyα+NV equivalent width of 69.3 ± 18.0A, based on a sample
of ∼ 40 quasars at z ∼ 4 selected from the SDSS (see also Schneider, Schmidt & Gunn 1991 and
§4). The line strength of this quasar is more than a factor of two larger than the average at high
redshift. It is one of the strongest-lined quasars at z > 4 yet known, and the strongest-lined quasar
at z > 5.
Second, SDSS J1623+3112 has a complete Gunn-Peterson (1965) trough. Following Becker et
al. (2001), Fan et al. (2002) and White et al. (2003), we define the transmitted flux ratio as:
T (zabs ) ≡ Df obs
ν E ,
ν /f con
(1 + zabs − 0.1) × 1216A < λ < (1 + zabs + 0.1) × 1216A,
(1)
where f con
is the continuum level extrapolated from the red side of Lyα emission. Using
ν
the spectrum presented in Figure 2, we find that at z = 6.05, the transmitted flux ratio
T (zabs = 5.95 − 6.15) = 0.004 ± 0.008, consistent with zero flux in the Gunn-Peterson trough
region. We also detect a complete Lyβ Gunn-Peterson trough in this quasar. This is the fourth
quasar with a complete Gunn-Peterson trough, after SDSS J1306+0524 (z = 6.28, Becker et al.
2002), SDSS J1148+5251 (z = 6.42, White et al. 2003) and SDSS J1048+4637 (z = 6.18, Fan et
al. 2003).
In al l quasars at z > 6.1, complete Gunn-Peterson troughs are detected, starting from
zabs = 5.95 ± 0.1, and extending to the highest redshift not affected by the quasar proximity effect.
15 SDSS J104845.05+463718.3 was reported to have a redshift of 6.23 in Fan et al. (2003). Subsequent observations
show that it is a Broad Absorption Line (BAL) quasar (Maiolino et al. 2004, Fan et al.
in preparation) and the
original redshift determination is biased; the best redshift estimate is z = 6.18 based on new observations.
– 8 –
The detection of a complete Gunn-Peterson trough in SDSS J1623+3112 further confirms the
rapid transition of the ionization state of the IGM at z ∼ 6 (e.g. Becker et al. 2001, Djorgovski et
al. 2001, Fan et al. 2002, White et al. 2003; see also Songaila 2004 for a different interpretation).
The S/N of the spectrum of SDSS J1623+3112 is considerably lower than that of the other
three quasars, therefore the optical depth limit we are able to place is not yet very strong. In a
subsequent paper, we will present a detailed analysis of the constraints on the evolution of IGM
properties using the five new quasars presented in this paper.
4. Quasar Composite Spectrum at z ∼ 6
The spectral energy distributions of luminous quasars show little evolution out to high
redshift. There is growing evidence from emission line ratio measurements that quasar broad
emission line regions have roughly solar or even higher metallicities at z > 4 (e.g., Hamann &
Ferland 1993, Dietrich et al. 2003a), similar to that in low redshift quasars. Dietrich et al. (2003b)
found the FeII/MgII ratio to have roughly the same value in a sample of z ∼ 5 quasars as at lower
redshift, suggesting that the metallicity of quasar emission line region remains high to even earlier
epochs.
The sample of twelve quasars at z > 5.7 from the SDSS provides the first opportunity to
study the evolution of quasar spectral properties at z ∼ 6, less than 1 Gyr after the Big Bang and
only 700 million years from the first star formation in the Universe (Kogut et al. 2003, Spergel et
al. 2003). Optical and infrared spectroscopy of some z ∼ 6 quasars already indicates a lack of
evolution in the spectral properties of these luminous quasars: Pentericci et al. (2002) show that
the CIV/NV ratio in two z ∼ 6 quasars are indicative of supersolar metallicity in these systems.
Freudling et al. (2003) and Barth et al. (2003) detected strong FeII emission in the spectra of four
z ∼ 6 SDSS quasars. In addition, the optical-to-X-ray flux ratios and X-ray continuum shapes
show at most mild evolution from low redshift (e.g. Brandt et al. 2002, Vignali et al. 2003). These
results, if confirmed with a larger sample, suggest that the accretion disk and photoionization
structure of quasars reached maturity very early on and are probably insensitive to the host galaxy
environment.
Figure 3 shows the composite of eleven of our twelve z > 5.7 quasar spectra. We omit SDSS
J0005-0006 due to its low S/N. To produce the composite, we simply redshift all the spectra to
zero, scale the continuum level of each quasar based on its m1450 magnitude, and average all the
scaled spectra with equal weighting. The composite covers rest-frame wavelengths from 1100A
to 1450A. Also plotted in Figure 3 is the low-redshift SDSS quasar composite of Vanden Berk et
al. (2001). The effective redshift of the low-redshift composite in this redshift range is z ∼ 2.
Blueward of Lyα emission, the strong IGM absorption at z ∼ 6 almost completely removes the
quasar flux. Redward of Lyα emission, there is no detectable difference in the intrinsic UV spectral
properties. The continuum shape is consistent with the power law fν ∝ ν −0.4 measured by Vanden
Berk et al. (2001). Clearly, spectral coverage in the near-IR is needed to put stronger constraints
– 9 –
on the continuum shape. Pentericci et al. (2003) use IR photometry of a sample of quasars at
z = 3.5 − 6 to measure a continuum shape of fν ∝ ν −0.5 , independent of redshift. The strengths
of the emission lines at z ∼ 6, including NV 1240A, OI 1300A, CII 1335A, and SiIV+OIV 1400A,
are also comparable to those at low redshift. This composite does not go red enough to cover the
CIV 1549A line, so we cannot test whether the Baldwin (1977) effect exists at these redshift. The
strength of the red wing of the Lyα emission line shows no evolution from that at low redshift.
The weaker Lyα emission in the blue wing is due to the strong IGM absorption. The average
FWHM of emission lines is ∼ 6000 km s−1 , also consistent with the low-redshift average.
5. Luminosity Function of z ∼ 6 Quasars
In Papers I and II, we estimated the comoving density of quasars at z ∼ 6 using a sample
of six quasars, covering a total area of 2870 deg2 . We repeat the calculation here, including
the additional three quasars in the complete sample; as explained above, SDSS 0002+2550 and
SDSS J0005-0006 were not selected as part of the flux-limited complete sample and will not be
included in the quasar luminosity function calculation. These nine quasars form a complete sample
satisfying the selection criteria in Eq. (1) over a total area of 4578 deg2 . Following Papers I and
II, we calculate the selection function of z ∼ 6 quasars using a Monte-Carlo simulation of quasar
colors, taking into account the distribution of quasar emission line and continuum properties, Lyα
absorption, the SDSS photometric errors and Galactic extinction. The selection function as a
function of redshift z and absolute magnitude M1450 is illustrated in Figure 7 of Paper II for the
Λ-model. The total spatial density of quasars at z ∼ 6 is derived using the 1/Va method. We
find that at the average redshift of hz i = 6.07, ρ(M1450 < −26.4) = (10.5 ± 3.9) × 10−10 Mpc−3
for the Ω = 1 model, and ρ(M1450 < −26.7) = (6.4 ± 2.4) × 10−10 Mpc−3 for the Λ-model. The
results, which are consistent with those in Paper II with smaller error bars, are plotted in Figure 4,
together with the measurements at lower redshifts from the 2dF survey (Boyle et al. 2000, Croom
et al. 2004) at z < 2.2, from Schmidt et al. (1995) at 2.7 < z < 4.8 and from Fan et al. (2001a) at
3.6 < z < 5.0. The comoving density of luminous quasars at z ∼ 6 is 30 times smaller than that
at z ∼ 3.
Following Paper II, we derive the bright-end slope from the luminosity distributions of the
sample using a maximum likelihood estimate. Assuming a single-power law luminosity function:
Ψ(M1450 ) = Ψ∗10−0.4[M1450 +26](β+1) ,
(2)
we find that for the Λ-model, Ψ∗ = (3.3+3.8
−1.6 ) × 10−9Mpc−3 . The best-fit bright-end slope is
β = −3.2, with a 68% confidence range of [–2.5, –4.0] and a 95% confidence range of [–2.2, –4.2].
This slope is steeper than that measured in Paper II, where the best-fit value was –2.3 with a
68% range of [–1.6, –3.1]. The reason for this 1-σ change is that all three new quasars included in
the sample have zAB ∼ 20, close to the detection limit. Their inclusion, when correcting for the
lower completeness at the faint end, drives the best-fit luminosity function to steeper slopes. For
– 10 –
example, the average selection probability of quasars with the redshift and luminosity of SDSS
J1623+3112 is of the order 10%; the strong Lyα emission boosts the z -band magnitude and makes
the z − J color bluer, making the selection of this ob ject easier than an average z = 6.2 quasar.
The best-fit slope omitting SDSS J1623+3112 is –2.8. The slopes derived here and in Paper II
differ only at the 1-σ level. The errors are large because of the small number of ob jects in the
sample, especially at the faint end. This underlines the need for a large sample to put strong
constraints on the quasar luminosity function at the highest redshift.
The slope of the quasar luminosity function has important implications. The total quantity
of ionizing photons emitted by the high-redshift quasar population is determined by the quasar
luminosity function. Fan et al. (2001) found that z ∼ 6 quasars are not likely to be the sources
responsible for reionizing the universe, or keeping the universe ionized at high redshift, assuming
a slope of β > −3.5. Interestingly, we are approaching this limit with the current analysis.
In a flux-limited sample, the lensing probability of the brightest observed ob jects is boosted
by magnification bias (e.g. Turner et al. 1984). The theoretical prediction of the fraction
of strongly-lensed quasars at high redshift could be be of order unity for a sufficiently steep
luminosity function (e.g. Wyithe & Loeb 2002a, Comerford, Haiman & Schaye 2002). Fan et al.
(2003) and Richards et al. (2004) used the lack of multiply-imaged quasars among the SDSS z ∼ 6
quasar sample to constrain the shape of the quasar luminosity function to be β > −4.6 at the 3-σ
level. The lensing fraction increases by a factor of ∼ 5 by assuming a slope of β = −3.3 rather
than β = −2.2 (Wyithe & Loeb 2002b). The quasar emissivity and lensing fraction also depends
strongly on the faint-end quasar slope, which is currently completely unknown.
In this paper, we present the first quasar discovered in the faint quasar survey using co-added
catalogs from the SDSS Southern Survey region. In Fall 2003, we also started to use co-added
images of the SDSS Southern Survey from ∼ 8 SDSS observations to select candidates. This
overcomes the limitation of the co-added catalog which requires the ob ject to be detected in
a single exposure. We have generated a preliminary photometric catalog using the SExtractor
software; the integration of the co-added imaging into the SDSS photometric pipeline is currently
underway. Using the catalogs generated by co-added imaging, we were able to recover SDSS
J000552.34–000655.8. The remaining candidate identification is still in progress and will be
reported in the future. In the next few years, the faint quasars selected from the Southern Survey
will be combined with the bright quasars in the Main Survey region to study the evolution of
quasar population at z ∼ 6 down to much fainter luminosities to probe the evolution of faint
quasars.
Funding for the creation and distribution of the SDSS Archive has been provided by the
Alfred P. Sloan Foundation, the Participating Institutions, the National Aeronautics and Space
Administration, the National Science Foundation, the U.S. Department of Energy, the Japanese
Monbukagakusho, and the Max Planck Society. The SDSS Web site is http://www.sdss.org/.
The SDSS is managed by the Astrophysical Research Consortium (ARC) for the Participating
– 11 –
Institutions. The Participating Institutions are The University of Chicago, Fermilab, the Institute
for Advanced Study, the Japan Participation Group, The Johns Hopkins University, Los Alamos
National Laboratory, the Max-Planck-Institute for Astronomy (MPIA), the Max-Planck-Institute
for Astrophysics (MPA), New Mexico State University, University of Pittsburgh, Princeton
University, the United States Naval Observatory, and the University of Washington. We thank the
staffs at Apache Point Observatory, the MMT, the Bok Telescope, Kitt Peak, the Hobby-Eberly
Telescope, and Magellan for their expert help. We acknowledge support from NSF grant AST
03-07384, a Sloan Research Fellowship and the University of Arizona (X.F.), NSF grants AST
00-71091 and AST 03-07409 (M.A.S.) and NSF grants AST 99-00703 and AST 03-07582 (D. P.
S.).
REFERENCES
Abaza jian, K., et al. 2003, AJ, 126, 2081
Abaza jian, K., et al. 2004, AJ, in press (astro-ph/0403325)
Anderson, S. F., et al. 2001, AJ, 122, 503
Baldwin, J. A. 1977, ApJ, 214, 679
Barth, A. J., Martini, P., Nelson, C. H., & Ho, L. C. 2003, ApJ, 594, L95
Becker, R. H., White, R. L., & Helfand, D. J. 1995, ApJ, 450, 559
Becker, R. H. et al. 2001, AJ, 122, 2850
Boyle, B.J., Shanks, T., Croom, S. M., Smith, R . J., Miller, L., Loaring, B., & Heymans, C. 2000,
MNRAS, 317, 1014
Brandt, W.N., et al. 2002, ApJ, 569, L5
Comerford, J., Haiman, Z., & Schaye, J. 2002, ApJ, 580, 63
Condon, J. J., Cotton, W. D., Greisen, E. W., Yin, Q. F., Perley, R. A., Taylor, G. B., &
Broderick, J. J., 1998, AJ, 115, 1693
Croom, S. M., Smith, R. J., Boyle, B. J., Shanks, T., Miller, L., Outram, P. J., & Loaring, N. S.,
2004, MNRAS, in press (astro-ph/0403040)
Dietrich, M., Appenzeller, I., Hamann, F., Heidt, J. Jaeger, J., Vestergaard, M., & Wagner, S. J.,
2003a, A&A, 398, 891
Dietrich, M., Hamann, F. Shields, J. C., Constantin, A., Heidt, J., Jaeger, M., Vestergaard, M., &
Wagner, S. J., 2003b, ApJ, 722, 732
Djorgovski, S. G., Castro, S., Stern, D., & Mahabal, A. A. 2001, ApJ, 560, L5
– 12 –
Fan, X. et al., 2000, AJ, 120, 1167
——, 2001a, AJ, 121, 31
——, 2001b, AJ, 121, 54
——, 2001c, AJ, 122, 2833
——, 2002, AJ, 123, 1247
——, 2003, AJ, 125, 1649
Finkbeiner, D. et al. 2004, AJ, submitted
Freudling, W., Corbin, M. R., & Korista, K. T. 2003, ApJ, 587, L67
Fukugita, M., Ichikawa, T., Gunn, J.E., Doi, M., Shimasaku, K., & Schneider, D.P. 1996, AJ, 111,
1748
Gunn, J. E., & Peterson, B. A. 1965, ApJ, 142, 1633
Gunn, J.E., et al. 1998, AJ, 116, 3040
Hamann, F., & Ferland, G., 1993, ApJ, 418, 11
Hill, G. J., et al. 1998, Proc. SPIE, 3355, 433
Hogg, D., et al. 2001, AJ, 122, 2129
Kogut, A. et al. 2003, ApJS, 148, 161
Lupton, R.H., Gunn, J.E., & Szalay, A. 1999, AJ, 118, 1406
Lupton, R. H., Gunn, J. E., Ivezic, Z., Knapp, G. R., Kent, S. M., & Yasuda, N., 2001, ASP
Conf. Ser. 238: Astronomical Data Analysis Software and Systems X, 10, 269
Maiolino, R., Oliva, E., Ghinassi, F., Pedani, M. Mannucci, F., Mujica, R., & Juarez, Y., 2004,
A&A, submitted (astro-ph/0312402)
Pentericci, L., et al. 2002, AJ, 123, 2151
Pentericci, L., et al. 2003, A&A, 409, 47
Pier, J. et al. 2003, AJ, 125, 1559
Richards, G. T., et al. 2002, AJ, 123, 2925
Richards, G. T., et al. 2004, AJ, 127, 1305
– 13 –
Schlegel, D.J, Finkbeiner, D.P., & Davis, M. 1998, ApJ, 500, 525
Spergel, D. N. et al. 2003, ApJS, 148, 175
Schmidt, M., Schneider, D. P., & Gunn, J.E., 1995, AJ, 110, 68 (SSG)
Schneider, D. P., Schmidt, M., & Gunn, J.E. 1991, AJ, 102, 837
Skrutskie, M. F., et al., 1997, The Impact of Large-Scale Near-IR Sky Surveys, ed. F. Garz´on, N.
Epchtein, A. Omont, B,. Burton, & P. Persei (Dordrecht: Kluwer), 25
Smith, J., et al. 2002, AJ, 123, 2121
Songaila, A., 2004, AJ, in press (astro-ph/0402347)
Stoughton, C. et al. 2002, AJ, 123, 485
Vignali, C., Brandt, W.N., Schneider, D.P., Garmire, G.P., & Kaspi, S. 2003, ApJ, 125, 2876
Turner, E. L., Ostriker, J. P., & Gott, R. 1984, ApJ, 284, 1
Vanden Berk, D. E., et al. 2000, AJ, 122, 549
White, R. L., Becker, R. H., Fan, X., & Strauss, M. A., 2003, AJ, 126, 1
Wyithe, J. S. B., & Loeb, A., 2002a, Nature, 417, 923
Wyithe, J. S. B., & Loeb, A., ApJ, 577, 57
York, D. G., et al. 2000, AJ, 120, 1579
This preprint was prepared with the AAS LATEX macros v4.0.
– 14 –
Table 1. Photometric Properties of Five New z > 5.7 Quasars
ob ject
redshift
i
z
J
SDSS run
J000239.39+255034.8
J000552.34−000655.8
J141111.29+121737.4
5.80 ± 0.02
5.85 ± 0.02
5.93 ± 0.02
J160254.18+422822.9
J162331.81+311200.5
6.07 ± 0.02
6.22 ± 0.02
21.47 ± 0.11
23.40 ± 0.34
23.43 ± 0.37
22.85 ± 0.30
22.78 ± 0.38
24.52 ± 0.62
18.99 ± 0.05
20.54 ± 0.10
19.63 ± 0.07
19.65 ± 0.08
19.89 ± 0.10
20.09 ± 0.10
> 16.5
19.87 ± 0.10
18.95 ± 0.05
18.46 ± 0.05
19.15 ± 0.10
4152
multiple
3836
3996
3705
3918
The SDSS photometry (i, z ) is reported in terms of asinh magnitudes on the AB system. The
asinh magnitude system is defined by Lupton, Gunn & Szalay (1999); it becomes a linear scale in
flux when the absolute value of the signal-to-noise ratio is less than about 5. In this system, zero
flux corresponds to 24.4 and 22.8, in i, and z , respectively; larger magnitudes refer to negative flux
values. The J magnitude is on a Vega-based system.
Table 2. Continuum Properties of new z > 6 Quasars
ob ject
redshift
AB1280
AB1450
M1450
(Ω-model)
M1450
(Λ-model)
E (B − V )
(Galactic)
J000239.39+255034.8
J000552.34−000655.8
J141111.29+121737.4
J160254.18+422822.9
J162331.81+311200.5
5.80
5.83
5.93
6.07
6.22
19.09
20.30
20.04
19.93
20.20
19.02
20.23
19.97
19.86
20.13
–27.40
–26.21
–26.49
–26.63
–26.40
–27.66
–26.46
–26.75
–26.82
–26.67
0.037
0.033
0.025
0.014
0.022
– 15 –
Fig. 1.— SDSS z -band images of the five new z > 5.7 quasars. Each side of the finding chart is
160′′ . The arrow at the lower left indicates the direction of North on the finding chart; East is 90◦
counterclockwise from North.
– 16 –
Fig. 2.— Spectra of the five new quasars at z > 5.7. The spectrum of SDSS J0002+2550 is a
60-min exposure taken with the MMT-Red Channel; the spectrum of SDSS J0005-0006 is a 40-min
exposure taken with Magellan II and B&C spectrograph; the spectrum of SDSS J1411+1217 is
a 120-min exposure taken with KPNO-4m and MARS; the spectrum of SDSS J1602+4228 is a
120-min exposure taken with the MMT-Red Channel; and the spectrum of SDSS J1623+3112 is a
co-addition of two 120-min exposures using MARS and the Red Channel. The fluxes are scaled to
reproduce the z -band magnitude as measured by the SDSS. All spectra are binned to a dispersion
of 5A per pixel; the spectral resolutions are between 500 and 1000, depending on the spectrograph
used.
– 17 –
Fig. 3.— The composite spectrum of eleven z ∼ 6 quasars (solid line). The spectrum of SDSS
J0005-0002 was not included because of its low S/N. The spectrum of each quasar is redshifted,
scaled according to its m1450 magnitude, and averaged with equal weighting. For comparison, we
also plot the low-redshift quasar spectral composite from Vanden Berk et al. (2001). The effective
redshift in the 1000 – 1500A range in the Vanden Berk et al. composite is about 2. The quasar
intrinsic spectrum redward of Lyα emission shows no detectable evolution up to z ∼ 6, in terms
of both the continuum shape and emission line strengths. On the blue side of Lyα emission, the
strong IGM absorption at z ∼ 6 removes most of the quasar flux.
– 18 –
Fig. 4.— The evolution of the quasar comoving spatial density at M1450 < −26.7 in the Λ-model.
The filled circle represents the result from this survey. The error-bar in redshift indicates the
redshift range covered by the i-dropout survey. The dashed and dotted lines are the best-fit models
from Fan et al. (2001b) and Schmidt et al. (1995, SSG), respectively. The solid line is the best-fit
model from the 2dF survey (Croom et al. 2004) at z < 2.3.
|
0707.0468 | 1 | 0707 | 2007-07-03T18:10:34 | A Spitzer Study of the Mass Loss Histories of Three Bipolar Pre-Planetary Nebulae | [
"astro-ph"
] | We present the results of far-infrared imaging of extended regions around three bipolar pre-planetary nebulae, AFGL 2688, OH 231.8+4.2, and IRAS 16342$-$3814, at 70 and 160 $\mu$m with the MIPS instrument on the Spitzer Space Telescope. After a careful subtraction of the point spread function of the central star from these images, we place constraints on the existence of extended shells and thus on the mass outflow rates as a function of radial distance from these stars. We find no apparent extended emission in AFGL 2688 and OH 231.8+4.2 beyond 100 arcseconds from the central source. In the case of AFGL 2688, this result is inconsistent with a previous report of two extended dust shells made on the basis of ISO observations. We derive an upper limit of $2.1\times10^{-7}$ M$_\odot$ yr$^{-1}$ and $1.0\times10^{-7}$ M$_\odot$ yr$^{-1}$ for the dust mass loss rate of AFGL 2688 and OH 231.8, respectively, at 200 arcseconds from each source. In contrast to these two sources, IRAS 16342$-$3814 does show extended emission at both wavelengths, which can be interpreted as a very large dust shell with a radius of $\sim$ 400 arcseconds and a thickness of $\sim$ 100 arcseconds, corresponding to 4 pc and 1 pc, respectively, at a distance of 2 kpc. However, this enhanced emission may also be galactic cirrus; better azimuthal coverage is necessary for confirmation of a shell. If the extended emission is a shell, it can be modeled as enhanced mass outflow at a dust mass outflow rate of $1.5\times10^{-6}$ M$_\odot$ yr$^{-1}$ superimposed on a steady outflow with a dust mass outflow rate of $1.5\times10^{-7}$ M$_\odot$ yr$^{-1}$. It is likely that this shell has swept up a substantial mass of interstellar gas during its expansion, so these estimates are upper limits to the stellar mass loss rate. | astro-ph | astro-ph |
A Spitzer Study of the Mass Loss Histories of Three Bipolar
Pre-Planetary Nebulae
Tuan Do, Mark Morris
Physics and Astronomy Department, University of California, Los Angeles, CA 90095-1547
Raghvendra Sahai and Karl Stapelfeldt
Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109
ABSTRACT
We present the results of far-infrared imaging of extended regions around
three bipolar pre-planetary nebulae, AFGL 2688, OH 231.8+4.2, and IRAS
16342−3814, at 70 and 160 µm with the MIPS instrument on the Spitzer Space
Telescope. After a careful subtraction of the point spread function of the central
star from these images, we place constraints on the existence of extended shells
and thus on the mass outflow rates as a function of radial distance from these
stars. We find no apparent extended emission in AFGL 2688 and OH 231.8+4.2
beyond 100 arcseconds from the central source. In the case of AFGL 2688, this
result is inconsistent with a previous report of two extended dust shells made on
the basis of ISO observations. We derive an upper limit of 2.1 × 10−7 M⊙ yr−1
and 1.0 × 10−7 M⊙ yr−1 for the dust mass loss rate of AFGL 2688 and OH 231.8,
respectively, at 200 arcseconds from each source. In contrast to these two sources,
IRAS 16342−3814 does show extended emission at both wavelengths, which can
be interpreted as a very large dust shell with a radius of ∼ 400 arcseconds and
a thickness of ∼ 100 arcseconds, corresponding to 4 pc and 1 pc, respectively,
at a distance of 2 kpc. However, this enhanced emission may also be galactic
cirrus; better azimuthal coverage is necessary for confirmation of a shell. If the
extended emission is a shell, it can be modeled, with some assumptions about
its dust properties, as enhanced mass outflow at a dust mass outflow rate of
1.5 × 10−6 M⊙ yr−1 superimposed on a steady outflow with a dust mass outflow
rate of 1.5 × 10−7 M⊙ yr−1. Because of the size of the possible shell, it is likely
that this shell has swept up a substantial mass of interstellar gas during its ex-
pansion, so these estimates are upper limits to the stellar mass loss rate. We find
a constant color temperature of 32 K throughout the circumstellar envelope of
IRAS 16342−3814, which is consistent with heating by the interstellar radiation
field.
– 2 –
Subject headings: (ISM:) planetary nebulae: individual (AFGL 2688, OH 231.8+4.2,
IRAS 16342−3814) - stars: AGB and post-AGB - stars: mass loss
1.
Introduction
Pre-planetary nebulae (PPNe) represent a fleeting stage of stellar evolution. These
transitional objects arise between the rapid mass loss phase at the end of the asymptotic
giant branch (AGB) and the ionized planetary nebula stage. AGB stars are important
because they are the Galaxy's main mechanism for the replenishment of dust and gas into
the interstellar medium (ISM), ejecting most of their main sequence mass within a few times
105 years because of their high mass loss rates. By looking at the circumstellar material
around PPNe, we can see an imprint of the activity of the earlier AGB. PPNe are also in
a stage where the geometry of the mass loss from the central star usually changes from
spherically symmetric to an axially symmetric bipolar outflow, as illustrated in the HST
images of AFGL 2688 (Sahai et al. 1998). In these images, a spectacular bipolar structure is
superimposed on a set of concentric shells of gas and dust ejected near the end of the AGB.
In the study reported here, we have attempted to characterize the mass loss behavior of the
progenitor AGB stages of three systems that are now bipolar.
Using the direct detection of infrared emission from dust lost over as much as 105 years
during the AGB, it is possible to study the mass loss histories of these objects. There has
been a dearth of observations of large extended shells of evolved stars out to the distances
sampled in this project primarily because of the difficulty in observing in the far infrared
where the cool dust has its peak emission. With the Spitzer Space Telescope (Werner et al.
2004), the far infrared is now accessible at a resolution and sensitivity sufficient to potentially
resolve the structures of these dust shells, and thereby to test theories of mass loss in AGB
stars.
Several tantalizing cases were observed by IRAS, showing that large dust shells exist
around some evolved stars. For example, Gillett et al. (1986) found that R CrB shows a
very large shell with a radius of 4 parsecs, and Hawkins (1990) reported a dust shell with
a diameter of 30 - 40 arcminutes (∼ 1 pc) around W Hya. However, these observations
could not resolve the structure that may be present in the shells, or address the possibility
of multiple shells. If the mass loss during the AGB is constant, we should expect to see a
smooth envelope with column density declining as 1/b, where b is the displacement from the
star in the sky. However, if the mass loss rate fluctuates as the star goes through thermal
pulses during the AGB, as the models suggest (e.g., Vassiliadis & Wood 1993), then we
should see enhanced emission in the form of multiple dust shells around our objects. For
– 3 –
example, Speck et al. (2000) reported on the basis of ISO data that there are large concentric
shells in AFGL 2688 and AFGL 618. This result is among a small set of observations
showing evidence for periodic mass loss caused by thermal pulsation as well as the time scale
between thermal pulses predicted by AGB evolutionary models (e.g., Vassiliadis & Wood
1993). Confirmation of these shells at higher resolution and greater sensitivities motivated
our selection of AFGL 2688 as a target for this study. There have been no previous claims
of large-scale dust emission from OH 231.8+4.2 (hereafter OH 231.8) or IRAS 16342−3814.
However, they are presently very young PPNe with high mass loss rates of approximately
10−4M⊙ yr−1 (Alcolea et al. 2001; Sahai et al. 1999), which suggests that the emission from
dust produced during the AGB may be readily visible.
Here we report on the lack of very extended emission, spherically symmetric or otherwise,
from AFGL 2688 and OH 231.8. However, we do see possible evidence for a very large diffuse
dust shell around IRAS 16342−3814 with a radius of 400 arcseconds, though this extended
emission might also be from galactic cirrus.
2. Observations and Data Reduction
AFGL 2688 (the Egg Nebula), OH 231.8, and IRAS 16342−3814 were observed with the
MIPS instrument (Rieke et al. 2004) at 70 and 160 µm on the Spitzer Space Telescope. These
PPNe were observed along two mutually perpendicular scan paths in order to sample dust
emission out to about 800 arcseconds from the central source, and to determine background
levels. At 70 µm, the scan paths are 15′ ×3.0′ in one direction and 11′ ×7.7′ in the other, while
at 160 µm, the scan paths are 15′ × 2.6′ and 10.5′ × 6.5′. The pixel scale is 9.2 arcseconds
at 70 µm and 16 arcseconds at 160 µm. However, the central sources were not directly
observed in order to avoid persistence artifacts from these bright sources. This observing
strategy limits the azimuthal coverage of the sources, but should give an adequate estimate
of the presense of well-defined shells or asymmetries such as large-scale bipolarity.
The basic science calibrated data (BCD) was reduced using the software package MOPEX1
from the Spitzer Science Center (SSC). MOPEX was used for rejecting outliers as well as
for co-adding and mosaicking the individual BCD frames to create the final image. Further
reduction was done manually to remove striping due to bright latents as described in the
MIPS data handbook2. The correction involved finding the median value measured for every
1http://ssc.spitzer.caltech.edu/postbcd/
2http://ssc.spitzer.caltech.edu/mips/dh/
– 4 –
pixel from a series of BCD frames far from the central source, where we assume there to be
no extended emission, then subtracting this median from each data frame before co-adding
and mosaicking. The median subtraction removes both the bright latents and the uniform
component of background emission. Figure 1 shows the final mosaicked images.
The most prominent features of the 70 µm images of the bipolar targets are the diffrac-
tion spikes from the point spread function (PSF) of the central source. Although we did
not image the central star, the wings of the PSF are still present and bright out to about
200 arcseconds from the source. At 160 µm, the PSF is less pronounced but we see much
more background than at 70 µm from galactic cirrus. The galactic cirrus emission presents
a problem in determining the true sky background since it covers much of our fields. For
both the 70 and 160 µm images, we use the median surface brightness of regions of low
and uniform brightness far from the central source as an estimate of the sky background.
This procedure removes a uniform background from the images, retaining possible enhanced
emission from the source and galactic cirrus. The uncertainty in the background (see below)
at 160 µm is higher than at 70 µm because it is more difficult to find a large patch of uniform
surface brightness to estimate the sky background.
2.1. Sensitivities
The sensitivity of our data is estimated by examining the standard deviation of a patch
of uniform background far from the source both before and after the median background
subtraction method described above. For AFGL 2688, before background subtraction, the
mean of a patch in the eastern scan path at 70 µm is 14.1 MJy Sr−1 with a standard deviation
of 1.11 MJy Sr−1. Some of the variance in the background is likely from the effects of bright
latents from the detector pixels, which are removed by the median background subtraction
described above. After median subtraction, the mean background is at 0.6 MJy Sr−1 with
a standard deviation of 0.92 MJy Sr−1. The mean surface brightness at 160 µm before
background subtraction is 31.45 MJy Sr−1 with a standard deviation of 1.67 MJy Sr−1.
After background subtraction, the mean is at 0.33 MJy Sr−1 with a standard deviation of
1.51 MJy Sr−1. The sensitivity for OH 231.8 is better than for AFGL 2688, probably because
of the lower sky background levels. The sensitivity for IRAS 16342−3814 is comparable to
that of AFGL 2688 at 70 µm, but worse at 160 µm due to a higher sky background and
greater presence of galactic cirrus. The mean background and sensitivities for all three
sources at 70 and 160 µm are given in Tables 1 and 2, respectively.
– 5 –
2.2. PSF Subtraction
Since the PSF pattern from the central source is so prominent at 70 µm, we attempted
to fit a model PSF to our images in order to subtract the bright central source in search
of extended dust emission near the source. The stinytim software provided by the SSC was
used to generate model PSFs. We chose to use the default MIPS throughput curve along
with a blackbody spectrum at several temperatures ranging from 20 to 100 K. See Figure
2 for examples of the model PSFs. Multidimensional fitting was carried out for our images
in order to minimize the residuals of the PSF subtraction using various x-y shifts as well as
the scaling of the PSF flux values. The best fit was found by minimizing the square root of
the sum of the squares of the differences in flux. We found several limitations in the model
PSFs when attempting to subtract the PSF from the images. Even for the best-fit PSFs, the
subtraction still leaves some obvious PSF residuals at a level of 10−4 relative to the peak as
extrapolated from the model PSF. These residuals show that the wings of the model PSF are
not well characterized at that flux level. The PSF subtraction also tends to leave portions
of the region within the Airy ring with negative flux values. The most likely reason is a
non-linear pixel response as the pixels approach saturation so that although the model has a
higher brightness value closer to the core of the PSF, the actual pixel values are leveling off
for our bright sources. Unfortunately, there are presently no empirical data on the MIPS 70
or 160 µm PSF to compare to the model PSF at distances greater than 100 arseconds from
the peak. Beyond about 240 arseconds, PSF features are no longer present in the MIPS
images so our ability to detect extended emission there is only limited by background and
integration time.
The results of PSF subtraction for AFGL 2688 are shown in Figure 3. Note the remaining
features near the north and at a position angle of ∼ 50◦ east of north, about 150 arcseconds
from location of the source in the PSF subtraction in Figure 3a. This residual can also been
seen in the roughly the same region of the 70 µm PSF subtraction of OH 231.8 in Figure 4.
It seems unlikely that these features are physical, but the northern feature in AFGL 2688
does align with its bipolar outflows seen in the optical and infrared, but which are observed
within the inner 10 arcseconds of the PPN. The blob in AFGL 2688 at a position angle
of ∼ 50◦ also roughly aligns with the outflow direction seen in previous CO observations
(Skinner et al. 1997), also at around 10 arcseconds from the central star. The fact that the
features we see are at nearly 200 arcseconds from the source and lie almost on top of the
diffraction rays of the original PSF casts doubt upon their reality. This dilemma cannot
be resolved until a well sampled empirical PSF out to several arcminutes is available. We
may be seeing possible features in the wings of the actual PSF that the model has failed to
reproduce.
– 6 –
As a substitute for an empirical PSF, we used OH 231.8 to subtract the PSF from AFGL
2688. Figure 3b shows the result. The residuals from using OH 231.8 as an empirical PSF
appear to be much smaller than those obtained using the model PSF, reducing the residual
of 30 MJy Sr−1 to less than 10 MJy Sr−1 at 100 arcseconds. Note that there are almost
no PSF features, such as diffraction spikes, remaining in Figure 3b compared to Figure 3a.
However, it may be problematic to use OH 231.8 +4.2 as an empirical PSF because it may
have extended emission as well, although it would be remarkably fortuitous if both sources
had extended emission having precisely the same morphology. It is also problematic to scale
OH 231.8 to the same surface brightness as AFGL 2688 because in order to subtract the PSF,
the whole image needs to be scaled, thus scaling the background as well. Because AFGL
2688 is over twice as bright as OH 231.8, but the background in AFGL 2688 is not, scaling to
the level of the AFGL 2688 PSF would overemphasize the background in OH 231.8 so that
the residuals after the PSF subtraction might be dominated by the background in OH 231.8.
We can nevertheless see that the PSF subtraction using OH 231.8 results in residuals at a
level ∼ 10 times below that from using the model PSF at 150 arcseconds, which suggests
that OH 231.8 is either a point source at 70 µm or that its extended structure has the same
orientation, scale, and shape as in AFGL 2688.
The two-dimensional PSF subtraction at 160 µm was done in a similar way using the
model stinytim PSF. There are PSF features remaining after this subtraction as well (see
Figure 3); the ends of the diffraction spikes at 200 arcseconds from the source in the original
image are not completely removed. We did not attempt a PSF subtraction of the 160 µm
image of IRAS 16342−3814, because that the PSF of the central source is weak enough at
this wavelength that its features are below the background and the extended structures.
2.3.
1-D PSF Subtraction
In order to investigate the hypothesis of spherical shells around a central source, we
have determined an azimuthal average of the intensity around the central source. Such an
average would show enhancements for spherically concentric shells projected as circularly
symmetric features from episodic mass loss during the AGB phase. An azimuthal average
also has the advantage that it greatly reduces the parameter space necessary to fit the model
PSF. The source position and intensity are found by fitting to the brightest Airy ring visible
in our images, which is the second brightest Airy ring of the PSF (see Figure 2), since the
source itself does not appear in our images. The PSF Airy ring used for scaling the PSF
is located 76 arcseconds from the source at 70 µm and 170 arcseconds from the source at
160 µm. We also find that the azimuthal averages provide a subtraction with less residuals
– 7 –
than using the two dimensional subtraction because localized irregularities of the PSF are
averaged away.
The one dimensional PSF subtractions for AFGL 2688 and IRAS 16342−3814 at 70
µm were done using both the model PSF and OH 231.8, and are shown in Figure 5. The
profiles of AFGL 2688 and OH 231.8 follow each other almost exactly out to beyond 200
arcseconds from the source. This further indicates that OH 231.8 and AFGL 2688 are both
strongly centrally concentrated at 70 µm, unless they both have exactly the same azimuthal
excesses. They both also match the model PSF very well out to 100 arcseconds beyond
which both appear to have slight excess (< 1% of the extrapolated peak of the PSF), above
the model PSF. Once again, this excess in the wings is more likely due to the model PSF
not reproducing the wings of the actual PSFs rather than resulting from physical emission
associated with the sources. At 160 µm, we only use the model PSF for subtraction because
of possible contamination due to galactic cirrus in OH 231.8.
3. Results and Analysis
3.1. AFGL 2688
3.1.1. 70 µm
At 70 µm, the PSF-subtracted images of both AFGL 2688 and OH 231.8 are similar. The
PSF residuals extend to about 150 arcseconds from the source, which may be obscuring some
physical extended features. However, there are no obvious circularly symmetric residuals
in the PSF subtraction even with the confusion near the central source, which would be
expected if there had been spherically symmetric mass loss above a dust mass loss rate of
∼ 2.1 × 10−7M⊙ yr−1 (see discussion section). Previous studies with ISO by Speck et al.
(2000) led to the claim of large shells around AFGL 2688 having radii of 150 and 300
arcseconds on the basis of a one-dimensional scan of the Egg Nebula with ISOPHOT at
120 and 180 µm. We find, with better spatial resolution and azimuthal coverage using
Spizer, that at 70 µm, there are no signs of shell-like extended emission in our field. At 150
arcseconds from the source, the PSF subtraction residuals have a surface brightness of about
16 MJy Sr−1 using the model PSF and about 1 MJy Sr−1 using OH 231.8 as the empirical
point source(Figure 5). Note also that there is no discernible excess emission above the
fluctuations in the background (σ ∼ 0.92 MJy Sr−1) at 300 arcseconds from the source.
– 8 –
3.1.2. 160 µm
The 160 µm image (see Figure 3) shows roughly uniform enhanced emission to the edge
of our field of view at 1000 arcseconds from the source in the eastern scan path; the northern
scan show enhanced emission out to about 300 arcseconds, then dropping off to background
levels. The enhanced emission in the two paths is asymmetric as it does not fall off with
distance in the eastern scan path. There does not appear to be any structure to the emission
in either scan, other than that which may be attributable to the PSF diffraction spike in
the north. Because the residual extended emission shows no symmetry about the central
star, and because it shows no intensity falloff in the eastern scan path, this emission most
likely arises from irregularly distributed galactic cirrus in this direction; this is supported by
the relatively lower quality IRAS 100 µm image, which shows substantial extended cirrus
beyound our survey region. The ISOPHOT observation of a possible shell at 300 arcseconds
was based on a 53′ × 3′ linear scan with 30′′ × 92′′ pixels, at a position angle of 8 degrees east
of north centered on the source, which in our case would be sampling the emission in the
northern scan. We see with better azimuthal coverage with MIPS that the emission in both of
our scans is likely strongly contaminated with galactic cirrus. The presence of the inner shell
reported by Speck et al. at 150 arcseconds cannot be directly confirmed by our observations
because the 160 µm PSF Airy ring at 170 arcseconds from the source overlaps that region.
The surface brightness at the Airy ring is about 17 MJy Sr−1 after background subtraction
(but before the PSF subtraction). This is comparable to the background subtracted surface
brightness of 20 (30) MJy Sr−1MJy Sr−1 at 120 (180) µm reported by Speck et al. (2000).
In the azimuthally averaged surface brightness of our 160 µm image (Figure 6), we see
a similar result as the 70 µm data. Speck et al. reported a brightness value of the putative
shell at 300 arcseconds from the source to be 20 MJy Sr−1 at 120 µm and 60 MJy Sr−1
at 180 µm, without background subtraction. The 120 µm emission is about 10 MJy Sr−1
above their background as extrapolated from the surface brightness far from the star in their
plot. At 180 µm, the excess emission is about 15 MJy Sr−1 above the background. In the
MIPS 160 µm image, we find that the surface brightness at 300 arcseconds from the source
is about 2 MJy Sr−1 after background subtraction. Although the comparison is not at the
same wavelength, the excess emission we detect is 7 times below that measured by ISO.
When comparing the two scans in our observations separately, we see that the eastern scan
has roughly uniform brightness of 4 MJy Sr−1 out to 1000 arcseconds, while the emission in
the northern scan drops off at 400 arcseconds from the source from about 2 MJy Sr−1 to the
level we use as the background in the data reduction. Both the asymmetry in the enhanced
emission in both scans and the low surface brightness in the Spitzer data at 300 arcseconds
from the source lead us to conclude that the shell at this distance reported on the basis of
ISO data is probably not associated with the source.
– 9 –
3.2. OH 231.8+4.2
Neither the 70 nor 160 µm images of OH 231.8 show any signs of spherically symmetric
extended emission. The PSF subtraction at 70 µm leaves similar residuals as for AFGL
2688, though the brightnesses of these residuals are much lower because the central source is
intrinsically less bright. The strongest PSF residuals are along the diffraction spikes, similar
in shape to those from AFGL 2688. This further suggests that these residuals are not physical
features, but are rather PSF artifacts not accounted for by the stinytim model. In the 160
µm azimuthally averaged plots, both scans appear to have rather uniform emission, with the
northern scan having slightly higher surface brightness at approximately 2 MJy Sr−1, while
the eastern scan has an average surface brightness of 1 MJy Sr−1 (Figure 7).
The 160 µm image shows some slightly clumpy emission along the northern scan path,
slightly beyond the PSF diffraction spike (see Figure 1) at 320 arcseconds. The clumps, with
a width of about 80 arcseconds, have a surface brightness of about 3.6 MJy Sr−1 on a 2
MJy Sr−1 diffuse background. The IRAS 100 µm image shows diffuse galactic cirrus in this
region, which may be the origin of these clumps. They are aligned with the bipolar axis, but
unfortunately also with the diffraction ray so it may also be a PSF feature. Since this study
is concerned with investigating spherically symmetric emission, we will defer the question
of their nature. The azimuthally averaged surface brightness shows no enhanced emission
attributable to the source (Figure 8).
3.3.
IRAS 16342−3814
In contrast to the previous two sources, IRAS 16342−3814 does show some evidence for
what may be a large patchy shell at both 70 and 160 µm. The enhanced emission out to
a radial distance of ∼ 400 arcseconds is consistent with rough circular symmetry about the
star in the coverage area available (Figure 1). This radius corresponds to a physical size of
about 4 parsecs, assuming the distance to IRAS 16342−3814 to be 2 kpc (Sahai et al. 1999).
This possible shell structure is rather patchy, with a large arc in the western scan at 70 µm.
At 160 µm, the same arc is present, but is more diffuse. Nevertheless, the azimuthal average
shows an excess above the background at ∼ 400 arcseconds from the central source. The
brightest portion of this feature occurs at a position angle of 250 degrees east of north, which
is the same direction as the axis of the bipolar nebula seen in the optical. The ratio of the
70 to 160 µm average surface brightness, is roughly constant (∼ 0.3) as a function of radial
distance from the source. This ratio yields a color temperature of ∼ 32 ± 2 K, and, if a ν 1
emissivity law is assumed, a dust temperature of 26 ± 2 K. The constant color temperature
suggests that the dust is heated primarily by the interstellar radiation field.
– 10 –
Because there are a number of interstellar cirrus features in the general direction of
IRAS16342-3814, we cannot definitively argue that the patchy arc of emission in the scan
paths around this source is caused by mass loss from the star. The IRAS 100 µm image of
this region with an overlay of the coverage of our data (figure 9) shows that there is indeed
a substantial likelihood that this emission is cirrus. With this caveat in mind, we proceed in
the analysis below on the assumption that the extended emission has resulted from mass loss
from the star to obtain some indicative numbers for the mass loss rate necessary to produce
such a shell. In the following analysis, we will use the term "mass outflow rate" to indicate
the total mass flowing outward per unit time across a spherical shell at a particular radius;
this includes both the stellar component and the interstellar medium being swept up; we
will use the term mass loss rate to refer to mass loss from only the star.
Making a few assumptions about the mass outflow and the dust, we can set some limits
on the mass outflow rate of IRAS 16342−3814. Assuming that the mass loss is spherically
symmetric, we can fit a model of the expected brightness profile to the observed radial profile
of the source. Figures 10 and 11 show the azimuthally averaged radial profiles at 70 and 160
µm, respectively. We fit only the 70 µm radial profile because there is less contamination by
galactic cirrus compared to 160 µm. Following the derivation by Gillett et al. (1986), if we
assume a 1/r2 density profile, appropriate for constant mass outflow, where r is the distance
from the star, and a constant expansion velocity, ve, then the surface brightness Iν(b) has
the form:
.
M κνBν(T )
Iν(b) =
cos−1(
b
Rmax
)
for Rmin ≤ b ≤ Rmax
Iν(b) =
.
M κνBν(T )
2πveb
2πveb
(cid:20)cos−1(
b
Rmax
) − cos−1(
b
Rmin
)(cid:21)
for b < Rmin
(1)
where b is the projected physical distance from the source in the sky, κν is the opacity
coefficient in units of cm2 g−1, Bν(T ) is the blackbody function, and Rmax and Rmin are the
outer and inner radii of the shell. We further assume that the outflow velocity is constant
at 15 km s−1, and that the dust is composed of small astronomical silicate grains with κν
= 104 cm2 g−1 at 70 µm (using a = 0.1 µm, ρ = 2.3 g cm−3, and Qν
abs = 2.99 × 10−3 from
Draine & Lee (1984)). We also adopt a constant temperature of 26 K throughout the shell
and envelope. Figure 12 gives a comparison between different models having various mass
outflow rates. The best-fit model is a shell with a 4.2 pc radius and a thickness of 1 pc, with
a dust outflow rate of 1.5×10−6M⊙ yr−1 superimposed on a smooth envelope with a constant
dust outflow rate of 1.5 × 10−7M⊙ yr−1. This implies a rather high gas outflow rate on the
order of 3 ×10−4M⊙ yr−1 in the shell and 3 ×10−5M⊙ yr−1 in the smooth envelope, assuming
– 11 –
a gas to dust mass ratio of 200. Using OH 231.8 as the empirical PSF, the fit only requires a
shell component with a mass outflow rate of 3 × 10−4M⊙ yr−1 without an underlying smooth
envelope. These fits depend on the choice of PSF, but once the PSF is chosen, the mass
outflow rates are insensitive to residuals from PSF subtraction, since we only fit for the
region beyond 150 arcseconds from the central source (see Figure 10). However, our value
for the mass outflow rate is highly dependent on the assumed temperature of the dust. For
typical ISM dust temperatures between 22 and 35 K, the inferred dust outflow rate is related
.
M ∝ T −7.6 at 70 µm. If T > 26 K, we would
empirically to the adopted temperature by:
infer a substantially lower mass outflow rate.
Given the model thickness of the shell and assuming a typical expansion velocity of 15
km s−1, we estimate the maximum duration of the enhanced mass loss event which produced
the shell to be about 65,000 years. This is an upper limit because the internal velocity
dispersion in the shell broadens the shell as it expands. A 1 km s−1 dispersion, for example,
would reduce the duration of the mass loss that produced the shell by about a third. If we
use an age of 40,000 years for the duration of the enhanced mass loss and approximate the
shell as spherical, the total dust mass would be about 0.04 M⊙, giving a total shell mass of ∼
8 M⊙, assuming a gas-to-dust mass ratio equal to 200. The velocity of the shell is probably
slower than the molecular outflow velocity of a typical AGB star and the total amount of
mass lost by the AGB star is likely smaller than this amount because of the interaction with
the ISM out to the distances we are observing (see below).
The mass of the shell can be estimated directly from the infrared emission using the
equation M = FνD2/(Bν(T )κν), where Fν is the total flux density in the shell and D is the
distance to the source. The integrated flux density from the limited azimuthal coverage that
we have between radii of 300 and 400 arcseconds is 6.4 Jy at 70 µm, corresponding to a dust
mass of 0.015 M⊙ in our observed portion of the shell, assuming a temperature of 26 K for
the dust. We have about one fifth of the full azimuthal coverage with data at this distance
from the star, and can estimate the total dust mass of the shell if we approximate the shell
as isotropic with an averaged surface brightness of 2.5 MJy Sr−1 at 70 µm from the radial
profile, with a width of 100 arcseconds, and a temperature of 26 K. The total dust mass of
the extrapolated shell is then about 0.03 M⊙, comparable to the above estimate.
Because the shell is so large, the amount of interstellar matter that may have been swept
up and accumulated in the shell during the AGB could be significant. For an ISM hydrogen
density of 1 cm−3, a 4 parsec radius shell would have swept up about 7 M⊙, which would
account for almost all the mass that we may be measuring. The accumulation of interstellar
material could potentially lower the stellar mass loss rate estimated from our model by a
large factor. We therefore emphasize that the mass outflow rate estimate is only an upper
– 12 –
limit to the stellar mass loss rate.
One of the difficulties in addressing the mass loss history of IRAS 16342−3814 with the
current data is its location in the Galaxy; with a scale height of only 150 pc, there is possible
confusion with diffuse galactic cirrus. An unfortunate alignment with background emission
is an alternative to the existence of a shell produced by mass loss. Further observations with
better azimuthal coverage are needed to test the circumstellar shell hypothesis.
4. Discussion
The extended dust emission in AFGL 2688 and OH 231.8 seems to show that these two
objects do not have as long a mass loss history as one might have anticipated. Our results
for these two sources probe the region beyond ∼ 100 arcseconds from the central star, which
corresponds to 1.0 × 104 yrs ago for AFGL 2688 and 2.8 × 104 yrs ago for OH 231.8, given
a distance of 420 pc (Ueta et al. 2006) and 1.2 kpc (Jura & Morris 1985), respectively, and
an expansion velocity of 20 km s−1. We can can set a limit to the dust mass loss rate by
inverting equation 1:
M =
Iν(b, T )2πveb
κνBν(T ) cos−1(b/Rmax)
(2)
The dust mass loss rate can be simplified as a function of the observed surface brightness I
and the projected distance from the source in the sky b with the assumption that b ≪ Rmax:
M (70µm) ∼ 3.8 × 10−8M⊙ yr−1(cid:18) I70 µm
M (160µm) ∼ 2.9 × 10−8M⊙ yr−1(cid:18) I160 µm
1 MJy Sr−1(cid:19)(cid:18)100 cm2g−1
1 MJy Sr−1(cid:19)(cid:18)20 cm2g−1
κ70 µm (cid:19)(cid:18) b
κ160 µm (cid:19)(cid:18) b
1 pc(cid:19)(cid:16)
1 pc(cid:19)(cid:16)
15 km s−1(cid:17)(cid:18) T
15 km s−1(cid:17)(cid:18) T
30 K(cid:19)−7.6
30 K(cid:19)−3.6
ve
ve
If b ∼ Rmax, then the full form of equation 2 must be used because the cos−1(b/Rmax)
term in the denominator of equation 2 becomes important and will cause the dust mass
loss rate inferred from a given surface brightness value to increase drastically. For the cases
such as AFGL 2688 and OH 231.8 where we do not see a clear envelope associated with the
source, we cannot be sure that the condition b ≪ Rmax holds since Rmax is indeterminate.
The equations above also show the temperature dependence in the form of a power law to
approximate the blackbody function between 22 and 35 K. From these equations we see that
the temperature dependence at 160 µm is less steep than at 70 µm, but the contamination
by galactic cirrus near the sources makes estimating upper limits problematic at 160 µm.
We can establish an upper limit to the mass loss rates for AFGL 2688 and OH 231.8 at 70
µm based up on the surface brightness of the residual left from the PSF subtraction at 200
– 13 –
arcseconds from the source, which is a compromise between a location far enough from the
source that PSF subtraction errors are small and being close enough to the source that there
could plausibly be a circumstellar envelope. For AFGL 2688, 200 arcseconds corresponds to
a radial distance from the source of 0.4 parsecs, using a distance to the source of 420 parsecs
(Ueta et al. 2006). The surface brightness after PSF subtraction with the model PSF is
about 14 MJy Sr−1 at 70 µm. Assuming an expansion velocity of 20 km s−1 as observed in
CO by Skinner et al. (1997), a dust temperature of 30 K, κν = 104 cm2 g−1 appropriate for
carbon dust (Draine & Lee 1984), and that the possible envelope has a radius significantly
greater than 0.4 pc, then the dust mass loss rate upper limit is 2.1 × 10−7 M⊙ yr−1. Similarly
for OH 231.8, the residual emission from the PSF subtraction is 3 MJy Sr−1 at 200 arcseconds
from the central star, which corresponds to 1.3 parsecs at a distance of 1.3 kpc. This surface
brightness implies a dust mass loss rate of 1.0 × 10−7 M⊙ yr−1 assuming an outflow velocity
of 15 km s−1, κν = 98 cm2 g−1 (appropriate for silicate dust), and a temperature of 30 K.
There are several potential explanations for the lack of extended emission seen in this
study. One is that, even with Spitzer's increased sensitivity, the dust emission is below the
threshold for detection. Since the distances we are studying are far from the central star,
we would expect the temperature of the dust to be determined by the ambient interstellar
radiation field. If the dust temperature is about 20 K, typical for the ISM (Mathis et al.
1983), and an emissivity ∝ ν 1, the surface brightness would be 12 times stronger at 160
µm, than at 70 µm where the resolution and sensitivity of MIPS is better. We would be
more likely to detect cool dust emission at 160 µm than at 70 µm. However, confusion with
the galactic cirrus is also likely at 160 µm because it is about the same temperature as any
hypothetical extended dust emission associated with the source that is heated primarily by
the interstellar radiation field. For comparison, our detection of the possible shell from IRAS
16342-3184 has a ratio of I160µm/I70µm ∼ 3.3, corresponding to a color temperature of 32 K,
which requires a higher than average interstellar radiation field.
The upper limits to the mass loss rates derived in this study can be compared with
theoretical AGB evolutionary models, particularly those that predict enhanced mass loss
rates due to thermal pulsation near the end of the AGB. The spatial coverage in this study
between ∼ 200 to 1000 arcseconds from the central star translates into a probe of the history
of mass loss between 2 × 104 yrs to 1 × 105 years ago for AFGL 2688 and 6 × 104 yrs to
3 × 105 yrs ago for OH 231.8, using an expansion velocity of 20 km s−1 for both sources. The
models by Vassiliadis & Wood (1993) show that for a 2.0 M⊙ progenitor, during the last few
×105 yrs of AGB evolution, there are several thermal pulses which result in enhanced mass
loss rates peaking at about 1.3 × 10−5 M⊙ yr−1 during the end of each pulse. Using a gas to
dust mass ratio of 200, we find that for AFGL 2688 and OH 231.8, the upper limit to the
total mass loss rate during the above time intervals is about 4 × 10−5 M⊙ yr−1 and 2 × 10−5
– 14 –
M⊙ yr−1, respectively. These limits are close to the sensitivity necessary to see shells that
may be the result of thermal pulsation on the AGB. Although we do not detect the shells
reported by Speck et al. (2000), which they attribute to thermal pulses, the signatures of
thermal pulses may still be present, but below our current sensitivity.
We also consider the possibility that OH 231.8 does not show extended emission in
the far infrared because of interactions with a binary companion, which caused the central
star to lose mass more rapidly and more recently than during the evolution of a lone AGB
star. A companion to QX Pup (the central star of OH 231.8), is evidenced by optical
spectra consistent with a companion of stellar type A0 V (S´anchez Contreras et al. 2004).
QX Pup also shows a peculiar paradox of being an M9III (Cohen 1981) AGB star, while its
bipolar activity and morphology display all the signs of post-AGB activity of typical PPNe.
Having a close companion would enhance the mass loss rate and provide a mechanism for
generating the bipolar outflows (Morris 1987), thus shortening its mass loss history enough
that we should not be surprised to see no emission far from the source.
In contrast to
the collimated outflow from OH 231.8 (Alcolea et al. 2001), AFGL 2688 has a spherically
symmetric envelope seen in 13CO (Yamamura et al. 1996) and evidenced by the partial,
concentric, circular arcs present in HST images (Sahai et al. 1998). Since a binary interaction
would not cause the past spherically symmetric mass loss, a possible companion is probably
not a good explanation for initiating mass loss, although the present bipolar morphology of
AFGL 2688 is consistent the possibility that a binary interaction has altered the mass loss
in more recent times.
Tracing the mass loss history during the AGB phase should be relatively straightforward
via mapping the emission from the circumtellar envelope. But in practice, it has been
difficult because molecular-line observations are ultimately limited by the photodissociation
of molecules in the outer envelope regions, and far-infrared observations of dust emission have
been limited by the generally low angular resolution of the space-based telescopes which have
been available for this purpose in the past (IRAS and ISO). Hence the reported detections of
very extended emission in a few dying stars with IRAS and ISO has generally been recognized
as an important milestone in the study of mass loss. However, such detections have also raised
the very important question of why only a few select objects like Y CVn (Izumiura et al.
1996) or RY Dra (Young et al. 1993) – not particularly known for having high mass loss rates
– show extended envelopes, whereas large numbers of stars having high CO-determined mass
loss rates do not reveal the presence of such envelopes. Is this because the radial density
law for most of these high mass-loss stars is significantly steeper than in objects like Y CVn
or RY Dra, and in particular steeper than r−2 (an issue of profound importance for the
evolutionary times of stars through the AGB phase and theories of mass-loss), or are the
claimed detections of extended envelopes really a result of poorly characterised instrumental
– 15 –
artifacts? The Spitzer data presented in this paper clearly show that the presence of these
shells cannot be confirmed at a level well below the intensities expected from Speck et al.'s
results. Our non-detections call into question not only the ISO results on AFGL 2688, but
all results on the detection of extended envelopes in other objects using the same linear scan
technique described by Speck et al.. More detailed mapping of many high mass-loss rate
objects like AFGL 2688, to search for such shells, is crucially needed.
5. Summary
Spitzer observations of extended envelopes of expanding, dusty outflows from bipolar
AGB and post-AGB stars reveal that there may be a very large dust shell having a radius of
400 arcseconds around IRAS 16342−3814. The combination of the presence of nearby cirrus
emission and the limited azimuthal coverage in the images makes the conclusion for a shell
uncertain, but if the shell is indeed the result of an episodic mass loss event, then it would
represent one of the largest dust shells found so far.
Our observations of AFGL 2688 at 70 µm do not show the dust shells at 150 and
300 arcseconds from the source reported by Speck et al. (2000). Since the dust shell may
be very cool, the non-detection at 70 µm does not alone rule out dust shells. However,
we find that there is only a slight excess at 160 µm of 2 MJy Sr−1 above the background
at the reported location of the outer shell (∼ 300 arcseconds). With greater azimuthal
coverage than was previously obtained with ISOPHOT we find that there is substantial
contamination by galactic cirrus in the region at 160 µm, with galactic cirrus emission above
the sky background present throughout the entire eastern scan path of our observations.
Unfortunately, at 160 µm we can only set an upper limit to the emission from a dust shell
at 150 arcseconds from the source because one of the Airy rings of the PSF overlaps that
region. We also see no extended emission from OH 231.8 at either 70 or 160 µm other than
that attributable to galactic cirrus. In fact, using OH 231.8 as an empirical point source for
PSF subtraction from AFGL 2688 at 70 µm appears to substantially reduce PSF subtraction
residuals compared to using the model PSF.
The limitation of our method of observation is that we have only two radial directions to
probe possible extended emission. For cases like IRAS 16342−3814 where the shell is patchy,
better azimuthal information would help to resolve whether the origin of the emission is from
the star or from galactic cirrus. The MIPS data show that observing very extended emission
in the far infrared is possible, but difficult because of the prominence of galactic cirrus
emission at these wavelengths.
– 16 –
This work is based on observations made with the Spitzer Space Telescope, which is
operated by the Jet Propulsion Laboratory, California Institute of Technology under a con-
tract with NASA. Support for this work was provided by NASA through an award issued
by JPL/Caltech. MM and RS was partially funded for this work from a GO Spitzer award
and an LTSA award (no. 399-20-40-06) from NASA.
Alcolea, J., Bujarrabal, V., S´anchez Contreras, C., Neri, R., & Zweigle, J. 2001, A&A, 373,
REFERENCES
932
Cohen, M. 1981, PASP, 93, 288
Draine, B. T., & Lee, H. M. 1984, ApJ, 285, 89
Gillett, F. C., Backman, D. E., Beichman, C., & Neugebauer, G. 1986, ApJ, 310, 842
Hawkins, G. W. 1990, A&A, 229, L5
Izumiura, H., Hashimoto, O., Kawara, K., Yamamura, I., & Waters, L. B. F. M. 1996, A&A,
315, L221
Jura, M., & Morris, M. 1985, ApJ, 292, 487
Mathis, J. S., Mezger, P. G., & Panagia, N. 1983, A&A, 128, 212
Morris, M. 1987, PASP, 99, 1115
Rieke, G. H., et al. 2004, ApJS, 154, 25
Sahai, R., Hines, D. C., Kastner, J. H., Weintraub, D. A., Trauger, J. T., Rieke, M. J.,
Thompson, R. I., & Schneider, G. 1998, ApJ, 492, L163+
Sahai, R., Te Lintel Hekkert, P., Morris, M., Zijlstra, A., & Likkel, L. 1999, ApJ, 514, L115
S´anchez Contreras, C., Gil de Paz, A., & Sahai, R. 2004, ApJ, 616, 519
Skinner, C. J., Meixner, M., Barlow, M. J., Collison, A. J., Justtanont, K., Blanco, P., Pina,
R., Ball, J. R., Keto, E., Arens, J. F., & Jernigan, J. G. 1997, A&A, 328, 290
Speck, A. K., Meixner, M., & Knapp, G. R. 2000, ApJ, 545, L145
Ueta, T., Murakawa, K., & Meixner, M. 2006, ApJ, 641, 1113
– 17 –
Vassiliadis, E., & Wood, P. R. 1993, ApJ, 413, 641
Yamamura, I., Onaka, T., Kamijo, F., Deguchi, S., & Ukita, N. 1996, ApJ, 465, 926
Young, K., Phillips, T. G., & Knapp, G. R. 1993, ApJ, 409, 725
Werner, M. W., et al. 2004, ApJS, 154, 1
This preprint was prepared with the AAS LATEX macros v5.2.
– 18 –
Object
AFGL 2688
OH 231.8
IRAS 16342−3814
Mean Before
Subtraction
(MJy Sr−1)
σ
(MJy Sr−1)
Mean After
Subtraction
(MJy Sr−1)
σ
(MJy Sr−1)
14.1
10.2
30.3
1.1
0.9
1.3
0.6
0.1
0.3
0.9
0.6
0.9
Table 1: The 70 µm mean surface brightness and sensitivity for a region of uniform back-
ground far from the source, before and after the median background removal described in
the text.
– 19 –
Object
AFGL 2688
OH 231.8
IRAS 16342−3814
Mean Before
Subtraction
(MJy Sr−1)
σ
(MJy Sr−1)
Mean After
Subtraction
(MJy Sr−1)
σ
(MJy Sr−1)
31.4
23.1
90.1
1.7
0.7
2.9
0.3
-0.2
1.0
1.5
0.7
2.3
Table 2: The 160 µm mean surface brightness and sensitivity for a region of uniform back-
ground far from the source, before and after the median background removal described in
the text.
– 20 –
70 µm
160 µm
140
120
100
80
60
40
20
0
64
56
48
40
32
24
16
8
0
14
12
10
8
6
4
2
0
−2
8
8
6
2
L
G
F
A
8
.
1
3
2
H
O
4
1
8
3
-
2
4
3
6
1
S
A
R
I
20.0
17.5
15.0
12.5
10.0
7.5
5.0
2.5
0.0
14
12
10
8
6
4
2
0
40
35
30
25
20
15
10
5
0
Fig. 1.- The three PPNe in our sample after mosaicking and background subtraction. The
units of the colorbars are in MJy Sr−1. The bars in the images are 200′′ in length for scale
and the × marks the location of the point source. Top: AFGL 2688. Center: OH 231.8.
Bottom: IRAS 16342−3814. Note the prominence of the PSF features at 70 µm in all the
images. At 160 µm, the PSF is not apparent in IRAS 16342−3814. North is up and east is
to the left.
– 21 –
Fig. 2.- Model stinytim PSFs in a log stretch to emphasize the Airy rings and diffraction
spikes. Left: The 70 µm PSF. Right: The 160 µm PSF. The lines in the figures are 200
arcseconds in length. For scale, the surface brightness of the diffraction spikes ∼ 200 arc-
seconds from the center is ∼ 10−4 times lower than at the center. See Figures 3 and 5 for
azimuthally averaged plots of the model PSF.
– 22 –
Fig. 3.- PSF subtracted images of AFGL 2688. The color stretches are linear, with the
units in MJy Sr−1. The x marks the location of the point source while the line represents
200′′ for scale. a) PSF subtraction using the stinytim model PSF at 70 µm. The three PSF
diffraction spikes visible in Figure 1 appear to subtract differently, leaving different levels
of residuals. b) PSF subtraction at 70µm using OH 231.8 as a PSF, showing almost no
residuals except for the region at about 60 degrees east of north. c) PSF subtraction using
the stinytim PSF at 160 µm. The feature in the northern scan path coincides with the
position of the diffraction spike in the model PSF. The images have the same orientation as
Figure 1, with north being up.
– 23 –
Fig. 4.- Left: PSF subtraction of OH 231.8 using the model PSF at 70 µm. The residuals
are very similar to those resulting from the PSF subtraction of AFGL 2688 in Figure 3.
Right: PSF subtraction at 160 µm. The northern scan path has more structure than the
eastern scan, probably from galactic cirrus. The units of the colorbars are in MJy Sr−1. The
x marks the location of QX Pup, the central star of OH 231.8. North is up as in Figure 1.
– 24 –
AFGL 2688 70 microns Azimuthal Plots
10000
AFGL 2688 70 microns PSF Subtraction
30
AFGL 2688 70
OH 231.8
Model PSF
Subtraction using Model PSF
Subtraction using OH 231.8
)
r
S
/
y
J
M
(
s
s
e
n
h
g
i
r
t
B
1000
100
10
0
100
200
300
Distance from Center (arcseconds)
)
r
S
/
y
J
M
(
s
s
e
n
h
g
i
r
t
B
25
20
15
10
5
0
400
−5
0
100
200
300
400
Distance from Center (Arcseconds)
Fig. 5.- Left: The azimuthally averaged surface brightness of AFGL 2688, OH 231.8, and
the model PSF at 70 µm. The model PSF and OH 231.8 are scaled to the Airy ring from the
AFGL 2688 PSF at 76 arcseconds. Right: The result of the PSF subtractions. The surface
brightness drops to background levels beyond about 250 arcseconds. The excess emission
left from the PSF subtraction using the stinytim model PSF is very similar to the residuals
left after the same model PSF subtraction of OH 231.8 in Figure 7. The PSF subtraction
using OH 231.8 as a template PSF instead of the model PSF is shown with a dashed line; the
excess emission from the subtraction using the model PSF is almost completely eliminated.
– 25 –
AFGL 2688 160 microns Azimuthal Plots
50
AFGL 2688 160 microns PSF Subtraction
20
Average of both Scans
Eastern Scan Path
Northern Scan Path
Model PSF
ISOPHOT 180 microns
ISOPHOT 120 microns
200
400
600
800
Distance from Center (arcseconds)
)
r
S
/
y
J
M
(
s
s
e
n
h
g
i
r
t
B
40
30
20
10
0
0
)
r
S
/
y
J
M
(
s
s
e
n
h
g
i
r
t
B
15
10
5
0
1000
−5
0
Average of both Scans
Eastern Scan Path
Northern Scan Path
ISOPHOT 180 microns
ISOPHOT 120 microns
200
400
600
800
1000
Distance from Center (Arcseconds)
Fig. 6.- Left: Azimuthally averaged surface brightness of AFGL 2688 at 160 µm. The
two scan paths of the observations are plotted separately to show the differences in their
background levels. The solid line represents the average of the two scans. Beyond about
750 arcseconds from the source, the northern scan path stops so only the eastern path is
represented beyond this distance. The PSF is scaled to the brightness of the Airy ring at
170 arcseconds from the source. Right: The result of PSF subtraction. The region less
than 200 arcseconds from the source may be affected by the PSF subtraction so we cannot
make strong quantitative statements about the emission there at this time, except that the
surface brightness of this region is likely to be about 2 MJy Sr−1.
ISOPHOT data from
Speck et al. (2000) are also plotted for comparison after subtracting the background offset
determined from the median of the values beyond about 400 arcseconds from the source.
The reported ISOPHOT intensity at 300 arcseconds is almost eight times higher than the
surface brightness seen by MIPS at 160 µm after PSF subtraction.
– 26 –
Fig. 7.- Left: the azimuthally averaged surface brightness of OH 231.8 and the model PSF
at 70 µm. The model PSF has been scaled to match the Airy ring emission of OH 231.8 at a
radius of 76 arcseconds. Right: the result of subtraction of the model PSF. The shape of the
residuals is almost the same as those left from the PSF subtraction of AFGL 2688 (Figure
5), falling off to background levels beyond 250 arcseconds from the source.
– 27 –
OH 231.8 160 microns Azimuthal Plots
10000.0
OH 231.8 160 microns PSF Subtracted
10
Source
Eastern Scan Path
Northern Scan Path
Model PSF
100
Distance from Center (arcseconds)
500
400
200
300
)
r
S
/
y
J
M
(
s
s
e
n
h
g
i
r
t
B
8
6
4
2
0
600
−2
0
1000.0
)
r
S
/
y
J
M
(
s
s
e
n
h
g
i
r
t
B
100.0
10.0
1.0
0.1
0
Average Azimuthal Brightness
Eastern Scan Path
Northern Scan Path
100
Distance from Center (Arcseconds)
200
400
300
500
600
Fig. 8.- Left: the azimuthally averaged surface brightness of OH 231.8 along with the
model PSF at 160 µm. Right: the 1-D PSF subtraction. There is no apparent enhanced
emission other than attributable to fluctuations from the galactic cirrus.
– 28 –
100
96
92
88
84
80
76
72
68
)
1
−
r
S
y
J
M
(
s
s
e
n
t
h
g
i
r
B
e
c
a
f
r
u
S
Fig. 9.- The MIPS 70 µm field of view of IRAS 16342−3814 overlaid on the IRAS 100 µm
image of the surrounding regions. Note the abundance of galactic cirrus. The bar above the
frame is 400 arcseconds in length, representing the radius of the potential shell seen with
MIPS around IRAS 16342−3814.
– 29 –
Fig. 10.- Azimuthal average and PSF subtraction of IRAS 16342−3814 at 70 µm. Left:
As with Figure 5, this plot shows the source plotted along with the radial profiles of OH
231.8 and the model PSF. Unlike AFGL 2688, IRAS 16342−3814 shows a definite excess
in its radial profile above the background and PSF profile. Right: the difference between
using the model and OH 231.8 for the PSF subtraction. Both give about the same level of
excess emission of ∼ 2.5 MJy Sr−1 between 300 to 400 arcseconds from the source, where we
see a patchy shell in the MIPS 70 and 160 µm images.
– 30 –
Fig. 11.- Left: The azimuthally averaged surface brightness of IRAS 16342−3814 and the
model PSF scaled to the brightness of the source at the location of the 170 arcsecond radius
Airy ring of the 160 µm PSF. Right: the result of PSF subtraction. The excess emission
seen in the MIPS images is clearly present out to 400 arcseconds.
– 31 –
Fig. 12.- Plot of the expected surface brightness for various mass loss rates. The solid line
shows the expected surface brightness for a smooth constant mass loss envelope with a radius
of 4.2 pc while the dashed lines show the radial profile of a single shell of 1 pc thickness at
the same mass loss rate. The diamonds show the radial profile of IRAS 16342−3814 after
the subtraction of the model PSF. We find that the best fit to the radial brightness profile
obtained from subtracting the model PSF is a two-component model with a shell of radius
4.2 pc and thickness 1 pc requiring a dust mass loss rate of 10−6 M⊙ yr−1 superimposed
upon an envelope from a constant dust mass loss of 10−7 M⊙ yr−1 (dashed blue line). The
asterisks show the radial profile found from using OH 231.8 as a substitute for the model
PSF in the PSF subtraction will fit using only the shell component with a dust mass loss
rate of 10−6 M⊙ yr−1 (dashed orange line). Points closer than a radial distance of ∼ 150
arcseconds are unreliable because they are most affected by PSF subtraction residuals and
are not used in the fit.
|
astro-ph/9510120 | 1 | 9510 | 1995-10-23T18:50:31 | The velocity gradient in the pseudo-photosphere of the peculiar supergiant HD101584 | [
"astro-ph"
] | In this paper preliminary results are presented based on a study of the low and high resolution ultraviolet spectrum of the peculiar supergiant (post-AGB star) HD101584. By a comparison of the low resolution spectrum (1200-3200ang) with standard stars, the star is classified as an A7I, indicating an effective temperature of 8150 K, where literature quotes spectral type F0I. The Doppler shift of the FeII absorption lines in the high resolution spectrum (2500-3000ang) show a relation with the line optical depth. This suggests an expanding accelerating wind, c.q. pseudo-photosphere. The relation is extended by a factor 10^5 in optical depth by using available data from optical HeI and NI lines. The relation suggests that the radial heliocentric velocity of the star is at least 54.5km/s. From the Halpha line a velocity of 96km/s is measured for the terminal velocity of the wind. | astro-ph | astro-ph |
1
THE VELOCITY GRADIENT IN THE PSEUDO-PHOTOSPHERE
OF THE PECULIAR SUPERGIANT HD 101584
ERIC J. BAKKER
SRON Laboratory for Space Research Utrecht
Sorbonnelaan 2
3584 CA UTRECHT
The Netherlands
Abstract
In this paper preliminary results are presented based on a study of the low and
high resolution ultraviolet spectrum of the peculiar supergiant (post-AGB star)
HD 101584. By a comparison of the low resolution spectrum (1200 − 3200 A) with
standard stars, the star is classified as an A7I, indicating an effective temperature
of 8150 K, where literature quotes spectral type F0I. The Doppler shift of the FeII
absorption lines in the high resolution spectrum (2500 − 3000 A) show a relation
with the line optical depth. This suggests an expanding accelerating wind, c.q.
pseudo-photosphere. The relation is extended by a factor 105 in optical depth by
using available data from optical HeI and NI lines. The relation suggests that the
radial heliocentric velocity of the star is at least 54.5 km s−1. From the Hα line a
velocity of 96 km s−1 is measured for the terminal velocity of the wind.
1
Introduction
The star HD 101584 (b = 6o) is classified as a 7.01 visual magnitude F0Iape with
(B − V) = +0.39 (Hoffleit 1983), indicating an effective temperature of 7700 K , log g =
1.7 , and thus (B − V)0 = 0.17 (Landolt-Bornstein 1982). Far- and near-infrared pho-
tometry reveals a strong infrared source at the position of the star (Humphreys & Ney
1974; Parthasarathy & Pottasch 1986). Molecular line observations show bipolar out-
flow for the OH maser (te Lintel Hekkert et al. 1992) and a very complex structure for
the CO(J = 1 → 0) transition (Trams et al. 1990; Loup et al. 1990; van der Veen et
al. 1992). The molecular line emission in OH and CO is normally discussed in terms
of evolved stars, and fits the idea that HD 101584 is a post-AGB star.
The first to classify the star HD 101584 as a post-AGB star were Parthasarathy
& Pottasch (1986). Their conclusion was based on the strong infrared excess of the star
which seems to be due to a large amount of dust around the star. An extensive study
by Trams et al. (1991) shows the resemblance of the infrared excess of HD 101584 with
other known post-AGB stars. It is now reasonably well established that HD 101584 is
a post-AGB star.
2 The ultraviolet spectrum
2.1 The low resolution IUE spectrum
The low resolution IUE spectra (9AA) of HD 101584, a A7I and a F0I standard star are
shown in figure 1. Before fitting the ultraviolet energy distribution to a standard star
the spectrum was smoothed over 12.6 A and corrected for interstellar and circumstellar
Typeset for LHLS workshop
2
Figure 1: The low resolution IUE spectrum of HD 101584 (upper), a standard A7I
star (middle), and a standard F0I star (lower). The spectra are smoothed over 12.6 A,
normalized, and dereddened
extinction (Mathis 1990) using a colour excess, E(B − V) = 0.27, derived for spectral
type A7I.
The low resolution ultraviolet spectrum was best fitted to the standard A7I star,
HD14873. An even better fit is possible if also an A6I star would be available in the
reference atlas (Heck et al. 1984). The spectrum of HD 101584 shows no flux lower
then 1400 A and can therefore not be fitted with a spectrum of a star of spectral type
A5I or hotter.
By comparing the spectrum of HD 101584 with the standard F0I star, α Lep,
an excess of flux for the program star between 1400 and 1700 A indicates a higher
temperature of the star. The slope of the continuum of the spectrum confirms the
supergiant nature of the star. The data on the spectrum of HD 101584 and on the two
reference stars is in table 1.
2.2
the high resolution IUE spectrum
In an extensive study of the high resolution (0.3 A) IUE spectrum of HD 101584 a large
number of absorption features between 2500 and 3000 A has been identified (Bakker
1994). The main conclusions from this work are that the spectrum of HD 101584 has
in principle the same absorption features as the F0 supergiant, α Lep, but the lines are
intrinsically broader and the lines are asymmetric in shape. This study limits itself to
the measured radial velocities of the absorption features as derived from the measured
Doppler shift of the core of the absorption profile.
Typeset for LHLS workshop
3
Table 1: Data on the low resolution spectra
HD number Name
Spectral E(B-V) Normalized on
[erg cm−2 s Hz]
type
4.1 10−23
A7I
6.0 10−22
α Lep F0I
2.8 10−23
A7I
0.27
0.04
0.25
HD 101584
HD 36673
HD 148743
Table 2: Stellar parameters for HD 101584 based on literature and on the low resolution
ultraviolet IUE spectra studied in this work
Work
Hoffleit 1983 F0I
This study
A7I
Spectral Effective
Type
Temperature
7700
8150
log g
(B − V)0 E(B − V)
1.7
1.9
+0.17
+0.12
0.22
0.27
Teff , log g, and (B − V)o from Landolt-Bornstein (1982) and are based on the given
spectral type.
In probing the photosphere of a star the line optical depth is a measure for the
depth in the photosphere seen. The stronger the optical depth τ the more outside layers
of the photosphere will be probed. The range in optical depth from the ultraviolet
spectrum is limited to about a factor 104. By incorporating some of the available
optical data, which are from much weaker lines, the deeper layers of the photosphere
can be probed as well and the relation can be extended over a much wider range of τ
by a factor of 105 to 109. There are however two assumptions in making the relation
valid for a larger range of τ . The first is solar abundance ratios, and that NI, HeI,
and FeII are the dominant ionization stages of these elements in the stellar wind (or
pseudo-photosphere). The second is that the effective temperature derived from the
energy distribution in the ultraviolet represents the real temperature of the gas. If the
first assumption is violated, the separate elements will shift horizontally in the fig. 2. If
the second assumption is violated, the relation within FeII will change, and there will
occur a small horizontal shift between the different elements. The data on FeII, NI, HeI
and Hα will be published in a separate paper which is in preparation (Bakker 1995).
Figure 2 shows the relation between the logarithm of the strength of an absorption
line and the heliocentric radial velocity measured for that line. The crosses are the FeII
lines from the ultraviolet spectrum, the triangle the optical nitrogen line (8680.24 A),
and the asterisk the optical helium line (5875.618 A).
Vhelio(FeII) = 49.1 − 5.1 ×(cid:20)log N + log gf −
5040χ
Teff (cid:21)(cid:2)km s−1(cid:3)
(1)
A first order approximation of the relation for the FeII line is given by eq. 1. The
reason that the HeI and NI line are not used for this first order approximation is that is
seems logical to assume a constant velocity for weak lines. These lines are formed in the
deeper layers of the photosphere and are therefore least affected by the unknown force
which accelerates the pseudo-photosphere outwards. The dashed line in fig. 2 shows the
first order approximation based on only the FeII lines. It is surprising to see that the
HeI and NI line fit this relation almost perfectly.
Typeset for LHLS workshop
4
Table 3: Main parameters of the chemical elements used in probing the pseudo-
photosphere
Element
He
N
Fe
Solar Abundance
log N
10.93
7.96
7.60
Ion
HeI
NI
FeII
Excitation
Ionization
Energy (eV) Energy (eV)
24.587
14.534
16.16
20.87
10.29
7.870 → 4.48
Solar Abundance, and ionization energies are from Allen (1985).
Figure 2: Relation between the strength of an absorption line and the radial velocity
measured for that line, indicating that we are looking at an accelerating expanding
wind. The dashed line is a first order fit based on the FeII lines. The HeI and NI line
seem to fit this relation exceptionally well. Where θ is 5040/Teff, χ is the excitation
potential of the lower level of the transition, and A is the abundance relative to H by
number
Typeset for LHLS workshop
5
3 Terminal velocity of the stellar wind
An upper limit on the maximum out streaming velocity (a lower limit on the blue shift
of a line) is given by the absorption part of the Hα line profile. The edge of the Hα
line profile is pretty steep, implying that hydrogen column density at that velocity does
not slowly decrease due to expansion (and thus dilution) of the gas, but rather that the
edge in the profile is due to hydrogen at the terminal velocity of the stellar wind. The
maximum out streaming velocity is 120 km s−1, and the out-streaming velocity derived
from the Doppler shift of the core of the absorption profile is 96 km s−1. The latter
velocity is the terminal velocity of the wind. The first is the net velocity maximum due
to turbulent motion in the wind and its expansion. If we assume that the velocity of the
star is best represented by the velocity of the HeI line, a minimum heliocentric velocity
of −41 km s−1 is expected. The dotted line in fig. 2 represents the terminal velocity of
the wind as determined from the Hα profile.
4 Discussion
The data in the literature concerning the radial velocity of HD 101584 is very confusing.
It's variations are not well understood and is a fruitful base for wild speculations. From
this study a little light is shed on these variations. The relation as shown in figure 2
reveals that variations in radial velocity within one spectrum (FeII lines) can be un-
derstood in terms of an expanding accelerating photosphere or wind. In the following
discussion this line absorbing region will be called pseudo-photosphere. This relation
can be extended to lower optical depth by incorporating the helium (5875.618 A) and
nitrogen (8680.24 A) optical absorption lines. That this relation holds for helium sug-
gests that this line is from the same star and not from a yet unseen hot companion
star. To produce a helium absorption line the star has to be much hotter than spectral
type A7I, probably even a B-type star. This contradicts the fact that the low resolution
ultraviolet spectrum is best fitted with a A7I reference star
Although the relation does not seem to go to a constant velocity for weaker lines
(HeI and NI) is seems reasonable to assume that the velocity for the HeI line is the
stellar velocity. By monitoring the velocity of the HeI absorption line it should be
possible to make a statement about the binary nature of the star. Radial velocity
measurements of stronger lines do not only have a contribution of the radial velocity of
the star due to binarity (if this would be the case), but will also have a contribution
from the pseudo-photosphere. These two contributions will be very hard to disentangle.
A maximum velocity of the out streaming wind as determined from the Hα profile is
96 km s−1. This means that velocities of absorption lines are to be expected in the
range between −41 km s−1 and 54.5 km s−1.
Table 4 summarizes the stellar parameters of HD 101584 as determined in this
study from ultraviolet spectra.
Acknowledgements The author would like to thank Dr. Norman Trams for making
the HIRES spectrum of HD 101584 available for this work. Dr. Christoffel Waelkens
was so kind to make the optical data of HeI, NI, and Hα lines available. Dr. Rens
Waters and Prof. Dr. Henny Lamers made a large contribution to this work by having
many discussion with the author. The author was supported by grant no. 782-371-040
by ASTRON, which receives funds from the Netherlands Organization for the Advance-
ment of Pure Research (NWO). This research has made use of the Simbad database,
operated at CDS, Strasbourg, France.
Typeset for LHLS workshop
6
Table 4: Improved data on HD 101584
Ultraviolet spectral type
Effective temperature
E(B − V)
Heliocentric velocity of star
Terminal velocity of wind
A7I
8150 K
0.27
54.5 km s−1
96 km s−1
54.5 km s−1
Maximum helio. velocity expected
Minimum helio. velocity expected −41 km s−1
References
[1985] Allen C.W.: 1985, Astrophysical Quantities, third edition
[1994] Bakker E.J.: 1994, A&AS 103, 189
[1995] Bakker E.J.: 1995, A&A in accepted
[1984] Heck A., Egret D., Jaschek M., Jaschek C.: 1984, IUE Low-Dispersion Spectra
Reference Atlas, Part 1. Normal Stars
[1983] Hoffleit D, Saladyga M., Wlasuk P.: 1983, Supplement to the Bright Star Cata-
logue, Yale University Observatory, USA
[1974] Humphreys R.M., Ney E.P.: 1974, ApJ 190, 339
[1982] Landolt-Bornstein 1982, Numerical Data and Functional Relationships in Science
and Technology, New Series, Group VI, Vol. 2b.
[1992] te Lintel Hekkert P., Chapman J.M., Zijlstra A.A.: 1992, ApJ 390, L23 prepa-
ration
[1990] Loup C., Forveille T., Nyman L. A, Omont A.: 1990, A&A 227, L29
[1990] Mathis J.S.: 1990, ARA&A 28. 37
[1986] Parthasarathy M, Pottasch S.R.: 1986, A&A 154, L16
[1990] Trams N.R., van der Veen W.E.C.J., Waelkens C., Waters L.B.F.M., Lamers
H.J.G.L.M.: 1990, A&A 233, 153
[1992] Van der Veen W.E.C.J., Trams N.R., Waters L.B.F.M.: 1992, A&A, submitted
[1991] Trams N.R., Waters L.B.F.M., Lamers H.J.G.L.M., Waelkens C., Geballe T.R.,
Th´e P.S.: 1991, A&AS 87, 361
|
astro-ph/9511091 | 1 | 9511 | 1995-11-20T14:08:43 | The Ionization Fraction in Dense Clouds | [
"astro-ph"
] | We present submillimeter observations of various molecular ions toward two dense clouds, NGC 2264 IRS1 and W 3 IRS5, in order to investigate their ionization fraction. Analysis of the line intensity ratios by the way of statistical equilibrium calculations allows determination of the physical parameters: n(H2)~(1-2)e6 cm-3 and T(kin)~50-100 K. Column densities and abundances are also derived. Together, the abundances of the observed ions provide a lower limit to the ionization fraction, which is (2-3)e-9 in both clouds. In order to better constrain the electron abundance, a simple chemical model is built which calculates the steady state abundances of the major positive ions, using the observed abundances wherever available. With reasonable assumptions, good agreement within a factor of two with the observations can be achieved. The calculated electron fraction is x(e)= (1.0-3.3)e-8 in the case of NGC 2264 and x(e)=(0.5-1.1)e-8 for W 3 IRS5. In the first case, the high abundance of N2H+ requires a rather high cosmic ray ionization rate >1e-16s-1, even if all nitrogen is assumed to be in gas phase N2. For W 3 IRS5, ionized metals such as Fe+ and Mg+ could provide 60% of the electrons. | astro-ph | astro-ph | THE IONIZATION FRACTION IN DENSE CLOUDS
C. de Boisanger
, F.P. Helmich
, and E.F. van Dishoeck
;
Commissariat (cid:18)a l'Energie Atomique, Centre d'Etudes de Bruy(cid:18)eres-le-Ch^atel,
Service PTN, F- Bruy(cid:18)eres-le-Ch^atel, France.
Leiden Observatory, P.O.-Box , RA Leiden, The Netherlands
Received August ; Accepted: October
Astronomy and Astrophysics, in press.
5
9
9
1
v
o
N
0
2
1
9
0
1
1
5
9
/
h
p
-
o
r
t
s
a
Abstract
We present submillimeter observations of various molecular ions toward two dense
clouds, NGC IRS and W IRS, in order to investigate their ionization frac-
tion. Analysis of the line intensity ratios by the way of statistical equilibrium cal-
culations allows determination of the physical parameters: n(H
) (cid:24) ( (cid:0) )
cm
(cid:0)
and T
(cid:24) (cid:0) K. Column densities and abundances are also derived. Together,
kin
the abundances of the observed ions provide a lower limit to the ionization fraction,
which is ( (cid:0) )
in both clouds. In order to better constrain the electron abun-
(cid:0)
dance, a simple chemical model is built which calculates the steady state abundances
of the ma jor positive ions, using the observed abundances wherever available. With
reasonable assumptions, good agreement within a factor of two with the observations
can be achieved. The calculated electron fraction is x
= (: (cid:0) :)
in the case
e
(cid:0)
of NGC and x
= ( : (cid:0) :)
for W IRS. In the (cid:12)rst case, the high
e
(cid:0)
abundance of N
H
requires a rather high cosmic ray ionization rate >
s
,
+
(cid:0)
(cid:0)
even if all nitrogen is assumed to be in gas phase N
. For W IRS, ionized metals
such as Fe
and Mg
could provide % of the electrons.
+
+
Key words: ISM: molecules { ISM: clouds { ISM: individual: W IRS, NGC IRS
Introduction
The ionization fraction of dense clouds is a basic parameter of interstellar physics,
since it determines the e(cid:14)ciency with which magnetic (cid:12)elds couple with the gas and
ultimately controls the rate of collapse and star formation. Moreover, electrons and
ions play a ma jor role in interstellar chemistry, and their abundances provide infor-
mation on the nature and importance of sources which ionize the gas. It is well known
that in the di(cid:11)use gas, interstellar photons with (cid:21) >
A provide most of the ioniza-
(cid:23)
tion so that the electron fraction is of order
, equal to the abundance of the ma jor
(cid:0)
supplier C
. This fraction is thought to drop by several orders of magnitude inside
+
dense clouds, where the ionization mostly results from cosmic ray interactions. The
ma jor ions here are thought to be H
, H
, He
, HCO
, H
O
and possibly metal
+
+
+
+
+
ions (e.g. Fe
, Mg
). Unfortunately, except for HCO
, all of these species are very
+
+
+
di(cid:14)cult to observe directly for spectroscopic reasons.
The (cid:12)rst attempts to derive the electron abundance in dense clouds date back more
than years (Watson ; Gu(cid:19)elin et al. , ; Wootten et al. ). The basic
idea was to use the observed DCO
/HCO
ratio as a measure of the H
D
/H
ratio,
+
+
+
+
which, together with a model for the deuterium fractionation and the H
recombina-
+
tion, gives a limit on the electron abundance. Typical electron fractions derived in
this way were
(cid:0)
, with an order of magnitude variation from cloud to cloud
(cid:0)
(cid:0)
(Langer ). However, this method was abandoned in , when measurements
by Adams & Smith ( ) found the dissociative recombination rate of H
with e
+
to be negligibly small, thereby invalidating the earlier determinations and raising the
upper limits to
. The H
rate has subsequently been a heated topic of discussion
(cid:0)
+
in astrochemistry for more than a decade, preventing further studies of the ionization
fraction.
There are several reasons why a new attack on this problem is now both timely and
warranted. First, the latest set of laboratory measurements on the H
dissociative
+
recombination rate seem to converge to the same (relative rapid) value under inter-
stellar conditions (Amano ; Canosa et al. ; Larsson et al. ), although
some lingering uncertainties remain (Smith &
Spanel ). Second, many more ions
(cid:21)
can now be observed with the new sensitive, high frequency receivers: apart from
+
+
+
+
HCO
and its isotopes, these include HCS
, N
H
and H
O
(Phillips et al. ),
whereas good upper limits on H
D
exist for a number of regions (van Dishoeck et al.
+
). Third, the dissociative recombination rates of HCO
, HCS
, and N
H
have
+
+
+
also recently been measured in the laboratory (Rowe et al. ). Fourth, the exci-
tation and optical depth of the lines in dense clouds is now much better constrained
by observations of several submillimeter lines, thereby minimizing the uncertainties in
abundance determinations and physical conditions.
We present here observations of various ions (HCO
, H
CO
, DCO
, HCS
and
+
+
+
+
+
N
H
) in two di(cid:11)erent dense clouds, NGC and W IRS, in order to investi-
gate their ionization fraction. The clouds are chosen to have a range in density and
temperature (as determined from our earlier H
CO data) and to have complementary
+
+
H
O
and/or H
D
emission observations available (Phillips et al. ; van Dishoeck
et al. ), as well as infrared absorption line data on H
(Black et al. ) and
+
CO/H
(Lacy et al. ). Previous studies in cold dark clouds such as TMC-, with
n(H
) =
cm
, have found :
(cid:20) x
(cid:20) :
, depending on the amount of
e
(cid:0)
(cid:0)
(cid:0)
PAHs and metals included in the modeling (Schilke et al. ). The sources studied
in this work have higher temperatures and densities up to two orders of magnitude,
and it will be interesting to investigate whether the electron fraction in them is lower,
as predicted by theory. The observations and results are presented in Sections {,
and their analysis in Section .
Together, the abundances of these ions provide a (cid:12)rm lower limit to the ionization
fraction of the cloud.
In order to constrain the electron abundance, more detailed
modeling is needed. The basic idea is that the molecular ions are formed by protona-
tion of abundant neutral species in reactions with H
:
+
+
+
X + H
! XH
+ H
where X=CO, H
O, N
, CS: : : Their destruction is governed by reactions with CO,
H
O and O and by dissociative recombination with an electron in the gas phase:
+
XH
+ e ! X + H
Thus by measuring XH
and the abundances of the parents X for a su(cid:14)cient number
+
of species, it should be possible to constrain both the electron fraction and the H
+
abundance (or equivalently the cosmic ray ionization rate (cid:16)
) in dense clouds. This
H
+
technique was pioneered by Wootten et al. ( ) based on the HCO
/CO ratio, but
is extended here to include many more ion/neutral pairs. The actual modeling, which
involves a more detailed network of reactions, is presented in Section .
Observations
The sources studied in this work are both associated with active regions of star for-
mation. NGC IR is a well studied molecular cloud core (d (cid:24) pc) (Schwartz
et al. ; Kr(cid:127)ugel et al ). Its shape on an optical plate resembles a dark cone
(the \Cone Nebula"). Presumably an intermediate mass star is forming here. The
infrared source is bright (Allen ) and has been used as background light source
for infrared absorption line studies of gas and solid phase species. In particular, an
upper limit of :
exists on the H
abundance in the gas (Black et al. ), and
(cid:0)
+
a lower limit of .
on the CO/H
ratio (Lacy et al. ).
(cid:0)
W IRS is the brightest infrared source (L = :
L
(Ladd et al. )) in the
(cid:12)
W Giant Molecular Cloud core (d (cid:24) : kpc). A large amount of dense and warm
material surrounds the massive star forming inside (Dickel ; Dickel et al. ;
Hasegawa et al. ; Hayashi et al. ; Tieftrunk et al. ; Helmich et al. ).
Because of the enormous visual extinction (A
> mag) the ultraviolet radiation
V
from the young stars is not expected to in(cid:13)uence the ionization balance deep inside
the cloud.
The observations of NGC were performed primarily at the James Clerk Maxwell
Telescope (JCMT
) in February , with additional data taken in April, June and
November . The facility receivers at GHz (A), GHz (Bi) and GHz
(C) were employed. These are double side-band receivers with upper and lower side-
bands . GHz (A, Bi) or . GHz (C) apart. No systematic side-band checks
were performed, since the spectra are simple and often there is only a single line in the
spectrum. The image frequencies were compared against spectral-line catalogues and
no obvious blending of lines was found. As the backend, the Digital Autocorrelation
Spectrometer (DAS) was used in di(cid:11)erent spectral resolutions (mostly . , . and
. MHz/channel). To improve the signal to noise, the spectra were smoothed to
. MHz/channel and in some cases to . MHz/channel. The former corresponds
to (cid:24) . km s
at GHz. Integration times were generally minutes on+o(cid:11) for
(cid:0)
and GHz receivers but up to an hour for lines in the GHz window, because
of the higher system temperatures. This resulted in typical (cid:27) noise in T
of (A){
A
(cid:3)
(Bi) mK and { (C) mK per resolution element ( . MHz). The pointing
was checked regularly on OMC- and was found to be within
.
The calibration at the JCMT was performed with the chopper-wheel method.
In
February , the beam e(cid:14)ciency was somewhat uncertain, because of inaccuracies
in the shape of the telescope surface. However, careful comparison with earlier runs
showed a reduction by only (cid:24) %. This e(cid:11)ect is corrected for by using an e(cid:11)ective
main beam-e(cid:14)ciency of ., . and . for receiver A, Bi and C, respectively.
During other runs, values of ., . and ., respectively, were used. In general
we expect the absolute calibration to be accurate to about %; relative uncertainties
can be signi(cid:12)cantly smaller for all runs. The beam size of the JCMT at , and
GHz is (cid:24)
,
and
, respectively. For most spectra, position-switching by
was found to be appropriate, except for the
CO lines for which a larger switch
was used. Additional data on other species toward NGC IRS will be presented
by Schreyer et al. ( ).
A limited number of spectra of NGC were taken at the Caltech Submillimeter
Observatory (CSO
) in January with the GHz (H
CO at GHz) and
GHz (C
O {) receivers. The low ( MHz) and high ( MHz) resolution
acousto-optical spectrometers were used as backends. Pointing is reproducible within
on the CSO. Main beam e(cid:14)ciencies of . and . were adopted, respectively.
The CSO data at GHz refer to a beam of (cid:24)
; those at GHz to a
beam.
The James Clerk Maxwell Telescope is operated by the Observatories on behalf of the Particle
Physics and Astronomy Research Council of the United Kingdom, the Netherlands Organisation for
Scienti(cid:12)c Research, and the National Research Council of Canada.
The Caltech Submillimeter Observatory is operated by the Californian Institute of Technology
under funding from the U.S. National Science Foundation (AST - ).
Earlier observations by van Dishoeck et al. ( ) with the CSO have resulted in
upper limits on the H
D
abundance. These same spectra also reveal surprisingly
+
strong N
H
{ emission in this source in a
beam.
+
The observations presented here for W IRS form part of the large JCMT GHz
survey of three star-forming cores in the W cloud (IRS, IRS and W (H
O))
Figure : Observed spectra toward: a and b NGC IRS and c W IRS. Note
that the DCO
{ line toward NGC consists of two components. Only the
+
broad component was used in the analysis. The N
H
{ line represents the (cid:12)rst
+
+
detection of this line in the interstellar medium. The N
D
{ line may be present at
the (cid:27) level. Since W IRS is a well-known out(cid:13)ow source, only the central Gaussian
was used in the analysis. The line next to the HCS
{ feature at V
= (cid:0) km s
LSR
+
(cid:0)
is the
SO
(cid:0)
line from the image side-band.
;
;
between January and November . A complete log and details of the obser-
vations will be given in a separate publication concerning the spectral survey (Helmich
& van Dishoeck ; see also Helmich et al. for (cid:12)rst results). For all runs, the
facility receivers (A, A, Bi and C) were used. Prior to July , the acousto-
optical spectrometer AOSC ( MHz) was employed as the backend; subsequently
the DAS was used in di(cid:11)erent spectral resolutions, mostly in its MHz bandwidth
mode. Binning of two channels resulted in a typical resolution of . MHz/channel.
As for NGC , no systematic side band identi(cid:12)cation was performed, because the
line crowding in IRS, compared with the two other W sources in the spectral-line
survey, is such that there is little ambiguity and possible blends are easily identi(cid:12)ed.
The pointing was checked regularly on the nearby continuum source W (OH) and is
within
. The observations of W IRS were performed in the
beam-switch
mode, except for the
CO lines where a larger position switch was used.
Additional observations of the H
O
ion toward W IRS were obtained by Phillips
+
et al. ( ) using the CSO, whereas an upper limit on H
D
was obtained by van
+
Dishoeck et al. ( ). The latter study also reported detection of the N
H
{ line.
+
Results
Tables and summarize the resulting Gaussian (cid:12)t parameters for the various lines
for the two sources. Examples of spectra are given in Fig. a, b and c. Although the
pro(cid:12)les of the lower lines of DCO
and other species toward NGC suggest the
+
presence of two components (Gu(cid:19)elin et al. ; Schreyer et al. ), it was found
that the lines observed in this work are well represented by single Gaussians with little
spread in the line widths, the optically thick
CO { and HCO
{ lines being
+
exceptions. Most other lines are optically thin, or at most moderately optically thick.
For example, the C
S { line is weak, and the intensity ratio CS/C
S={ is only
slightly smaller than the cosmic [
S]/[
S] abundance ratio of . . As mentioned
above, N
H
is very strong in this source; the { line at GHz presented in Fig. b
+
is the (cid:12)rst detection of this line in any astrophysical source. N
D
{ appears present
+
as well, but only at the (cid:27) level.
The lines toward W IRS are also well represented by single Gaussians but there is a
much larger spread in the line widths. The molecules HCS
, C
S and C
S have much
+
narrower lines than e.g., H
CO
, HCO
and CO-isotopomers. It is easy to explain
+
+
the larger line widths for CO and HCO
by optical depth e(cid:11)ects, but this becomes
+
more di(cid:14)cult for the rarer species. Part of the explanation may lie in inaccurate
Table : Parameters of Gaussian (cid:12)ts for NGC
Line
T
(cid:1)V
T
dV
MB
MB
R
(K)
km s
K km s
(cid:0)
(cid:0)
CO (cid:0)
.
.
C
O (cid:0)
.
.
.
C
O (cid:0)
.
.
.
CS (cid:0)
.
.
CS (cid:0)
.
.
CS (cid:0)
(cid:20) .
C
S (cid:0)
.
.
.
HDO
(cid:0)
(cid:20) .
HCO
(cid:0)
.
.
.
+
+
H
CO
(cid:0)
.
.
.
+
H
CO
(cid:0)
.
.
.
DCO
(cid:0)
.
.
.
+
+
HCS
(cid:0)
.
.
.
+
N
H
(cid:0)
.
.
.
+
N
D
(cid:0)
.
.
.
[C i]
P
(cid:0)
P
.
.
.
Coordinates (B . ): (cid:11) = . , (cid:14) = .
All data were taken with the JCMT except for C
O (cid:0) .
The upper limits are (cid:27) in a . MHz channel.
determinations of the C
S and HCS
line widths, because these lines are so weak.
+
However, source structure can play a r^ole as well. A complete investigation is beyond
the scope of this paper; therefore, for each species its observed line width is adopted
in the analysis. The detection of C
S implies that even the C
S lines are slightly
optically thick. From the
CS line, however, it can be seen that the optical depth
must be small.
Table : Parameters of Gaussian (cid:12)ts for W IRS
Line
T
(cid:1)V
T
dV
MB
MB
R
(K)
km s
K km s
(cid:0)
(cid:0)
CO (cid:0)
.
.
CO (cid:0)
.
C
O (cid:0)
.
.
.
C
O (cid:0)
.
.
.
CS (cid:0)
.
.
.
C
S (cid:0)
.
.
.
C
S (cid:0)
.
.
.
C
S (cid:0)
.
.
.
C
S (cid:0)
.
.
.
CS (cid:0)
.
.
.
HCO
(cid:0)
.
.
+
+
H
CO
(cid:0)
.
.
.
+
H
CO
(cid:0)
.
.
.
+
HC
O
(cid:0)
.
.
.
DCO
(cid:0)
.
.
.
+
+
HCS
(cid:0)
.
.
.
+
HCS
(cid:0)
.
.
.
+
N
H
(cid:0)
< .
+
N
D
(cid:0)
< .
[C i]
P
(cid:0)
P
.
.
Coordinates (B . ): (cid:11) = ., (cid:14) = .
All data were taken with the JCMT.
The upper limits are (cid:27) in a . MHz channel.
Analysis
. Physical parameters
In order to derive the column densities from the observed line intensities, (cid:12)rst the
physical parameters in the sources have to be determined. Statistical equilibrium
calculations were performed along the lines described by Jansen et al. ( ) and
Jansen ( ). These models take the various collisional and radiative excitation and
de-excitation processes into account, and treat the radiative transfer with an escape
probability method. Because the observed lines are mostly optically thin, the precise
details of the radiative transfer method do not a(cid:11)ect the results. The ratios of some
H
CO lines are particularly sensitive to temperature variations (see e.g. Jansen et al.
; Helmich et al. ; Mangum & Wootten ) and are thus very well suited
for kinetic temperature determinations. Lines from linear rotors such as CS, C
S, or
ions like H
CO
, on the other hand, are more sensitive to density.
+
Line intensity ratios were calculated for a range of densities and temperatures for
H
CO, CS, H
CO
and N
H
, and are plotted in Fig. a, together with the observed
+
+
ratios for NGC (see also Schreyer et al. ). The contour plot of the CS {/{
line ratio is produced for the slightly optically thick case appropriate for this source.
All other ratios refer to optically thin lines. From the H
CO
(cid:0)
and
(cid:0)
lines
at GHz measured in the same frequency setting, a kinetic temperature of
K is
+
(cid:0)
+
found. Furthermore it is seen from Fig. a that the measurements for the N
H
and CS
ratios indicate a similar density of ((cid:0))
cm
. Thus, these species probably probe
(cid:0)
the same region. The density indicated by H
CO
appears to be somewhat lower
+
than that found from the two other species. This di(cid:11)erence is di(cid:14)cult to account for
in terms of optical thickness, but could be due, e.g., to uncertainties in the collisional
rate coe(cid:14)cients or source structure. The measurements are, however, within the errors
in agreement with each other and give a number density of n(H
)=( (cid:6) :)
cm
.
(cid:0)
Maps of the CS { and { and
CO { lines by Schreyer et al. ( ) show that
the temperature cannot be much lower than that found from H
CO. Note that the
above single temperature and density analysis assumes that all species and lines are
similarly distributed in the beam. No corrections are made for the fact that the higher
frequency JCMT data refer to a smaller beam than those at lower frequencies.
For W IRS the temperature T
(cid:25)
K and density n(H
)(cid:25)
cm
derived
kin
(cid:0)
+
(cid:0)
in Helmich et al. ( ) from various formaldehyde lines was adopted. The H
CO
+
{/{ and C
S {/{ line ratios, presented in Fig. b, show that these molecules
probe about the same density, suggesting that they are present in the same gas. It is
Figure : Contour plot showing ratios of lines. Shaded regions refer to the observed
range toward the source: a NGC , b W IRS. All ratios refer to optically thin
lines, except the CS {/{ ratio toward NGC , where N (CS) = :
cm
(cid:0)
and (cid:1)V = : km s
was adopted. The N
H
ratio for W IRS is uncertain and
(cid:0)
+
is therefore plotted as an upper limit.
assumed that the emission of all molecules described here (cid:12)lls our beams.
The inferred densities toward our two sources are indeed two orders of magnitude
larger than those found in dark clouds such as TMC-, which have been the sub ject
of previous studies of the electron fraction.
. Column densities
The total beam-averaged column densities for each species are calculated by (cid:12)tting
the absolute intensities, using the temperature and density derived from our statistical
equilibrium calculations. The results are given in Tables and for the two sources.
Since the HCO
{ line is optically thick toward NGC , the column density
+
of HCO
is determined from H
CO
, using the [
C=
C] ratio of obtained by
+
+
Mitchell & Maillard ( ) for this source. The
CO column density was calculated
from the C
O { line assuming that the CO/C
O ratio is the same as the cosmic
[
O]/[
O] ratio of . The lower limit on the CO/H
ratio found by Lacy et al.
( ) of :
then gives an upper limit on the total beam-averaged column-
(cid:0)
density N (H
)(cid:25) :
cm
. Although the CO/H
ratio derived from the pencil-
(cid:0)
beam infrared absorption line observations does not necessarily apply to the whole
volume sampled by the emission data, the CO column density derived from the two
methods agree within a factor of two.
The total (beam-averaged) column density toward W IRS is derived from the C
O
{ line. In contrast with Helmich et al. ( ), a CO/H
abundance ratio of :
(cid:0)
was used instead of
. The value in Helmich et al. ( ) was chosen because the
(cid:0)
study of Tielens et al. ( ) of the solid CO seemed to indicate a lower value. The
subsequent study of Lacy et al. ( ) however suggests a much larger ratio for warm
sources like NGC IRS. Since the gas and grain temperatures toward W IRS
are even higher than those toward NGC , a higher value is probably more appro-
priate. This value decreases the total beam-averaged column density from
cm
(cid:0)
to :
cm
, and at the same time increases the abundances by a similar factor.
(cid:0)
It is stressed again, however, that the derived column-densities and abundances are
uncertain for this source due to the unknown structure and distribution of the various
species within the beam. The relative values should be more reliable, however.
For both clouds the observation of the [C i] GHz line provides good upper limits
on the beam averaged column density and abundance of atomic carbon since a large
fraction of this emission probably arises from the outer skin of the cloud.
Tables and include column densities and abundances (w.r.t. H
) for other relevant
Table : Column densities and abundances in NGC
Species
N
x
(cid:0)
(cm
)
CS
:
:
(cid:0)
(cid:0)
C
S
:
:
HDO
(cid:20) :
(cid:20) :
(cid:0)
+
(cid:0)
H
CO
:
:
HCO
:
:
+
(cid:0)
DCO
:
:
+
(cid:0)
+
(cid:0)
HCS
:
:
+
(cid:0)
N
H
:
:
+
(cid:0)
N
D
:
:
+
(cid:0)
H
D
< :
< :
+
H
< :
< :
(cid:0)
H
O
< :
< :
(cid:0)
C
(cid:20) :
(cid:20) :
(cid:0)
species obtained from the literature for the two sources. In the case of NGC
we adopted the H
D
column density derived by van Dishoeck et al. ( ), with
+
slightly di(cid:11)erent total H
density and temperature. The H
O column density is taken
from Wannier et al. ( ) who derived H
O/CO < . in NGC from searches
for H
O using the Kuiper Airborne Observatory in a .
beam. For H
O
toward
+
W IRS, the data given in Phillips et al. ( ) were reanalysed with the above
density and temperature. From observations of the H
O GHz line presented by
Cernicharo et al. ( ), Phillips et al. ( ) determined x(H
O) (cid:24)
(cid:0)
. The
(cid:0)
(cid:0)
O
limit of <
cm
for that source is based on a deep search for the
O
O
(cid:0)
GHz line (Keene, van Dishoeck & Phillips, unpublished result).
Models
In order to estimate the electron fraction, we built a simple chemical model which
calculates the steady state abundances of the ma jor positive ions, namely: H
, H
,
+
+
+
+
+
+
+
+
+
+
+
+
+
H
, He
, C
, H
O
, HCO
, DCO
, N
H
, H
D
, HCS
, N
D
and H
DO
. The
electron density is set equal to the sum of the densities of the positive ions, including
Table : Column densities and abundances in W IRS
Species
N
x
(cid:0)
(cm
)
CS
:
:
(cid:0)
(cid:0)
C
O
:
:
+
(cid:0)
H
CO
:
:
HCO
:
:
+
(cid:0)
DCO
:
:
+
(cid:0)
+
(cid:0)
HCS
:
:
+
(cid:0)
N
H
(cid:20) :
(cid:20) :
+
(cid:0)
N
D
(cid:20) :
(cid:20) :
+
(cid:0)
H
O
:
:
+
(cid:0)
H
D
(cid:20) :
(cid:20) :
O
(cid:20) :
(cid:20) :
(cid:0)
C
(cid:20) :
(cid:20) :
(cid:0)
metals or PAH ions. Only gas phase reactions are included. The temperature, density
of molecular hydrogen and abundances of CO and CS are those derived from the
observations for each cloud.
The abundances of O, O
, H
O, N, N
, D and HD are free parameters, taking into
account the constraints imposed by other observations or models, and allowing for
a certain amount of depletion on grains. The main motivation of this approach is
that detailed chemical models generally give O
and H
O abundances higher than
the upper limits derived from the observations (x (cid:24)
(cid:0)
) (see Tables and
(cid:0)
(cid:0)
). Moreover these models usually do not allow for grain surface reactions and do
not consider deuterated varieties.
In our case the abundance of O
does not play
an important role since this species contributes solely to the destruction of He
, H
+
+
and C
, for which we do not have observed abundances, and for which the calculated
+
abundances are low in any case. Other free parameters are the cosmic ray ionization
rate (cid:16)
and the abundance of ionized metal x
+
.
H
M
Concerning the reaction rates, special attention has been paid to the dissociative re-
combination rate coe(cid:14)cients of HCS
, HCO
, N
H
and H
, which have recently been
+
+
+
+
studied in the lab. For HCO
and H
, the rate coe(cid:14)cients measured by Rowe et al.
+
+
( ) and Amano ( ) are very close: from their work, we derive the temperature-
dependent values:
k
+
= :
(T = )
cm
s
;
HCO
(cid:0)
(cid:0) :
(cid:0)
k
+
= :
(T = )
cm
s
:
(cid:0)
(cid:0) :
(cid:0)
H
In fact, Smith &
Spanel ( ) obtain a lower value k
+
=
cm
s
at K,
(cid:21)
(cid:0)
(cid:0)
H
+
but use of this value does not a(cid:11)ect our results since H
is mainly destroyed by CO,
due to the high density and the high abundance of CO.
For N
H
, the rate measured by Amano ( ) can be (cid:12)tted by:
+
k
= :
(T = )
cm
s
:
+
N
H
(cid:0)
(cid:0) :
(cid:0)
For HCS
, Rowe et al. ( ) found a rate coe(cid:14)cient of ( (cid:0) )
cm
s
at
+
(cid:0)
(cid:0)
K, which is about one order of magnitude larger than found in previous studies (Millar
et al. ; Talbi et al. ). The original motivation for slow HCS
dissociative
+
recombination came from the observed high HCS
/CS abundance ratios. Rowe et al.
+
( ) claim that they can still reproduce this high value with a rapid rate by assuming
that the branching ratio toward the H + CS channel represents only one third of the
total dissociative recombination process. Since, in addition to this assumption, the
in(cid:13)uence of the vibrational excitation of the HCS
ions in their experiment is unclear,
+
the original determination of Millar et al. is kept:
k
=
(T = )
cm
s
+
HCS
(cid:0)
(cid:0) :
(cid:0)
For the deuterated species (DCO
, H
D
, N
D
and H
DO
) the same dissociative
+
+
+
+
recombination rate coe(cid:14)cients as for the hydrogen-based ones have been adopted; the
main formation and destruction rate coe(cid:14)cients are those given by the UMIST group
(Millar et al. ). In particular, we followed the study of Millar et al. ( ) for the
temperature-dependence of the important exchange reaction, H
+ HD ! H
D
+ H
:
+
+
k
= :
(: (cid:0) T = ) cm
s
f
(cid:0)
(cid:0)
For the reverse reaction, Pagani et al. ( ) have shown that the rate depends on the
ortho-H
D
/para-H
D
ratio. Since it is beyond the scope of this study to calculate
+
+
this ratio in our model, we simply adopt: k
= k
e
.
r
f
(cid:0)=T
The best (cid:12)tting models for the two clouds are described in the next section. The
model parameters are summarized in Tables and .
. NGC IRS
The principal characteristics of the NGC observations are the high abundance
of N
H
, and the high HCS
/CS ratio equal to :
. As noted by van Dishoeck
+
+
(cid:0)
et al. ( ), the high N
H
abundance implies that virtually all nitrogen must be in
+
the gas phase in the form of N
. Also, the CO/H
ratio has to be close to the lower
limit derived by Lacy et al. ( ), i.e., CO/H
= :
, because the dominant
(cid:0)
destruction route for N
H
is by reaction with CO, and the main formation route
+
is by: N
+ H
! N
H
+ H
, and H
is also mainly destroyed by CO. H
, in
+
+
+
+
turn, is formed through cosmic ray ionization of H
, and its abundance is directly
proportional to (cid:16)
(Lepp et al. ; Black et al. ; Phillips et al. ).
H
As can be seen from Tables and , the best (cid:12)t requires that all nitrogen is in gas
phase N
for a rather high value of (cid:16)
(cid:25) :
s
and low abundances of atomic
H
(cid:0)
(cid:0)
oxygen and carbon (Model ). In fact, if x(O) (cid:28) :
and x(C) (cid:28) :
, the
(cid:0)
(cid:0)
abundance of N
H
can be written as:
+
"
#
+
x(N
)
(cid:16)
H
x(N
H
)
[
][
]
(cid:0)
(cid:0)
:
:
(cid:24)
(cid:0)
x(H
O)
x
e
x
e
( +
+
)( +
)
:
:
:
(cid:0)
(cid:0)
(cid:0)
Thus, the N
H
abundance can also be reproduced by an even higher value of (cid:16)
, but
H
+
with the fraction of nitrogen in N
decreased. The dependence of the ion abundances
on (cid:16)
is illustrated in Fig. a. The inferred values for (cid:16)
are larger than those normally
H
H
assumed for dense clouds, (cid:16)
(cid:25) ( (cid:0) )
s
, but could be due to e.g., the presence
H
(cid:0)
(cid:0)
of X-rays from the young stellar ob ject, which have a similar e(cid:11)ect on the ionization
(Casanova et al. , Farquhar et al. ). If the abundances of O and C are close
to the upper limits derived from observations and other estimates (Little et al. ,
; Pagani & Encrenaz ) (cid:16)
has to be very high: an extreme case is given by
H
Model , in which we have assumed x(O) = x(CO) and x(C) = . x(CO).
The main problem comes from reproducing the high HCS
abundance.
It can be
+
shown that its abundance can be expressed as:
"
#
+
+
+
+
x(HCS
)
x(H
O
) + :x(HCO
) + :x(H
)
:
x
+ :
x(O)
e
(cid:0)
(cid:0)
(cid:24) :
Figure : Ratios of calculated ion abundances to observed ones as functions of the
three dominant parameters: (cid:16)
, x(H
O) and x(M
): a in the case of NGC (Model
H
+
), b in the case of W IRS (Model ).
Table : Model inputs for NGC
Parameter Model Model
(cid:0)
(cid:0)
(cid:0)
(cid:16)
(s
)
:
:
H
+
(cid:0)
(cid:0)
x(M
)
(cid:20)
(cid:20)
x(N
)
:
:
(cid:0)
(cid:0)
x(H
O)
(cid:0)
(cid:0)
x(O)
(cid:20) :
:
(cid:0)
(cid:0)
x(C)
(cid:20) :
:
(cid:0)
(cid:0)
x(D)
:
:
(cid:0)
(cid:0)
x(HD)
:
(cid:0)
(cid:0)
so that x(HCS
) increases when the electron fraction decreases: it is one of the rea-
+
sons why the metal ion abundance has to be low in this case. Moreover, keeping in
mind that the abundance of H
is now (cid:12)xed { because (cid:16)
is (cid:12)xed from the N
H
H
+
+
observations { , x(HCS
) depends essentially on the abundances of HCO
and H
O
,
+
+
+
where x(HCO
) has been constrained from observations. As can be seen in Fig. a,
+
the dependence of the HCS
abundance on (cid:16)
is not large, because in the model H
O
H
+
+
is the dominant ion leading to the formation of HCS
.
+
The formation of HCO
invokes CO, H
and N
H
. Its destruction proceeds mainly
+
+
+
by dissociative recombination and by reaction with H
O and atomic deuterium, which
results in DCO
. H
O
is formed by reactions of H
O and O with molecular ions
+
+
(essentially HCO
and H
) and is destroyed by dissociative recombination.
+
+
At this point, the important remaining parameters are thus the abundances of H
O,
D and HD which govern the abundances of HCO
, H
O
and DCO
, the latter
+
+
+
via HCO
and H
D
. Fig. a shows the model HCO
and DCO
abundances as
+
+
+
+
functions of the H
O abundance. In order to reproduce the observed values a rather
high abundance of H
O of at least
is needed. Also, it is inferred that only to
(cid:0)
% of deuterium is in atomic form. The high value of x(H
O) seems in contradiction
with the work of Wannier et al. ( ). However, their results refer to a much larger
beam of (cid:24) :
than the region sampled by our observations. The presence of water
masers near IRS suggests that locally the H
O abundance may be signi(cid:12)cantly higher
(Henning et al. ).
Finally, as can be seen in Table , good agreement between the observed and calculated
abundances is achieved for N
H
, HCO
, DCO
, N
D
and HCS
.
+
+
+
+
+
Table : Model (cid:12)ts for NGC
x(model)/x(obs) Model Model
HCO
.
.
+
DCO
.
.
+
+
N
H
.
.
+
N
D
.
.
+
HCS
.
.
predictions
x
e
:
:
(cid:0)
(cid:0)
+
(cid:0)
(cid:0)
x(H
)
:
:
x(H
D
)
:
:
+
(cid:0)
(cid:0)
x(H
O
)
:
:
+
(cid:0)
(cid:0)
x(H
DO
)
:
:
+
(cid:0)
(cid:0)
x(NH
)
:
:
(cid:0)
(cid:0)
x(NH
D)
:
:
(cid:0)
(cid:0)
With these assumptions, the electron fraction is x
= (: (cid:0) :)
, the dominant
e
(cid:0)
ion being H
O
. Searches for this ion in this ob ject would therefore be very valuable.
+
A strict lower limit is provided by the sum of all observed ions: x
(cid:21) :
. The
e
(cid:0)
maximum value of x(M
), for which the ratio of the calculated to observed abundances
+
lies between . and , is
in the case of Model (Fig. a) and
in the case
(cid:0)
(cid:0)
of Model . The predicted abundance of H
is below the upper limit derived by Black
+
et al. ( ) and is in good agreement with the values estimated by van Dishoeck et
al. ( ) from H
D
, in their high density case, especially when it is assumed that
+
all nitrogen is in gas phase N
. Model predicts H
D
=H
and H
DO
=H
O
ratios
+
+
+
+
comparable to the observed ratios DCO
=HCO
(cid:24)
and N
D
=N
H
(cid:24)
+
+
(cid:0)
+
+
(cid:0)
(Table ). In the case of Model , the calculated N
D
=N
H
is higher due to the
+
+
higher abundance of D, and the H
D
=H
is lower due to the lower abundance of HD.
+
+
The model also predicts the abundance of NH
, which is close to the value of
(cid:0)
determined observationally by Schreyer et al. ( ), and that of NH
D. It should
be noted, however, that the current model does not include grain surface reactions,
which may be signi(cid:12)cant for the production of NH
and its deuteration.
. W IRS
The excitation analysis of x has shown that the density in W IRS is slightly
lower than that in NGC , but the temperature higher. Moreover the observed
+
(cid:0)
+
HCS
=CS ratio of
and the N
H
abundance are lower, suggesting a higher
electron fraction and/or a lower cosmic ray ionization rate. The adopted CO=H
ratio is also higher than toward NGC . The model analysis for W IRS cannot
proceed in the same way as for NGC , because no good determination of the N
H
+
abundance is available. Although N
H
{ is most likely detected by van Dishoeck
+
et al. ( ), the line falls at the edge of the spectrum and does not lead to a reliable
column density. Moreover, the { line was not detected. On the other hand, H
O
+
data are available for this source, in contrast with NGC , which together with
+
HCS
proved to be very useful in the analysis.
The abundance of HCS
can be written as:
+
"
#
+
+
+
+
x(HCS
)
x(H
O
) + :x(HCO
) + : x(H
)
:
x
+ :
x(O)
e
(cid:0)
(cid:0)
(cid:24) :
Using the observational constraints on HCS
, H
O
and HCO
, we can derive a
+
+
+
relation between x(H
), x(O) and x
from this expression.
e
+
The relation between x
, x(H
) and x(M
) is given by:
e
+
+
x
(cid:24) x(H
O
) + x(HCO
) + x(H
) + x(M
);
e
+
+
+
+
or
x
(cid:24)
+ x(H
) + x(M
):
e
(cid:0)
+
+
Together, these two formulae lead to an expression of x(H
) as a function of x(O )
+
and x(M
).
+
Further relations can be derived from
"
#
x(H
O
)
+
:
+
+
(cid:24)
[x(H
O)(x(HCO
) + :x(H
))
:
(cid:0)
x
e
+ :x(O)x(H
)]
+
which leads to x(H
O) as a function of x(O ) and x(M
).
In the same way an
+
expression for x(D) as a function of x(O ) and x(M
) is derived from:
+
"
#
x(HCO
)
x(N
H
) + : x(H
)
+
+
+
:
(cid:0)
x(H
O)
x(D)
x
e
:
:
:
(cid:0)
(cid:0)
(cid:0)
+
+
(cid:24)
Assuming that x(HD) = [D]/[H
](cid:0)x(D), with [D]/[H
] = :
, another expression
(cid:0)
of x(D) can be obtained from:
"
#
x(DCO
)
+
:
(cid:0)
x
e
x(H
O)
+
:
:
(cid:0)
(cid:0)
(cid:24)
[x(D)(x(HCO
) +
:
+
+
+
:x(H
)) + x(HD)x(H
)]
The two expressions of x(D) imply that:
x(O)
x(M
)
+
:
:
(cid:0)
(cid:0)
+
(cid:24) :
The maximum value of x(M
) is then around :
, corresponding to a low atomic
+
(cid:0)
oxygen abundance (Model ). The values of these two parameters being (cid:12)xed, we
can obtain the abundances of H
O, D and HD needed to reproduce the observations,
as presented in Table : the abundance of H
O adopted in the modeling is in good
agreement with the value derived by Phillips et al. ( ) from data of Cernicharo et
al. ( ), and only . to .% of deuterium is in atomic form. We also derive the
abundance of H
from x(O ) and x(M
), and then (cid:16)
since:
H
+
+
+
x(H
) (cid:24)
:
(cid:16)
H
x(O)
x(C)
x
e
+
+
+
(cid:0)
(cid:0)
(cid:0)
:
:
:
In the case of Model the values of x(O) and x(M
) have been chosen in order to get
+
a value of (cid:16)
close to the lower limit generally assumed in dark clouds.
H
As for NGC , the sensitivity of the calculated ion abundances to the main pa-
rameters, x(M
), x(H
O) and (cid:16)
are shown in Fig. b, in the case of Model . The
H
+
fraction of N
is chosen to give an abundance of N
H
close to the observed upper
+
limit: between and % of the nitrogen is then in gas phase N
, similar to the
values suggested by van Dishoeck et al. ( ).
With the values of the parameters (Table ), good agreement between the observed and
calculated abundances is achieved for HCO
, DCO
, N
H
, HCS
and H
O
(Table
+
+
+
+
+
Table : Model inputs for W IRS
Parameter Model Model
(cid:0)
(cid:0)
(cid:0)
(cid:16)
(s
)
:
H
+
(cid:0)
(cid:0)
x(M
)
x(N
)
:
(cid:0)
(cid:0)
x(H
O)
(cid:0)
(cid:0)
x(O)
:
(cid:20) :
(cid:0)
(cid:0)
x(C)
(cid:20) :
(cid:20) :
(cid:0)
(cid:0)
x(D)
:
(cid:0)
(cid:0)
x(HD)
:
:
(cid:0)
(cid:0)
Table : Model (cid:12)ts for W IRS
x(model)/x(obs) Model Model
HCO
.
.
+
DCO
.
.
+
+ (cid:3)
N
H
.
.
+ (cid:3)
(cid:0)
(cid:0)
N
D
:
:
+
HCS
.
.
+
H
O
.
.
+ (cid:3)
(cid:0)
(cid:0)
H
D
:
:
predictions
x
e
:
:
(cid:0)
(cid:0)
+
(cid:0)
(cid:0)
x(H
)
:
:
x(H
DO
)
:
:
+
(cid:0)
(cid:0)
x(NH
)
:
:
(cid:0)
(cid:0)
x(NH
D)
:
:
(cid:0)
(cid:0)
(cid:3)
Compared with the observed upper limit.
Table : Ratios of deuterated to hydrogenated ion abundances obtained in the models
Ratio
NGC
W IRS
Mod. Mod. Mod. Mod.
+
+
(cid:0)
(cid:0)
(cid:0)
(cid:0)
N
D
=N
H
:
:
:
:
+
+
(cid:0)
(cid:0)
(cid:0)
(cid:0)
H
D
=H
:
:
:
:
DCO
=HCO
:
:
:
:
+
+
(cid:0)
(cid:0)
(cid:0)
(cid:0)
H
DO
=H
O
:
:
:
:
+
+
(cid:0)
(cid:0)
(cid:0)
(cid:0)
NH
D=NH
:
:
:
:
(cid:0)
(cid:0)
(cid:0)
(cid:0)
). The calculated values of x(N
D
) and x(H
D
) are much lower than the observed
+
+
upper limits. The electron fraction is x
= ( : (cid:0) :)
and x(M
) = ( (cid:0) )
.
e
(cid:0)
+
(cid:0)
+
M
is the dominant ion, providing between and % of the electrons. The predicted
abundances of H
, NH
and NH
D are one or two orders of magnitude lower than
+
toward NGC , due to the higher values of x(CO) and x(O) and the lower values
of (cid:16)
and x(N
).
H
The predicted H
DO
=H
O
ratio is close to the observed ratio DCO
=HCO
(cid:24) :
,
+
+
+
+
(cid:0)
but is about one order of magnitude higher than the calculated H
D
=H
, N
D
=N
H
+
+
+
+
and NH
D=NH
ratios (Table ): these low values are mostly due to the low abun-
dance of atomic deuterium adopted in the models required to reproduce the observed
DCO
and HCO
abundances.
+
+
Conclusions and Discussion
We have observed submillimeter lines of several molecular ions and neutral species in
two dense clouds, NGC and W IRS, which allow accurate determination of the
physical conditions and abundances. Uncertainties remain due to the fact that possible
density, temperature and abundance variations have been ignored, so that the derived
n(H
), T
and abundances are averaged over the beam and along the line of sight.
kin
The absolute abundances are furthermore uncertain due to lingering uncertainties in
the CO/H
abundance ratio; the relative values should be more reliable, however.
In both clouds the lower limit on the electron fraction derived from the sum of the
abundances of the observed ions is x
>
at densities of ( (cid:0) )
cm
.
e
(cid:0)
(cid:0)
A more accurate determination of x
has been made, using a simple steady state
e
chemical model in which the most important formation and destruction routes of the
ions are taken into account, and in which as many species as possible are constrained by
observations. Although there are still uncertainties in the values of the reactions rates,
especially for dissociative recombination, we obtained agreement with the observations
to within a factor of two. The possibility exists that a chemical steady state has not
been reached in the observed sources, in which case comparison of observations with
models are more di(cid:14)cult since the initial conditions and the age of the cloud are
unknown. However, time-dependent models such as those of Millar et al. ( ) and
Shalabiea & Greenberg ( ) show little variation in the ma jor ion abundances and
the electron fraction with time.
More elaborated models using a full blown gas phase network of more than
reactions and including gas-grain interactions give comparable results: for a density
n(H
) =
cm
and elemental gas-phase abundances close to those derived for
(cid:0)
NGC and W IRS, Shalabiea & Greenberg ( ) obtain x
(cid:24) (: (cid:0) :)
e
(cid:0)
using a pure gas phase model or a dust/gas model (see their Model A). Similar results
are presented by Le Bourlot et al. ( ) with a pure gas-phase model, as well as by
Millar ( , private communication).
We did not consider in this work the interaction between electrons and grains. As
discussed e.g. by d'Hendecourt et al. ( ) this process leads in dense clouds to grains
with an average charge of (cid:0)e, due to the ine(cid:14)cient photoelectric e(cid:11)ect. Because the
grain abundance is typically
, this results in a decrease in the gas phase electron
(cid:0)
fraction by only a few percent at most. Moreover, this situation may be modi(cid:12)ed if
the internally produced ultraviolet photons due to the excitation of H
by cosmic rays
are taken into account, as suggested by preliminary results of Cervetto et al. ( ):
this ultraviolet radiation could be strong enough to give rise to a photoelectric e(cid:11)ect
su(cid:14)cient to drive the charge distribution of grains toward positive values.
The ionization balance in dense clouds may also be modi(cid:12)ed by the presence of large
molecules such as PAHs. In particular, the density of electrons may be reduced and
negative PAH
ions may contain most of the negative charge if the total PAH abun-
(cid:0)
dance and the thermal sticking coe(cid:14)cient of electrons on neutral PAH molecules are
high (Omont , Lepp & Dalgarno , Allamandola et al. ). This means that
the electron fractions derived in the preceding section are upper limits, but that the
degree of ionization could be higher, typically of the order of the total fraction of PAH
which is generally assumed to be x
(cid:24)
(cid:0)
, for PAH molecules containing
PAH
(cid:0)
(cid:0)
between and % of the total carbon and formed with about C atoms. The re-
duced electron fraction would imply an increase in the abundances of those ions which
are predominantly destroyed by dissociative recombination { H
O
, HCS
, HCO
and
+
+
+
DCO
in our case { , although the (slower) destruction by mutual neutralization with
+
(cid:0)
PAH
ions diminishes the e(cid:11)ect. Thus, the numerical results of Section could be
slightly modi(cid:12)ed if PAH molecules are included in the modeling. However, the large
uncertainties in the total PAH abundance in dense clouds and in the rates involving
these species prevent clear conclusions. In particular, Mendoza-G(cid:19)omez et al. ( )
argue that PAH molecules do not survive accretion onto grains inside dense clouds.
In summary for dense clouds with n(H
) (cid:24)
cm
, the inferred electron fraction lies
(cid:0)
in the range
< x
<
. Ionized metals such as Mg
or Fe
could provide
e
(cid:0)
(cid:0)
+
+
up to % of the electrons. These values of x
are consistent with those commonly
e
used in the literature and represented by the simple formula x
(cid:24)
n(H
)
e
(cid:0)
(cid:0)=
(McKee ). Comparison of our results to those obtained in lower density, colder
clouds (Schilke et al. ), does not show a clear dependence of x
on density between
e
(cid:0)
and
cm
, however. In fact, the analysis of the preceeding section shows that
other parameters which can (cid:13)uctuate from cloud to cloud also play a r^ole.
Acknowledgments
The authors are grateful to Remo Tilanus and Fred Baas for their support during
the JCMT observing runs, and to Katharina Schreyer for sharing her JCMT data on
NGC with us. They thank Tom Millar and Osama Shalabiea for many helpful
discussions concerning chemical models. One of us (CB) would like to acknowledge
E.F. van Dishoeck and her group for their hospitality in Leiden. This work was spon-
sored by a PIONIER grant from the Netherlands Organization for Scienti(cid:12)c Research
(NWO).
References
Adams, N.G., Smith, D. , ApJ , L.
Allamandola, L.J., Tielens, A.G.G.M., Barker, J.R., , A&AS , .
Allen, D.A. , ApJ , L.
Amano, T. , J. Chem. Phys. , .
Black, J. H., van Dishoeck, E. F., Willner, S. P., Woods, R. C. , ApJ , .
Canosa, A., Gomet, J. C., Rowe, B. R., Mitchell, J. B. A., Que(cid:11)elec, J. L. , J. Chem. Phys.,
, .
Casanova, S., Montmerle, T., Feigelson, E. D., Andr(cid:19)e, P., , ApJ , .
Cernicharo, J., Thum, C., Hein, H. et al., , A&A , L.
Cervetto, C., Cecchi-Pestellini, C., Aiello, S., Barsella, B., , in: Proceedings of the (cid:12)rst
Franco-British Conference on the Physics and Chemistry of the Interstellar Medium, CCP,
in press.
Dickel, H.R., , ApJ , .
Dickel, H.R., Dickel, J.R., Wilson, W.J., Werner, M.W., , ApJ , .
Farquhar, P.R.A., Millar, T.J., Herbst, E., , MNRAS , .
Gu(cid:19)elin, M., Langer, W.D., Snell, R.L., Wootten, H.A. , ApJ , L.
Gu(cid:19)elin, M., Langer, W.D., Wilson, R.W. , A&A , .
Hasegawa, T.I., Mitchell, G.F., Matthews, H.E., Tacconi, L., , ApJ , .
Hayashi, M., Kobayashi, H., Hasegawa, T., , ApJ , .
Helmich, F.P., Van Dishoeck, E.F., A&AS , in preparation.
Helmich, F.P., Jansen, D.J., De Graauw, Th., Groesbeck, T.D., van Dishoeck, E.F. , A&A
, .
d'Hendecourt, L., Allamandola, L.J., Greenberg, J.M., , A&A , .
Henning, T., Cesaroni, R., Walmsley, C.M., Pfau, W., , A&AS , .
Jansen, D.J., , PhD thesis, University of Leiden.
Jansen, D.J., Van Dishoeck, E.F., Black, J.H., , A&A , .
Kr(cid:127)ugel, E., G(cid:127)usten, R., Schulz, A., Thum, C. , A&A , .
Lacy, J.H., Knacke, R., Geballe, T.R., Tokunaga, A.T. , ApJ , L .
Ladd, E.F., Deane, J.R., Sanders, D.B., Wynn-Williams, C.G., , ApJ , .
Langer W.D., , in: Protostars and Planets II, eds. D.C. Black and M.S. Matthews
(University of Arizona Press, Tucson), p. .
Larsson, M., Danared, H., Mowat, J.R., et al. , Phys. Rev. Lett. , .
Le Bourlot, J., Pineau des For^ets, G., Roue(cid:11), E., Schilke, P. , ApJ , L.
Lepp, S., Dalgarno, A. , ApJ , .
Lepp, S., Dalgarno, A., Sternberg, A. , ApJ , .
Little, L.T., Gibb, A.G., Heaton, B.D., Ellison, B.N., Claude, S.M.X. , MNRAS , .
Little, L.T., Gibb, A.G., Kelly, M.L. , in: Proceedings of the (cid:12)rst Franco-British Conference
on the Physics and Chemistry of the Interstellar Medium, CCP, in press.
Mangum, J.G., Wootten, A. , ApJS , .
McKee, C.F. , ApJ , .
Mendoza-G(cid:19)omez, C.X., de Groot, M.S., Greenberg, J.M., , A&A , .
Millar, T.J., Bennett, A., Herbst, E. , ApJ , .
Millar, T.J., Adams, N.G., Smith, D., Clary, D.C. , MNRAS , .
Millar, T.J., Rawlings, J.M.C., Bennett, A., Brown, P.D., Charnley, S.B. , A&AS , .
Mitchell, G.F., Maillard, J.-P. , ApJ , L .
Omont, A. , A&A , .
Pagani, L., Encrenaz, P. , in : The structure and content of molecular clouds, eds.
T.L. Wilson and K.J. Johnston (Lecture Notes in Physics, Springer Verlag), p. .
Pagani, L., Salez, M., Wannier, P. G. , A&A , .
Phillips, T.G., Van Dishoeck, E.F., Keene, J. , ApJ , .
Rowe, B., Canosa, A., Sims, I. , J. Chem. Soc. Far. Tr. , .
Schilke, P., Adams, N.G., Ferguson, E. E. , A&A , .
Schreyer, K., Helmich, F.P., van Dishoeck, E.F., Henning, T. , A&A in preparation.
Schwartz P.R., Thronson H.A., Odenwald, S.F., et al., , ApJ ,
Shalabiea, O.M., Greenberg, J.M. , A&A , .
Smith, D.,
Spanel, P., , Chem. Phys. Letters , .
(cid:21)
Talbi, D., Hickman, A.P., Pauzat, F., Ellinger, Y., Berthier, G. , ApJ , .
Tieftrunk, A.R., Wilson, T.L., Steppe, H., et al., , A&A, in press.
Tielens, A.G.G.M., Tokunaga, A.T., Geballe, T.R., Baas, F. , ApJ , .
van Dishoeck, E.F., Phillips, T.G., Keene, J., Blake, G.A. , A&A , L.
Wannier, P. G., Pagani, L., Kuiper,T. B. H., et al., , ApJ , .
Watson, W.D. , in CNO isotopes in Astrophysics, ed. J. Audouze (Reidel, Dordrecht), p. .
Wootten, A., Loren, R.B., Snell, R.L., , ApJ , .
Wootten, A., Snell, R., Glassgold, A.E., , ApJ , .
|
astro-ph/0701505 | 1 | 0701 | 2007-01-17T16:47:39 | HI study of extremely metal-deficient dwarf galaxies. I. The Nancay Radio Telescope observations of twenty-two objects | [
"astro-ph"
] | The goal of this study is to measure parameters of the integrated HI emission for twenty-two dwarf galaxies with oxygen abundance 12+log(O/H) in the range of 7.42 to 7.65, which are representatives of the eXtremely Metal-Deficient (XMD) galaxy group. Some of them are expected to be similar to the well-known candidates for local young galaxies, IZw18 and SBS 0335-052 that have most of their baryon mass in the form of neutral gas. Therefore, the HI 21-cm line observations are crucial to understanding their group and individual properties. The Nancay Radio Telescope (NRT) with the upgraded focal receiver was used for observations of the 21-cm HI-line. This permitted the detection of the faintest sources with rms of ~1 mJy per 10.5 km/s resolution element. For eighteen detected galaxies we present the parameters of their integrated HI line emission and describe the data on individual objects in more detail. For four undetected XMD galaxies, we give upper limits on their M$(HI). For 70% of the twenty studied non low surface brightness XMD galaxies, we find evidence (both from HI and optical data) for their interaction with neighboring objects. In the brief discussion of the group HI properties of the observed subsample (the total O/H range is of 0.23 dex, or a factor of 1.7), we underline the broad distributions of the HI mass (range is of 2 orders of magnitude), of the ratio M(HI)/L_B (of 1 order of magnitude), and of the blue luminosity (range is of 2 orders of magnitude). We also obtained HI parameters of six galaxies that do not belong to the XMD sample. These data increase the number of XMD galaxies with known integrated HI parameters (or upper limits) by a factor of two. This allows us to address statistical properties of this group, which will be presented in a forthcoming paper. (Abridged). | astro-ph | astro-ph | Astronomy&Astrophysicsmanuscript no. 6137
May 11, 2018
c(cid:13) ESO 2018
7
0
0
2
n
a
J
7
1
1
v
5
0
5
1
0
7
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
H i study of extremely metal-deficient dwarf galaxies.
I. The Nanc¸ay Radio Telescope observations of twenty-two objects
S.A. Pustilnik1,2 and J.-M. Martin2,3
1 Special Astrophysical Observatory RAS, Nizhnij Arkhyz, Karachai-Circassia, 369167 Russia
2 Observatoire de Paris, GEPI, 5, place J. Janssen, 92195 Meudon Cedex, France
3 Observatoire de Paris, USN, route de Souesmes, 18330 Nanc¸ay, France
Received 27.07.2006; Accepted
2006
ABSTRACT
Aims. The goal of this study is to measure parameters of the integrated H i emission for twenty-two dwarf galaxies with oxygen
abundance 12+log(O/H) in the range of 7.42 to 7.65, which are representatives of the eXtremely Metal-Deficient (XMD) galaxy
group. Some of them are expected to be similar to the well-known candidates for local young galaxies, I Zw 18 and SBS 0335−052
that have most of their baryon mass in the form of neutral gas. Therefore, the H i 21-cm line observations are crucial to understanding
their group and individual properties.
Methods. The Nanc¸ay Radio Telescope (NRT) with the upgraded focal receiver was used for observations of the 21-cm H i-line. This
permitted the detection of the faintest sources with rms of ∼1 mJy per 10.5 km s−1 resolution element.
Results. For eighteen detected galaxies we present the parameters of their integrated H i line emission and describe the data on
individual objects in more detail. For four undetected XMD galaxies, we give upper limits on their M(H i). For 70% of the twenty
studied non-LSB (low surface brightness) XMD galaxies, we find evidence (both from H i and optical data) for their interaction with
neighboring objects. In the brief discussion of the group H i properties of the observed subsample (the total O/H range is of 0.23 dex,
or a factor of 1.7), we underline the broad distributions of the H i mass (range is of 2 orders of magnitude), of the ratio M(HI)/LB (of
1 order of magnitude), and of the blue luminosity (range is of 2 orders of magnitude). We also obtained H i parameters of six galaxies
that do not belong to the XMD sample. As a by-product, we have detected an H i-object with Vhel =188 km s−1, which is probably a
part of a new high-velocity H i cloud (HVC 234.3-16.8+208) near our Galaxy, in the direction of AM 0624 -- 261.
Conclusions. These data increase the number of XMD galaxies with known integrated H i parameters (or upper limits) by a factor of
two. This allows us to address statistical properties of this group, which will be presented in a forthcoming paper.
Key words. galaxies: dwarf -- galaxies: H i -- galaxies: star formation -- galaxies: abundances -- radio lines: galaxies
1. Introduction
The oxygen abundances in the great majority of actively star-
forming low-mass galaxies (BCGs - blue compact and H ii galax-
1 (e.g., Terlevich
ies) correspond to metallicities of (0.1 -- 0.5) Z⊙
et al. 1991; Izotov et al. 1992; Ugryumov et al. 2003; Kniazev
et al. 2004; Salzer et al. 2005; Pustilnik et al. 2006c, 2006b,
among others). Some BCGs, due to their low metallicities, were
originally considered as probable young galaxy populations.
However, it has been shown that the great majority of BCGs con-
tain a measurable number of old stars and thus, are old systems
(e.g., Papaderos et al. 1996; Doublier et al. 1999; Cairos et al.
2003). Only a few BCGs with the lowest known metallicities
(12+log(O/H) of 7.12 -- 7.54, or Z = Z⊙/34 -- Z⊙/13) of those with
well studied deep photometry show no evidence for stellar pop-
ulations older than one Gyr. They include I Zw 18 (Papaderos et
al. 2002; Izotov & Thuan 2004b; Ostlin & Mouhcine 2005); SBS
0335 -- 052 E and W (Papaderos et al. 1998; Pustilnik et al. 2004b;
Izotov & Thuan 2005), Tol 65 (Papaderos et al. 1999), and prob-
ably DDO 68 (Pustilnik et al. 2005a). These BCGs show an ad-
ditional blue color excess and a luminosity excess, which are
not accounted for by the color-metallicity (Guseva et al. 2003a)
Send offprint requests to: S. Pustilnik e-mail: [email protected]
1 Updated Z⊙ corresponds to 12+log(O/H) = 8.66, according to the
recent data by Asplund et al. (2004).
and the luminosity-metallicity relations for other BCGs (e.g.,
Pustilnik et al. 2003a; Kniazev et al. 2003). In addition, they
have very high gas mass-fractions µg = Mg/(Mg + Mstar): up to
0.95 -- 0.99 (van Zee et al. 1998; Papaderos et al. 2002; Pustilnik
et al. 2004b). These properties can be understood if these BCGs
are young objects, with non-cosmological ages.
The number of known eXtremely Metal-Deficient (XMD,
12+log(O/H) ≤ 7.65) galaxies has grown to ∼50 during the last
decade thanks to dedicated searches, conducted mainly with ob-
jective prism surveys of Emission Line Galaxies (e.g., Izotov
et al. 1992; Kniazev et al. 1998; Melbourne & Salzer 2004;
Ugryumov et al. 2003) and with spectroscopy of large num-
ber of faint galaxies from the Sloan Digital Sky Survey (SDSS,
presented, e.g., by Kniazev et al. 2003, 2004 and Izotov et al.
2006). Such 'unevolved' objects comprise only a small fraction
of known galaxies in the nearby Universe. Most of the known
XMD galaxies are of the BCG type.
Different evolutionary scenarios can lead to very low metal-
licity of galaxy ISM. They include: a) the loss of metals due to
superwinds related to powerful star formation (SF) events (e.g.,
Mac Low & Ferrara 1999); b) primordial gas inflow or inter-
galactic cloud accretion (e.g., Matteucci & Chiosi 1983; Combes
2005); c) the very slow astration and related metal production,
characteristic of conditions in some LSB galaxies (Low Surface
Brightness -- e.g., Legrand et al. 2000), and finally, d) truly young
2
S.A. Pustilnik & J.-M. Martin: H i in 22 extremely metal-deficient galaxies
galaxies in which the first SF episode took place less than ∼1
Gyr ago. In particular, for a few LSB dwarfs with sufficiently
bright H ii regions the measurements gave very low values of
12+log(O/H) of 7.3 to 7.5 (Ronnback & Bergvall 1995; van
Zee et al. 1996; van Zee 2000). Several dwarf irregular galax-
ies (DIG) from the Local Volume were claimed as XMD objects
long ago (e.g., Skillman et al. 1989). The recent data confirmed
the XMD classification for GR 8 and Leo A (van Zee et al. 2006),
DDO 53 (Pustilnik et al. 2003c) and Sextans A and B (Kniazev
et al. 2005). All nearby XMD LSB dwarfs and DIGs studied with
the Color-Magnitude Diagram (CMD) method appeared to have
a substantial fraction of stars belonging to an old stellar pop-
ulation. Thus, they are old systems. In contrast to these "non-
active" dwarfs, the CMD data for the second most metal-poor
BCG I Zw 18 imply that it is probably a young galaxy (Izotov
& Thuan 2004b; Ostlin & Mouhcine 2005; this result has been
questioned, however, by Momany et al. 2005). Therefore, low
ISM metallicity is a necessary but not sufficient property to as-
sign an object as a candidate for a young galaxy.
The high-z dwarf galaxies are near still or below the limit
of detectability with the most powerful instruments (e.g., Pello
et al. 2005; Yan et al. 2005; Bouwens & Illingworth 2006).
Local XMD galaxy properties are similar to those of young
dwarf galaxy populations at high redshift, so their detailed study
is an effective way to probe early galaxy evolution. Since the
cold/warm gas should be the main component of baryonic mat-
ter in unevolved galaxies, the 21-cm H i line is one of the primary
means to study their properties.
Our multi-wavelength study of the XMD galaxy sample in-
cludes long-slit spectroscopy, optical/NIR imaging, photome-
try, and H i 21-cm spectral observations. We present the mea-
surements of integrated H i parameters for 22 galaxies with
12+log(O/H) = 7.42 -- 7.65 here. These data are used for H i
mapping of some XMD galaxies at the Giant Meterwave Radio
Telescope (GMRT, India). In Sect. 2 we describe the sample.
Observations and data reduction are described in Sect. 3. The re-
sults are presented in Sect. 4. In Sect. 5 we discuss the results
and draw preliminary conclusions. The analysis of ∼40 XMD
galaxies with available H i data is deferred to a forthcoming pa-
per.
2. The sample
The list of XMD galaxies was compiled by Kunth & Ostlin
(2000, hereafter KO). A significant number of new XMD galax-
ies was added in the recent papers cited below. Metallicities of
the galaxies from the KO list, based on old, lower accuracy deter-
minations, were revisited, and four of them were excluded from
the XMD group (Kniazev et al. 2001; Pustilnik et al. 2002a).
Three of the 22 galaxies are well known. AM 0624 -- 261 is
classified as a blue LSB galaxy (Ronnback & Bergvall 1994,
1995). Two others - Tol 21 and Tol 65 are BCGs from the Tololo
survey (Masegosa et al. 1994; Thuan & Izotov 1997; Papaderos
et al. 1999; Fricke et al. 2001, and references therein). For
these three, only AM 0624 -- 261 has published H i data (S/N ∼4,
Gallagher et al. 1995). The remaining 19 galaxies were found
recently as XMD objects. Seven of them are from the HSS sam-
ple (Kniazev et al. 1998; Pustilnik et al. 1999, 2006b), four are
from the HSS-LM sample (Ugryumov et al. 2003; Pustilnik et
al. 2006a), and one is from the SBS sample (Izotov & Thuan
1999). Six more galaxies were found in the SDSS (Kniazev et
al. 2003). One galaxy is from the KISS sample (Lee et al. 2004).
No data on H i parameters of these galaxies were known in
the period of our NRT observations, except for HS 0122+0743
and SDSS J1121+0324. The former was suggested as a proba-
ble merger, so we tried to find some additional indications in its
H i profile. The latter was serendipitously detected in H i at the
VLA by Hibbard & Sansom (2003). The main parameters of all
observed galaxies with references to the sources are shown in
Table 1. Some B-magnitudes were derived from the photometry
presented in the Sloan Digital Sky Survey (SDSS) Data Release
4 by Adelman-McCarthy et al. (2006), with the use of the trans-
formation equation from g and r magnitudes as given by Smith
et al. (2002).
(North-South) at declination δ=0◦
3. Observations and reduction
The H i-observations with the Nanc¸ay2 radio telescope (NRT)
with a collecting area of 200×34.5 m are character-
ized by a half-power beam width (HPBW) of 3.7′ (East-
West) × 22′
(see also
http://www.obs-nancay.fr/nrt). The data were acquired
during the years 2002 -- 2004. We used the new antenna/receiver
system F.O.R.T. (e.g., Granet et al. 1997; Martin et al. 2002)
with improved overall sensitivity. The system temperature was
∼35 K for both the horizontal and vertical linear polarizations
of a dual-polarization receiver. The gain of the telescope was
1.5 K Jy−1 at declination δ=0◦. The new 8192-channel correla-
tor was used covering a total bandwidth of 12.5 MHz. The total
velocity range covered was about 2700 km s−1, with the channel
spacing of 1.3 km s−1 before smoothing. The effective resolution
after averaging of four adjacent channels and Hanning smooth-
ing was ≈10.5. km s−1. The observations consisted of separate
cycles of 'ON' and 'OFF' integrations, each of 40 seconds in
duration. 'OFF' integrations were acquired at the target decli-
nation, usually with the East R.A. offset of ∼15′ ×cos(δ). For a
few cases, when we clearly suspected confusion from a galaxy
near an 'OFF' position, we selected a different 'OFF' position to
avoid possible confusion.
For the flux calibration we used a noise diode. Its power
was regularly monitored throughout the observations by point-
ing at known continuum and line sources. The comparisons of
our measured fluxes with independent measurements of the same
objects by other telescopes indicates the consistency of the flux
scale to within 10%.
With an rms noise of ∼1 to 7 mJy per resolution element af-
ter smoothing (10.5 km s−1), we obtained a S/N ratio of 30 -- 40
for the HI line peak flux densities Fpeak of the brightest detected
objects, while for the faintest ones we had detections with a S/N
ratio of ∼3 -- 4. Total integration times per galaxy ('ON'+'OFF')
varied between 1 and 10 hours. For four of the twenty-two ob-
served XMD galaxies, we obtained only upper limits on their
Fpeak, and estimated limits on their integrated H i flux.
The data was reduced using the NRT standard programs
NAPS and SIR, written by the telescope staff (see description
on http://www.obs-nancay.fr/nrt/support). Horizontal and verti-
cal polarization spectra were calibrated and processed indepen-
dently and then averaged together. The error estimates were cal-
culated following Schneider et al. (1986). The baselines were
generally well-fit by a third order or lower polynomial and were
subtracted out. For a few nearby extended galaxies with an angu-
lar size comparable to the NRT horizontal HPBW, a correction
for resolution of the observed H i flux has been done. Having no
2 The Nanc¸ay Radioastronomy Station is part of the Observatoire
de Paris and is operated by the Minist`ere de l'Education Nationale
and Institut des Sciences de l'Univers of the Centre National de la
Recherche Scientifique.
S.A. Pustilnik & J.-M. Martin: H i in 22 extremely metal-deficient galaxies
3
Table 1. Parameters of the XMD sample and additional (below the line) galaxies
IAU style
name
(1)
Other
name or
prefix
(2)
HS
0017+1055
HS
0122+0743
SDSS
J0133+1342
SDSS
J0205 -- 0949
AM
0624 -- 261
HS
0846+3522
KISSB23
0937+2949
HS
0940+4025
HS
1013+3809
HS
1033+4757
HS
1059+3934
SDSS
J1105+6022
SDSS
J1121+0324
SDSS
J1201+0211
SBS
1159+545
SDSS
J1215+5223
Tol
1214 -- 277
Tol
1223 -- 359
HS
1442+4250
HS
1704+4332
HS
2134+0400
HS
2236+1344
1059+395
KUG
1102+6038 MCG
J1106+6015 MCG
J1202+0215
1342+4210
2236+136
HS
KUG
Type
Coord. (2000.0)
R.A.
h m s
(4)
00 20 21.35
01 25 34.18
01 33 52.56
02 05 49.13
06 26 16.99
08 49 40.26
09 40 12.85
09 43 36.34
10 16 24.59
10 36 25.11
11 02 09.81
11 05 53.62
11 21 52.80
12 01 22.32
12 02 02.36
12 15 46.56
12 17 17.09
12 25 46.96
14 44 11.35
17 05 45.39
21 36 58.95
22 38 31.15
11 02 00.14
11 05 34.05
11 06 47.99
12 02 20.94
13 44 59.49
22 39 21.95
Dec.
◦
′
′′
(5)
+11 12 20.7
+07 59 22.2
+13 42 09.4
-- 09 49 18.1
-- 26 15 56.3
+35 11 39.2
+29 35 28.8
+40 11 18.1
+37 54 44.5
+47 41 52.3
+39 18 45.3
+60 22 28.5
+03 24 21.2
+02 11 08.5
+54 15 50.1
+52 23 13.9
-- 28 02 32.6
-- 36 14 00.6
+42 37 34.7
+43 28 48.9
+04 14 04.1
+14 00 28.6
+39 19 06.4
+60 22 06.6
+60 15 46.8
+02 15 58.5
+41 55 05.4
+13 52 55.8
(3)
BCG
IP
BCG
SmPec
LSB
BCG
dIrr
BCG
BCG
BCG
BCG
BCG
BCG
BCG
BCG
LSB
BCG
BCG
BCG
BCG
BCG
BCG
BCG
LSB
dIrr
LSB/dI
BCG
GPair
Angul.
sizea
a25 × b25
(6)
...
59×30
...
96×48
49×26
...
36×12
...
...
...
...
30×18
43×12
...
...
72×40
...
...
60×12
...
...
...
...
57×34
51×17
...
...
...
Vopt
km s−1
(7)
5630±30
2926±30
2599±20
1908±18
495±17
2177±27
543±11
5310±40
1185±20
1541±21
2995±70
1340±20
1223±38
986±20
3537±40
162±15
7795±50
2811±30
660±15
2076±12
5070±50
6160±20
3067±60
--
1273± 3
--
3780±50
5184±15
Btot‡
mag
(8)
18.21
15.51
18.15
15.55
16.01
18.15
16.33
18.55
16.05
17.75
17.25
16.65
16.95
17.94
18.75
15.45
18.46
17.67
15.68
18.48
19.31
17.91
16.63
16.5:5
16.35
18.85
18.35
15.45
∗
M0
B
mag
(9)
-- 16.7
-- 17.7
-- 14.4
-- 16.6
-- 12.5
-- 14.6
-- 13.6
-- 16.0
-- 15.4
-- 14.3
-- 16.1
-- 15.1
-- 14.6
-- 13.5
-- 14.9
-- 12.9
-- 16.9
-- 15.6
-- 14.9
-- 14.5
-- 15.1
-- 16.9
-- 16.8
-- 15.1
-- 15.3
--
-- 15.5
-- 19.0
O/H† Alternative
name
(11)
UGC 993
KUG 0203-100
ESO 489-056
KUG 0937+298
KUG 1013+381
SBS 1102+606
CGCG 269 -- 049
Tol 21
Tol 65
UGC 9497
MCG+10-16-056
MCG+10-16-062
(10)
7.63
7.63
7.60
7.60
7.49
7.65
7.65
7.61
7.59
7.65
7.62
7.62
7.62
7.54
7.49
7.42
7.54
7.54
7.63
7.55
7.44
7.50
8.11
7.78
(†) -- In units 12+log(O/H). Data are from Ronnback & Bergvall (1995); Lee et al. (2004); Kniazev et al. (1998);
Kniazev et al. (2003); Fricke et al. (2001); Guseva et al. (2003b), Pustilnik et al. (2006a), Pustilnik et al. (2006b);
(‡) -- Photometric data are from: 1 Pustilnik et al. in prep.; 2 Ronnback & Bergvall (1994); 3 Salzer et al. (2002); 4 Kniazev
et al. (2003); 5 SDSS DR4; 6 Fricke et al. (2001); 7 Papaderos et al. (1999); 8 Gil de Paz et al. (2003)
(a) -- Optical diameters (in ′′, measured at the 25 B-mag. arcsec−2 level);
(∗) -- Corrected for AB according to Schlegel et al. (1998) and with the distances from Table 2.
information on the H i spatial distribution, we used the galaxies'
optical sizes and followed the procedure described by Thuan et
al. (1999). The latter takes into account the statistical relation
between the BCGs' optical size and their characteristic size in
H i.
4. Results and preliminary analysis
The H i-profiles of the studied galaxies, smoothed to 10.5 km s−1
are shown in Fig. 1. The related parameters are presented in
Table 2. The velocity profiles show significant diversity in form
and width. They are described individually in more detail below.
In Sect. 5 we briefly discuss some properties of the observed
galaxies as a group.
In Table 2 we present the following data. In Col. (1) -- the
galaxy IAU type name. In Col. (2) - the central heliocentric ve-
locity of the H i-profile, derived as the mid-point of profiles at
50% of the peak, with its rms uncertainty. The related estimate
of the distance, given in Col. (3), follows that of Karachentsev
et al. (2004) for their Catalog of Neighboring Galaxies, with the
adopted value of H0 = 72 km s−1 Mpc−1. The observed widths
of H i-profiles at the levels of 50 and 20 % of the peak value
with their r.m.s. errors are given in Cols. (4) and (5). Column
(6) gives the observed H i flux (the area under the profile). The
errors for the values in Cols. 4, 5, and 6 are calculated similarly
to the method suggested by Schneider et al. (1986). Where it is
appropriate, we show in Col. (7) H i flux, corrected for the beam
resolution. The respective formula (see below) was discussed in
detail, e.g., by Thuan et al. (1999). The correction for source ex-
tension in the north-south direction is negligible since the NRT
vertical beam (FWHM = 22′) is much larger than the largest size
of any of the target galaxies. The correction for extension in the
east-west direction (for NRT beam of 3.′7) is necessary for some
of the targets with optical diameters larger than 0.′5. We follow
the work by Thuan & Martin (1981), and assume that neutral gas
in these galaxies is distributed like an elliptical Gaussian. Then:
Fc = FH [1 + (a2 sin2 PA + b2 cos2 PA) / θ2]1/2,
where a and b are the FWHM diameters (in arc minutes) of
the assumed elliptical distribution of H i, which are adopted for
BCGs/H ii galaxies (following Lee et al. 2002) as 2.0 times of
the optical sizes a25 and b25 in Table 1. PA is the position angle
of the major axis, and θ is the HPBW of the Nanc¸ay telescope in
the East-West direction. In Col. 8 we present the logarithm of the
total H i mass in solar units, derived according to the well-known
formula below.
4
S.A. Pustilnik & J.-M. Martin: H i in 22 extremely metal-deficient galaxies
MHI = 2.36 × 105 Fc D2,
where MHI
is in M⊙, Fc is the object's integrated flux in
Jy km s−1, and D is the distance to the object in Mpc. Finally,
in Col. 9 we show the ratio M(HI)/LB (in solar units, MB⊙ =
5.48, or LB⊙ = 2.12×1033 erg s−1), where M(HI) is from Col. 8,
and LB corresponds to the absolute magnitude MB from Table 1.
Besides the target galaxies, we detected H i emission from
several galaxies appearing either in the 'ON' beam, or in the
'OFF' beam at the radial velocities outside the uncertainty
ranges for target galaxies. We examined the respective sky re-
gions in NED and for many cases found a candidate galaxy that
probably appeared in our H i spectrum. Their observed H i fluxes,
corrected for the offset attenuation, are given in Col. 7. Below we
give more details on the H i emission of individual galaxies from
this program, as well as other information related to the discus-
sion of their properties in Sect. 5.
4.1. HS0122+0743=UGC993
This object is considered by some authors as a galaxy pair. Its
morphology indeed suggests that we are witnessing merging of
the E and W components, which seem already to be in contact.
We aimed to find more evidence for two different components
in its H i profile and probable high-velocity tails. Previous H i
observations with a lower S/N ratio (Garcia et al. 1994; Lu et
al. 1993) indicated the separate faint component in the H i pro-
file with δV = -- 90 km s−1 relative to that of the main emission.
There was a hint in the Lu et al. (1993) profile of the very low
contrast broad velocity component. In our profile, the fainter
component with δV = -- 75 km s−1, W50 ∼35 km s−1 is well de-
tected, but no broad component is visible. The width of the H i
profile of this object (W50=50 km s−1) is rather narrow for its
high neutral gas mass (M(H i)=2.7×109 M⊙). According to the
optical morphology of this system, the narrow profile is prob-
ably not due to the object being seen close to face-on. The re-
cent GMRT H i mapping of this system shows clear evidence of
merging in this XMD object.
4.2. SDSSJ0133+1342
The galaxy is marginally detected in H i. Its narrow H i profile is
consistent with its low optical luminosity and a small H i mass.
On the DSS-2 image this galaxy is compact (maximum extent
is ∼16′′ or ∼3 kpc), with a bright central knot. Its morphology
is very disturbed. The galaxy is almost connected by a bridge to
a ∼3.5-mag (in B-band) fainter reddish galaxy at ∼13′′ N, and
looks like a merging system.
4.3. SDSSJ0205−0949=KUG0203−100
This object has a double-horn profile, typical of disk galaxies,
with W20, corresponding to the rotation velocity amplitude of
∼70 km s−1, which is consistent with its optical classification
in NED as SB(s)m pec. Optical morphology, especially in the
outer parts, is rather disturbed, suggesting significant interaction.
The disk is warped on both edges. A potential candidate for a
disturber is a LSB galaxy, 3.m5 fainter (with M0
B ∼ −12.5 if at
the same distance), at ∼1.′4 SW (∼11.5 kpc in projection). This
galaxy is the only candidate within the NRT beam that might
be responsible for a hint of emission in the H i profile at ∆V ∼
+100 km s−1.
The total B magnitude of SDSS J0205−0949 (as recalcu-
lated from its SDSS photometry) is 15.m46 (not B=18.m38, as in
Kniazev et al. 2003; the latter is likely based on the light picked
up by an optical fiber of only 3′′ aperture, thus missing a lot of
the light). With its absolute magnitude of M0
B = −16.64, it is
somewhat bright for a galaxy with such a low metallicity. While
it could be an analog of the luminous XMD BCG HS 0837+4717
(Pustilnik et al. 2004a), it is worth noting that its cited uncer-
tainty of O/H (0.09 dex) is quite large. Hence, its XMD classifi-
cation should be confirmed with higher quality spectral data. It
is worthwhile noting the high ratio M(HI)/LB=2.57, which is the
second highest in this sample.
4.4. AM0624 -- 261=ESO489-056
The first H i detection for this LSB galaxy was presented by
Gallagher et al. (1995) with the nominal S/N ratio of 4.3. We
achieved an S/N ratio of ∼36. All but one of the profile pa-
rameters of the old and new data are consistent. Only the in-
tegrated H i-flux is higher on our data by a factor of 1.35. The
most reasonable explanation of this difference is the rather low
S/N ratio of the old data. The measured W50 = 27 km s−1 (after
smoothing to 5.2 km s−1) translates (according to Staveley-Smith
et al. 1992, with the adopted isotropic velocity dispersion σ =
10 km s−1) to the disk maximal rotation velocity of 8 km s−1.
An inclination correction is applied for i = 44◦ (Ronnback &
Bergvall 1994). The total H i mass of this galaxy of 1.6×107 M⊙
is the lowest in this subsample. However, this is compatible with
what one expects for its very low blue luminosity, which is also
the lowest in the subsample. The ratio M(H i)/LB = 0.85 is typi-
cal of LSB galaxies (Schombert et al. 2001). We detect an addi-
tional H i source, seen at Vr sufficiently close to that of the Milky
Way. It is discussed in Sect. 5.3 as a possible high-velocity cloud
(HVC).
4.5. HS0846+3522
This galaxy is detected near the correct optical velocity of
2177 km s−1 with a S/N ratio of only 3.4. Its velocity in the
discovery paper (Pustilnik et al. 1999) is mistaken due to a
misprint (given as 2481 km s−1 instead of the correct value of
2184 km s−1). Parameters presented in Table 2 have uncertain-
ties of ∼50%. HS 0846+3522 has a disturbed morphology on
the DSS-2 images, probably due to interaction with an irregular
galaxy at 50′′ NEE (∼7 kpc in projection, B-mag ∼19.6 from
SDSS). Its redshift is unknown.
4.6. KISSB23=KUG0937+298
This galaxy is situated at ∼6 Mpc, in the inner rim region
of the nearby Lynx-Cancer void described in Pustilnik et al.
(2003b). Two more XMD galaxies are found within the same
void (Pustilnik et al. 2003b, 2005a). While the blue luminos-
ity of KISSB 23 is only a factor of 2.8 higher than that of
AM 0624 -- 261, this galaxy, with a similar H i mass has a sig-
nificantly broader H i profile: W50 ∼70 km s−1. Even for an
edge-on case this translates to an amplitude of Vrot (according
to Staveley-Smith et al. 1992) of about 35 km s−1 in comparison
to 8 km s−1 for AM 0624 -- 261. Despite having similar param-
eters of Lopt and M(H i), KISSB 23 is significantly more DM
dominated than AM 0624 -- 261. This is related to the enhanced
stability of its ISM to local intrinsic perturbations. However, this
galaxy shows enhanced SFR. The EW(Hβ) of its bright non-
central knot of ∼30 Å (Lee et al. 2004), for the adopted metal-
licity of z=0.001 corresponds to an age of an instantaneous star-
S.A. Pustilnik & J.-M. Martin: H i in 22 extremely metal-deficient galaxies
5
Fig. 1. The NRT H i profiles Sν (in mJy) vs. Vhel (km s−1) of all studied galaxies with the resolution of 10.5 km s−1. The vertical
arrows at the bottom with the horizontal bars show the optical velocity and its ±1σ uncertainty for weak detections or non-detected
objects.
burst of ∼10 Myr (Leitherer et al. 1999). The origin of such a
non-central SF "burst" is unclear. The possibility of a trigger by
an external perturber can be investigated through H i mapping
of this galaxy. Its ratio M(H i)/LB = 0.98 is rather high for dIrr
galaxies.
a factor of 2.5 lower than that of the outstanding XMD BCG
Tol 65.
4.8. HS1013+3809andUGC5540
4.7. HS0940+4025
This galaxy is marginally detected (S/N ratio of 2.5), so its pa-
rameters have large uncertainties. The object is very compact
and elongated, with the total extent on the DSS-2 ∼9′′ (3 kpc).
The SDSS image shows two very blue knots almost in contact
(∼3′′ separation), very similar in appearance to HS 2236+1344,
for which there are clear indications of a merger. The same
can be attributed to HS 0940+4025. Thus, more detailed study
of this BCG is necessary. For its rather small profile width
(∼50 km s−1), this galaxy has one of the largest M(H i), only
This BCG presumably comprises a physical pair with a ∼2.m3
brighter (MB = −17.7) Sc galaxy UGC 5540 at 8′ to the South
and ∼0.′5 to the West (in total ∼50 kpc in projection). The latter
was observed in the H i line with the 91 m Green Bank (here-
after GB) telescope and the NRT (Haynes & Giovanelli 1991;
Theureau et al. 1998). Taking into account the confusion of the
two galaxies in the NRT beam, some useful estimates of the in-
tegrated H i flux of HS 1013+3809 can be obtained. If we ac-
count for flux attenuation due to an offset of UGC 5540 from
the NRT pointing, we can estimate from the observed flux the
residual that should be attributed to the H i emission related to
6
S.A. Pustilnik & J.-M. Martin: H i in 22 extremely metal-deficient galaxies
HS 1013+3809. In particular, the value of its H i mass can be
derived.
these galaxies (in preparation) shows disturbed H i morphology
in both galaxies, implying strong interaction in the system.
The two cited measurements of UGC 5540 (F(H i,GB)=
5.37±0.56 and F(H i,NRT) = 6.20±1.20 Jy km s−1) are not sig-
nificantly different. However, the 0.83 Jy km s−1 higher value
of the NRT flux, in comparison to that measured at GB, may
indicate some contribution from HS 1013+3809 to the pointing
toward UGC 5540. Due to the narrower beam for GB, this is not
the case for F(H i,GB). Therefore, for further analysis we accept
that the correct F(H i) of UGC 5540 is given by GB data. Taking
into account the attenuation factor (of 0.655) due to the pointing
offset between the two galaxies, the expected contribution from
UGC 5540 at this position is F(H i) = 3.52±0.37 Jy km s−1.
F(H i) = 3.23±0.34 Jy km s−1 when the horizontal NRT beam
resolution effect on the extended H i emission of UGC 5540 is
applied. Subtracting this expected flux from the measured value
of 4.74±0.19 Jy km s−1, we estimate the flux for HS 1013+3809
of F(H i)=1.51±0.39 Jy km s−1. The derived value is consistent
with the NRT measured flux of UGC 5540. Indeed, the con-
tribution of HS 1013+3809 at the NRT pointing to UGC 5540
will be 0.99±0.25, and after addition with F(H i)=5.37±0.56 for
UGC5540 from GB data results in 6.36±0.61, very close to the
NRT measured value of 6.20 Jy km s−1. The estimated flux of HS
1013+3809 comprises only 32% of the measured H i flux. To re-
solve this galaxy pair in H i requires mapping with a beam-width
of ∼1′.
4.9. HS1033+4757
This very compact and low luminosity galaxy has a bright knot
that is somewhat displaced from the center of the main body. The
outer morphology is irregular with many filaments. Its H i profile
is rather broad and asymmetric, suggesting possible confusion.
We have checked all sufficiently bright galaxies falling within
the main NRT beam. All those for which there are SDSS spectra
appeared as background objects.
4.10. HS1059+3934andKUG1059+395
The measured H i flux for the NRT pointing at the position of
HS 1059+3934 is 1.39 Jy km s−1. Its H i profile is quite narrow
(W50=59 km s−1), indicating a small rotation velocity. The in-
clination correction (i=62◦, sin i=0.88) is not large, at least as
it can be estimated from the DSS-2 image, where the axial ra-
tio b/a ≈0.5. The nearest galaxy, which could potentially affect
the H i signal of HS 1059+3934, is KUG 1059+395, an irregu-
lar galaxy with a bright central knot, 1.m2 brighter than the target
galaxy, at 1.′9 to the West (25 kpc in projection). We obtained the
optical spectrum of KUG 1059+395 with the SAO 6m telescope
and found this to be a starburst galaxy with radial velocity close
to that of HS 1059+3934 (Pustilnik et al. 2006b).
To resolve the potential confusion, we carried out NRT ob-
servations of KUG 1059+395. The observed H i profile is cen-
tered at a velocity very close to that for HS 1059+3934. A small
extra feature is seen at the high velocity end. The derived inte-
grated flux (without the extra feature) is 1.38 Jy km s−1. Having
two independent measurements and a known attenuation factor
due to offset positions (0.475) for both pointings, we solved this
system in several iterations. The total flux is 0.95 Jy km s−1 for
HS 1059+3934 and 0.93 Jy km s−1 for KUG 1059+395. The ex-
tra feature detected for KUG 1059+395, with the total flux of 0.3
Jy km s−1, is probably an artifact. The recent GMRT mapping of
4.11. SDSSJ1105+6022anditsneighbors
There is some H i signal in the off-source position, which looks
as a negative amplitude profile centered at Vhel ∼1260 km s−1.
We have checked possible candidate galaxies, and found that the
most probable is MCG+10-16-062, a galaxy of B=16.3, which
is at ∆α = 7s.0 (52′′) and ∆δ=6.′7 from the off-source pointing.
Its radial velocity in SDSS is given as 1273±3 km s−1. Its offset
results in the attenuation factor of 0.75. This galaxy is situated
at 9.′4 SE from our target object.
Another potential source of confusion is a disturbed
DIG/LSB galaxy of a comparable size and total magnitude,
MCG+10-16-056 (mB ≈16.5), which is only at 2.′4 west from
the target XMD galaxy. Its redshift is unknown. Since this
can also contribute to the measured H i flux, we conducted an
NRT observation with pointing to MCG+10-16-056. It was de-
tected at a velocity of 1265 km s−1, which is very close to that
of MCG+10-16-062. The contribution of the latter in the off-
source position for this observation was also significant. Having
these 3 results, we solve this system by iterating the fluxes for
all three galaxies. This method converges within the observa-
tional errors. The derived parameters of these galaxies are given
in Table 2. The XMD galaxy and these two new H i detected
dwarfs (at the projection distances of 14 and 55 kpc) presumably
form an isolated 'dwarf' group. The preliminary results of the
GMRT mapping (in preparation) reveal H i-line emission from
all three described above galaxies, with H i morphology of SDSS
J1105+6022 and MCG+10-16-056 typical of a merging pair.
4.12. SDSSJ1121+0324=NGC3640A
This galaxy is well detected with a S/N ratio of ∼10. Its pro-
file, resembling a double-horn one, is consistent with its op-
tical appearance as an inclined disk. This disk has two bright
regions displaced from the center and rather disturbed outer
isophotes. Hibbard & Sansom (2003) show this galaxy (called
NGC 3640A) to be a probable companion of elliptical galaxy
NGC 3640 (at a projected distance of ∼70 kpc), which belongs
to galaxy group LGG 233. The H i-line VLA map of this region
shows no confusing galaxies. Accordingly, its integrated param-
eters, derived in that work, are consistent with those obtained on
the NRT data.
The B-magnitude of 17.m91 from Kniazev et al. (2003) relates
only to emission from the bright H ii region. Our independent
estimate of its magnitude (from SDSS) of B=17.m88 is consis-
tent with theirs. The B-magnitude for the rest of the galaxy (also
from SDSS photometry) is 17.m18. Summing up the light of both
objects results in Btot=16.m92.
4.13. SDSSJ1201+0211
This galaxy with rather disturbed morphology is a probable
merger with the fainter (B ∼20.4) companion at 17′′ NW (1.4
kpc in projection). It lies at the southern edge of the Virgo
Cluster. No known galaxies in NED with the radial velocities
within our velocity range are expected to be in the NRT beam.
A significant signal, corresponding to an object near the off-
source position, is detected at Vhel of ∼2110 km s−1. The latter
falls within the range of the Virgo cluster members in the di-
rection close to its center. However, for the distant southern pe-
S.A. Pustilnik & J.-M. Martin: H i in 22 extremely metal-deficient galaxies
7
Table 2. H i observed and derived parameters of the sample and additional (below the line) galaxies
IAU name
(1)
0017+1055
0122+0743
J0133+1342
J0205−0949
0624−261
0846+3522
0937+2949
0940+4025
1013+3809
1033+4757
1059+3934
J1105+6022
J1121+0324
J1201+0211∗
1159+545
J1215+5223
1214−277
1223−359
1442+4250
1704+4332
2134+0400
2236+1344
HI0624−2614
1059+395
1102+6038
J1106+6015
J1202+0215
HIJ1218−2801
1342+4210
2236+136
V(HI)
km s−1
(2)
5630±30
2899± 5
2580± 4
1885± 1
491± 1
2169± 3
505± 2
5353± 8
1169± 4
1541± 9
3019± 3
1333± 3
1171± 3
974± 3
3560±15
158± 1
7795±50
2790± 3
647± 1
2082± 8
5090± 4
6160±20
188± 2
3023± 5
1265± 8
1252± 8
2107± 8
7392± 3
3850±21
5122±10
Dist.1
Mpc
(3)
81.3
42.3
38.1
26.8
5.0
29.6
6.0
74.2
15.8
21.8
41.9
19.9
17.0
17.0
50.3
4.2
105.8
36.3
10.6
31.9
73.9
89.1
0.02
41.9
19.0
19.0
27.1
100.2
54.6
75.8
W50
km s−1
(4)
50‡
50± 6
33± 7
112± 1
27± 1
33± 6
77±10
40±16
86± 7
86± 7
59± 6
48± 6
89± 6
29± 7
...
27± 1
...
40± 6
85± 2
33±15
25± 9
50‡
37± 5
62±10
43± 9
57± 9
33±16
117± 6
70±42
105±16
W20
km s−1
(5)
...
123±16
39±11
133± 2
42± 2
40±10
98±15
53±24
130±12
144±29
91± 9
74±10
112±10
53±10
...
43± 2
...
56± 9
99± 2
59±24
33±14
...
58± 7
101±16
70±15
80±15
76±24
127±10
120±67
120±25
Jy km s−1
Jy km s−1
Fc
(7)
FH
(6)
<0.20
6.70±0.26
0.10±0.05
11.07±0.19
2.69±0.08
0.10±0.03
1.95±0.24
0.20±0.08
4.74±0.19
1.32±0.15
1.39±0.06
2.40±0.17
2.49±0.15
0.96±0.09
<0.10
5.14±0.14
<0.10
2.13±0.24
6.37±0.14
0.24±0.05
0.12±0.05
<0.15
0.77±0.06
1.38±0.09
2.30±0.17
2.10±0.18
0.83±0.10
0.46±0.06
0.40±0.10
0.30±0.08
7.60±0.30
12.19±0.22
2.79±0.08
2.05±0.25
1.51±0.39
2.48±0.18
2.67±0.16
5.24±0.16
7.05±0.16
2.58±0.19
0.97±0.12
0.63±0.17
Log(M(HI)
M⊙
(8)
<8.29
9.47
7.54
9.35
7.22
7.32
7.24
8.42
7.90
8.02
8.60
8.36
8.29
7.82
<7.72
7.34
<8.42
8.82
8.24
7.76
8.19
<8.35
>1.86
8.59
8.34
8.25
8.23
9.04
8.32
8.91
M(HI)/LB
M⊙/L⊙
(9)
<0.39
1.48
0.23
2.57
0.85
0.24
0.98
0.70
0.57
2.32
1.54
1.62
1.94
1.92
<0.43
1.01
<0.28
2.70
1.77
0.75
0.76
<0.25
0.86
1.52
1.02
4.5:
1.21
0.15
1 -- Distances adopted similar to Karachentsev et al. (2004) with H0 = 72 km s−1 Mpc−1.
‡ -- Adopted from the total ionized gas velocity amplitude
∗ -- Adopted to be the distance of Virgo cluster
0017+1055 -- σnoise = 1.8 mJy. For the F(HI) upper limit the value ∆V=50 km s−1 is used from Hα data.
J0205−0949 -- Correction factor of 1.10, calculated as for Sc galaxy NGC 5540, see Sect. 4.3.
0624 -- 2614 -- Distance from Karachentsev et al. (2004)
0937+2949 -- Distance is estimated as in Karachentsev et al. catalog (2004), where this galaxy is absent.
1013+3809 -- Resolved confusion with Sc galaxy of ∼2 mag. brighter, at 8′ to the S, see text.
1059+3934 -- Resolved confusion with KUG 1059+395 at 1.9′, see text.
J1105+6022 -- In a group of three dwarfs, see text.
1159+545 -- σnoise = 1.0 mJy.
J1215+5223 -- Correction factor of 1.02, accounting for H i structure from GMRT (Begum et al. 2006)
1214 -- 277 -- σnoise=1.0 mJy. FLASH J121809.29−275219.5 is probably a member of the same group.
1223 -- 359 -- A galaxy at 2.′1 to the East and 4′to the South.
2236+1344 -- σnoise=1.2 mJy. For the F(HI) upper limit the value ∆V=50 km s−1 is used from Hα data.
HI0624−2614 -- HVC? Assumed D = 20 kpc. Unknown correction for angular size/offset.
1059+395 -- Observed to resolve confusion for HS 1059+3934. See text.
1102+6038 -- MCG+10-16-056, observed to resolve confusion for SDSS J1105+6022.
J1202+0215 -- LSB/dI galaxy near J1201+0211 OFF-source position. Flux is corrected for offset attenuation.
HIJ1218−2801 -- A source near OFF-source position J121830.74−280125. No NED candidates.
2236+136 -- In case this is KUG 2236+136 seen in the OFF-source position, the offset attenuation is
of ∼2.1. However, this can be a blend since its velocities in UZC (5188 km s−1) and SDSS (5224 km s−1)
catalogs differ significantly from the value found here.
riphery this velocity is too high and this H i object is far behind
the Virgo cluster. Its W50 is ∼30 km s−1, suggesting very slow
rotation or large inclination. There are two candidate faint op-
tical galaxies within the NRT beam near this position. The first
one is an edge-on disk, at 100′′ N, 25′′ W, and the second is a
face-on LSBG, at 288′′ N, 25′′ W. Considering the very narrow
H i profile, the 'face-on' LSBG galaxy is a more probable coun-
terpart. However, its B ∼18.8 (from SDSS) implies a very high
ratio M(HI)/LB ∼4.5. This requires further study.
8
S.A. Pustilnik & J.-M. Martin: H i in 22 extremely metal-deficient galaxies
4.14. SDSSJ1215+5233=CGCG269−049
This galaxy belongs to the Canis Venatici (CnV) I galaxy cloud
at a distance of ∼4 Mpc (e.g., Karachentsev et al. 2003). Its nar-
row H i profile, with W50 =25 km s−1, is consistent with its
very low optical luminosity. The inclination correction for its
b/a =0.3 (Karachentsev et al. 2004) is small, ∼1.03. Its nearest
neighbor galaxy UGC 7298 (at the projected distance of 14.3
kpc) has a similar radial velocity of 146 km s−1. The distance
of UGC 7298 of 4.21 Mpc is determined through the tip of the
RGB (Karachentsev et al. 2004). We therefore consider the dis-
tance to SDSS J1215+5233 to be equal to 4.2 Mpc as well. These
galaxies seem to be a close pair and tracers of the recent/current
interaction might be visible in their H i morphology. Indeed, the
H i maps presented by Begum et al. (2006), give some hint of
disturbance in both galaxies of the neutral gas morphology and
its kinematics.
4.15. Tol1214−277=Tol21
This BCG is the most distant in our sample. Despite the long
integration and the low noise (σ ∼1 mJy for the effective ve-
locity resolution of 10.5 km s−1), no signal was detected in the
spectrum at or near the optical velocity. However, we have de-
tected an object near the off-source position (a negative signal
with a 'double-horn' profile) with W50 =117 km s−1, at Vhel
of 7392 km s−1. We have checked known galaxies close to the
off-source beam position (72.5 s in RA, or 16′ to the East)
and found no appropriate candidates in NED. Several galaxies
around Tol 21 have Vhel of 7600 -- 7740 km s−1, implying that
Tol 21 probably belongs to a loose group. If this H i object is a
galaxy with M(HI)/LB ∼1, its measured integrated flux F(HI) of
0.46 Jy km s−1 corresponds to Btot ∼18. Accounting for a prob-
able offset and a lower value M(HI)/LB, it can be even brighter.
The upper limit for the H i mass of Tol 21 is derived assuming
that its W50 < 100 km s−1 and the peak value of the H i profile
is lower than 2 mJy. The upper limit for the ratio M(H i)/LB is
low (<0.28). For most of other XMD galaxies this parameter is
(significantly) larger. This implies that this object belongs to the
lower end of the XMD galaxy H i mass function. Probably the
neutral gas "deficiency" is related to the detectability of Ly-α
emission in this BCG. This galaxy is the only one among four
XMD BCGs in which Ly-α emission was detected in the HST
observations (Kunth et al. 1994; Thuan & Izotov 1997). The very
faint galaxy G2 with close Vhel (Izotov et al. 2004a) at a projected
distance of 14.5 kpc is probably a companion of Tol 21.
4.16. Tol1223−359=Tol65
This galaxy is quite unusual since its H i profile is rather narrow
for its large H i mass of 0.66×109 M⊙. The inclination correc-
tion is difficult to infer since its optical morphology is strongly
disturbed at outer isophotes (Papaderos et al. 1999). It resembles
SBS 0335−052 E, whose H i mass is only a factor of 1.2 higher.
Their total H i linewidths are also similar.
Tol 65 is situated at
the outskirts of a galaxy group.
There are several galaxies with close radial velocities at an-
gular distances of 38′ to 53′. One irregular galaxy, FLASH
J122530.92−360714.5 (Kaldare et al. 2003), with Vhel =
2781±56 km s−1 and mb =15.7, at 7.′5 to the NW, could con-
tribute some H i emission to the observed flux. However, due to
offset attenuation, its measured flux will be ∼9% of the actual
value. Even for M(H i)/LB ∼1, this galaxy would contribute no
more than 20% to the total H i flux of Tol 65. The close LSBD
G1 (Papaderos et al. 1999) is about 5 magnitudes fainter and
would give only a minor contribution to the observed H i flux.
4.17. HS1442+4250=UGC9497
This galaxy appears to be an edge-on disk with two prominent
off-center H ii-regions and several fainter ones (Kniazev et al.
1998; Guseva et al. 2003b). It is one of the nearest XMD BCG
known. Its deep V, I photometry and long-slit spectroscopy show
no evidence of a stellar population older than 2 Gyr (Guseva et
al. 2003b). Its high S/N ratio H i double-horn profile is rather
wide, indicating significant rotation. Some asymmetry in its H i
profile at the high-velocity edge of ∼10% of the peak value may
indicate a low mass companion, or a galaxy within the NRT
beam. However, a better S/N ratio profile is necessary to con-
firm this. There is a fainter red galaxy almost in contact, but its
radial velocity is unknown, so this might be a background object.
4.18. HS1704+4332
This galaxy has quite a narrow profile, consistent with its low
luminosity. The galaxy is situated in the region of low density
luminous galaxies (L & L∗) with the nearest one being situated at
4.7 h−1 Mpc. The nearest known subluminous Updated Zwicky
Catalog (UZC) galaxy is at 3 h−1 Mpc.
4.19. HS2134+0400
This is the most metal-poor BCG and the faintest galaxy in B-
band in this subsample. The expected H i flux is hence also low.
There is a narrow peak (W50=25 km s−1) at a velocity close to
the optical one, but its amplitude is not very significant, of the or-
der of 3 σnoise. Its morphology on the deep B-band image looks
rather disturbed, with a plume at the W edge that suggests its
probable merger nature (Pustilnik et al. 2006a), since no disturb-
ing galaxies are present in its environment. The nearby small
irregular galaxy South of the BCG is a distant background ob-
ject.
4.20. Non-detectedXMDgalaxies
Of four undetected XMD galaxies three are the most distant in
the sample, with Vhel in the range of ∼5600 to 7800 km s−1. For
these four galaxies (HS 0017+1055, SBS 1159+545, Tol 21, and
HS 2236+1344), the Notes for Table 2 give estimates of σnoise
for the resolution of 10.5 km s−1 in the spectral regions near their
optical velocities. To derive upper limits of their M(H i), we as-
sumed that their peak flux densities Fpeak are lower than the vis-
ible peak flux density near the optical velocity plus 2σnoise (in
mJy), and that their profiles are boxcar function with W50 less
than 100 km s−1. Then their integrated H i flux F(H i), is less
than Fpeak× W50. For SBS 1159+545 the situation is complicated
since close to the optical velocity we have a faint negative signal,
indicating possible confusion with an object in the 'OFF'-source
beam. However, we failed to find a reliable optical counterpart
for this negative signal. If the negative peak near the optical ve-
locity is treated as noise, a 3σ upper limit of the peak flux density
for SBS 1159+545 is ∼1 mJy. We present its parameters based
on this value in Table 2. For W50 in the spectra of HS 0017+1055
and HS 2236+1344 we used the full amplitude of the ionized gas
velocity as measured in the Hα-line (∼50 km s−1) from the SAO
6 m telescope long-slit observations (in preparation).
S.A. Pustilnik & J.-M. Martin: H i in 22 extremely metal-deficient galaxies
9
The ratio M(H i)/LB < 0.25 for HS 2236+1344 is one of the
lowest among the XMD BCGs observed in this work. There are
two additional factors that could affect its value. They both are
related to its probable 'atypical' nature. As the analysis of its
morphology shows (two bright, almost contacting star-forming
knots with disturbed periphery and plumes) and the unusual ion-
ized gas kinematics (Pramskij et al. 2003; Pustilnik et al. in
preparation), this object is very likely an advanced merger. The
strengths of starbursts in mergers are the largest in comparison
to other cases since the main fraction of available gas can be in-
volved in SF. This implies that in mergers, the M(H i)/LB ratio is
shifted down from its 'typical' value. The second reason is that
conditions in the ISM of merging gas-rich galaxies are favor-
able for transforming the whole H i cloud population to H2 (e.g.,
Elmegreen 1993).
4.21. Non-XMDgalaxies
The parameters of the observed and detected non-XMD galaxies
are presented at the bottom of Table 2. We divide them into three
categories.
First: the targets previously considered as XMD galaxies, but
later eliminated from this group due to better precision O/H de-
termination. The only galaxy in this category is HS 1342+4210,
with the value of 12+log(O/H)=7.89. This is probably interact-
ing with a fainter companion at 13′′ SEE, as seen on the DSS-2
image.
Second: the targets observed to resolve the potential or cer-
tain confusion, either in the 'ON'-source, or in the 'OFF'-source
beam. They include KUG 1059+395, MCG+10-15-056 and
MCG+10-15-062.
Third: new H i sources which appeared occasionally in the
'OFF' beam in the range of ±1300 km s−1 near the velocity of
the target XMD galaxy. One of them, near the position of HS
2236+1344, could be tentatively identified (based on the infor-
mation from NED) with the cataloged galaxy KUG 2236+136.
Another faint galaxy, J1202+0215, detected near the position of
SDSS J1201+0211 (see Sect. 4.13) is not cataloged. For the faint
source HI J1218−2801 near Tol 21 we did not find any optical
counterpart.
5. Discussion and conclusions
As mentioned in the introduction, four evolution scenarios can
possibly explain the existence of XMD galaxies. For old galax-
ies with very low SFR (kind of LSBGs) and truly young galaxies
for which one expects the closed-box to be valid, the small range
of O/H in our sample corresponds, according to estimates, e.g.,
of Pilyugin & Ferrini (2000) to a gas mass fraction µ > 0.9.
However, if the low metallicity of a galaxy ISM is acquired due
to exchange with the intergalactic medium, the gas mass-fraction
can be significantly lower, e.g., MHI . Mstars. The estimate of
H i mass from the integrated H i flux is quite straightforward.
However, the estimate of the total stellar mass in star-bursting
galaxies requires the separation of the flux from the young pop-
ulation and surrounding H ii regions that often dominate hiding
an old, underlying component. Thus, to address the issue of the
gas-mass fraction, we need good surface photometry of the stud-
ied objects. We just mention that in addition to well-known very
gas-rich XMD galaxies I Zw 18 and SBS 0335 -- 052, the H i data
for Tol 65 from this paper, along with the surface photometry
from Papaderos et al. (1999), indicate that this galaxy belongs to
the same class. We postpone the related analysis based on photo-
metric data to a forthcoming paper. Below we discuss the large
Fig. 2. Relation between M(H i) and LB for the XMD galaxies
from this paper. Middle dashed line corresponds to positions of
objects with M(H i)/LB=1.0, while the upper and lower lines for
objects with this ratio equal 3 and 1/3, respectively. The arrows
show the upper limits on M(H i).
variance of H i parameters of the XMD galaxies and new evi-
dence for the importance of interactions to trigger starbursts in
the XMD BCG sample.
5.1. Varianceof H i andopticalproperties
We briefly summarize the data in Tables 1 and 2. The total range
of oxygen abundances in the sample is small, 0.23 dex, or a fac-
tor of 1.7. However, the total range of B-band luminosities in
the studied XMD galaxies is more than two orders of magnitude
(M0
B from -- 12.5 to -- 17.7, or log(LB/L⊙) = 7.19 to 9.27). The
range of H i mass in the sample XMD galaxies is comparable
to the optical luminosity range: log(M(H i)/M⊙)=7.22 to 9.47.
The latter implies that for galaxies in the very-low metallicity
regime this is not tightly connected with their other global pa-
rameters. For the blue luminosity, this can be partly understood
since the studied galaxy sample includes objects of various types
- LSBDs, DIGs, BCGs, and even merging pairs. The strong star-
bursts in BCGs can brighten the blue luminosity of a progenitor
galaxy by a factor of ∼10 or more (e.g., Kruger 1992), depend-
ing on the starburst strength, its age, and the preburst evolution-
ary state. This brightening, however, could account only for a
part of this large range of LB. Figure 2 shows that the LB and
M(H i) of our XMD galaxy sample are correlated, and have a
mean M(H i)/LB of 1.15. The scatter of the data is smaller than a
factor of ∼2.4. However, five of the sample galaxies show values
of M(H i)/LB that are below the mean by a factor of 3 -- 5. The lat-
ter can indicate evolution with significant metal loss, as for the
nearby old XMD DIGs Sextans A and B, with M(H i)/LB < 0.2.
5.2. InteractionsasaSFtriggerinXMDBCGs
The majority of known XMD galaxies are BCGs, low-mass
galaxies with sufficiently strong SF activity. One of the im-
portant questions on XMD BCGs is the nature of the trigger-
ing mechanisms for their SF bursts. Both external and intrin-
sic mechanisms are proposed to be responsible for the starbursts
observed in BCGs in general, e.g., Salzer & Norton (2000) ar-
gue that BCG starbursts are due to an intrinsic trigger in some
specific BCG progenitors -- gas-rich galaxies with enhanced gas
concentration index. However, if some XMD galaxies are really
young, the external trigger due to gravitational interactions with
10
S.A. Pustilnik & J.-M. Martin: H i in 22 extremely metal-deficient galaxies
other galaxies appears more natural, otherwise it is not clear
why they did not start SF much earlier. The interaction trig-
ger seems to be sufficiently important for starbursts in BCGs in
general (e.g., Taylor et al. 1993, 1995; Pustilnik et al. 2001b).
Therefore, it is natural to check whether sufficiently close com-
panions/neighbors of various masses exist in the vicinity of our
target galaxies.
The majority of galaxies in the studied sample appear to
have various indications of stronger or weaker interaction. First
of all there are direct indications of close companions (both in
terms of projected distance and radial velocity). This relates to
HS 1059+3934 and KUG 1059+395 at 25 kpc, HS 1013+3809
and UGC 5540 at 50 kpc, and SDSS J1105+6022 and MCG 10-
16-056 at 15 kpc. SDSS J1121+0324 is a probable companion
of E3 galaxy NGC 3640 at 74 kpc and a member of the galaxy
group LGG 233. SDSS J1215+5223 is paired with UGC 7298
at 14 kpc. Tol 21 has a tiny companion at 14 kpc (Izotov et al.
2004a). One more XMD BCG HS 0822+3542 has a companion
LSBG at 11 kpc (Pustilnik et al. 2003b; Chengalur et al. 2006).
Several other XMD galaxies HS 0122+0743, HS 2236+1344,
and HS 0837+4717 (Pramskij et al. 2003; Pustilnik et al. 2004a)
show clear evidence for various stages of a merger from both
their optical morphology and ionized gas kinematics.
Other XMD BCGs show only indications for interaction-
induced SF activity based on their morphology. In particular,
HS 0846+3522 and SDSS J1201+0211 have a disturbed exter-
nal morphology and fainter galaxies in their vicinity, suggesting
ongoing merging. HS 2134+0400 has a disturbed morphology of
the outer parts and a plume on the western edge. Another indica-
tion of interaction is an asymmetric H i profile, which is visible
in HS 1033+4757.
In summary, we note that the majority of XMD BCGs are ei-
ther certainly or probably interacting. Our H i studies gave sev-
eral new indications for the importance of tidal trigger to SF
activity. However, more detailed studies of H i morphology and
kinematics, as well as those of the ionized gas in these galaxies,
are necessary to unambiguously support this conclusion.
5.3. Probablenewhigh-velocitycloud
A well-detected H i source in the 'ON' beam in the direction
of AM 0624 -- 261, at Vhel = 188 km s−1 can be seen in Fig. 1.
Its parameters are given in Table 2 under the name HI 0624 --
2614. No candidate galaxies within the NRT beam that could be
counterparts for this H i source are found.
Therefore, the most probable interpretation of this feature
is the emission of a high-velocity H i cloud (HVC) near the
Milky Way. In the HVC name system, it would be HVC 234.3-
16.8+208 (which means galactic longitude and latitude in de-
grees and the local standard of rest velocity in km s−1). The
nearest known HVCs to this position are HVC 234.6-17.2+175,
HVC 233.6-17.2+175, and HVC 233.8-18.7+212 from the
HIPASS HVC catalog of Putman et al. (2002), at angular dis-
tances of 0.5◦, 0.8◦, and 2.0◦, respectively. The FWHMs of their
H i profiles are 33 -- 34 km s−1 in comparison to the value of
37±5 km s−1 measured for HI 0624 -- 2614. These nearby HVCs
have axial ratios of a/b ∼1.5, with the major axis FWHM of 0.8◦
to 2.4◦ and peak brightness temperatures Tb of 0.06 to 0.23 K.
Their total H i fluxes vary in the range of 9 to 97 Jy km s−1. The
NRT beam with FWHM = 0.37◦×0.06◦ is rather small for such
HVCs and due to resolution will pick up only a fraction of the
total H i flux. The flux loss can vary from a factor of ∼10 to sev-
eral hundred. Therefore, if this source is a part of a HVC, the
measured flux of this object 0.77 Jy km s−1 matches the known
range well. Note, that for an extended source with a character-
istic size of &10′, a large reduction of the measured signal will
occur due to subtraction of the source signal in the 'OFF' beam
(here at 13.′5). Thus, all parameters of HI 0624 -- 2614 are consis-
tent with the hypothesis that it is a portion of a HVC similar to
several such nearby objects.
5.4. Conclusions
1. We have measured the integrated H i parameters for 18 ex-
tremely metal-deficient (12+log(O/H) ≤7.65) dwarf galax-
ies, including 15 measured for the first time. For four more
XMD galaxies we obtained useful upper limits on their H i
mass and ratio M(H i)/LB.
2. For the total range of O/H of the studied XMD galaxies of
only 1.7 (0.23 dex), the distribution of their global parame-
ters vary widely. The range of M(H i), as well as of blue lumi-
nosity, is two orders of magnitude (from ∼2×107 to ∼3×109
M⊙, and MB from -12.5 to -17.7).
3. More than 2/3 of the twenty non-LSB XMD galaxies from
this study show evidence of interaction/merger with other
galaxies.
4. In the direction of AM 0624 -- 261 (l = 234.3, b = −16.8), an
H i-object is detected at a radial velocity Vhel = +188 km s−1
(Vlsr= +208 km s−1), with an integrated H i flux within the
NRT beam of 0.77 Jy km s−1 and W50=37 km s−1. A com-
parison with data on nearby high-velocity H i clouds (HVC)
shows that this object is probably a part of a new HVC (ten-
tative name HVC 234.3-16.8+208).
Acknowledgements. We are grateful to the NRT Comit´e de Programmes for time
allocation for this project. SAP is grateful to OdP and CNRS for support of
his visits to Observatoire de Paris where the main part of this work was per-
formed. SAP acknowledges the partial support from the RFBR grant No. 06-
02-16617. The Nanc¸ay Radio Observatory is the Unit´e scientifique de Nanc¸ay
of the Observatoire de Paris, associated as Unit´e de Service et de Recherche No.
B704 with the French Centre National de la Recherche Scientifique. The Nanc¸ay
Observatory also gratefully acknowledges the financial support of the Conseil
Regional de la R´egion Centre in France. We thank the anonymous referee for
useful comments and suggestions that allowed us to improve our presentation.
We acknowledge the use of the NED and LEDA databases to check the informa-
tion on the properties of the studied dwarf galaxies and their environment, and
the SDSS database for the use of photometry for many of the studied galaxies
and of spectral information on some objects in their environment.
References
Adelman-McCarthy J.K., et al. 2006, ApJS, 162, 38
Asplund, M., Grevesse, N., Sauval, A.J., Allende Prieto, C., & Kiselman, D.,
2004, A&A, 417, 751
Begum, A., Chengalur, J., Karachentsev, I., Kajsin, S. & Sharina, M. 2006,
MNRAS, 365, 1220
Bergvall, N., Ronnback, J., Masegosa, J., & Ostlin, G. 1999, A&A, 341, 697
Bouwens, R., & Illingworth, G. 2006, New Astronomy Reviews, 50, 152
Cairos, L.M., Caon, N. Papaderos, P., et al. 2003, ApJ, 593, 312
Chengalur, J.N., Giovanelli, R., & Haynes, M.P. 1995, AJ, 109, 2415
Chengalur, J.N., Pustilnik, S.A., Martin, J.-M. & Kniazev, A.Y. 2006, MNRAS,
371, 1849 (Paper III)
Combes, F. 2005, in Starbursts: From 30 Doradus to Lyman Break Galaxies, Ed.
by R. de Grijs & R.M. Gonzalez Delgado, ASS Library, vol. 329. Dordrecht:
Springer, p.167
Doublier, V., Caulet, A., & Comte, G. 1999, A&AS, 138, 213
Elmegreen, B.G. 1993, ApJ, 411, 170
Ekta, Chengalur, J.N., & Pustilnik, S.A. 2006, MNRAS, 372, 853 (Paper II)
Fricke, K., Izotov, Y.I., Papaderos, P., Guseva, N.G., Thuan, T.X. 2001, AJ, 121,
169
Gallagher III, J.S., Littleton, J.E., & Matthews, L.D. 1995, AJ, 109, 2003
Haynes, M.P. & Giovanelli, R. 1991, ApJS, 77, 331
Hibbard, J.E. & Sansom, A.E. 2003, AJ, 125, 667
Garcia, A.M., Bottinelli, L., Garnier, R., Gouguenheim, L., & Paturel, G. 1994,
A&AS, 107, 265
S.A. Pustilnik & J.-M. Martin: H i in 22 extremely metal-deficient galaxies
11
Pustilnik, S., Zasov, A., Kniazev, A., et al. 2003c, A&A, 400, 841
Pustilnik, S.A., Kniazev, A.Y., Pramskij, A.G., et al. 2004a, A&A, 419, 469
Pustilnik, S.A., Pramskij, A.G., & Kniazev, A.Y. 2004b, A&A, 425, 51
Pustilnik, S.A., Kniazev, A.Y., Pramskij, A.G. 2005a, A&A, 443, 91
Pustilnik, S.A., Engels, D., Kniazev, A.Y. et al. 2006a, Astronomy Letters, 32,
228
Pustilnik, S.A., Kniazev, A.Y., Pramskij, A.G., et al. 2006b, A&A, in preparation
Pustilnik, S.A., Engels, D., Kniazev, A.Y. et al. 2006c, A&A, in preparation
Putman, M.E., de Heij, V., Staveley-Smith, L., et al. 2002, AJ, 123, 873
Roberts, M.S., & Haynes, M.P., 1994, ARAA, 32, 115
Ronnback, J., & Bergvall, N. 1994, A&AS, 108, 193
Ronnback, J., & Bergvall, N. 1995, A&A, 302, 353
Salzer, J.J., Di Serego Alighieri, S., Matteucci, F., Giovanelli, R., Haynes, M.
1991, AJ, 101, 1258
Salzer, J.J., Gronwall, C., Sarajedini, V.L. et al. 2002, AJ, 123, 1292
Salzer, J.J., Rosenberg, J.L., Weisstein, E.W., Mazzarella, J.M., & Bothun, G.D.
2002, AJ, 124, 191
Salzer, J.J., Lee, J.C., Melbourne, J., et al. 2005, ApJ, 624, 661
Sargent, W.L.W., & Searle, L. 1970, ApJ, 162, L155
Schlegel, D.J., Finkbeiner D.P., & Douglas M. 1998, ApJ, 500, 525
Schneider, S.E., Helou, G., Salpeter, E.E., & Terzian, Y. 1986, AJ, 92, 742
Schombert, J., McGaugh, S.S., & Eder, J., 2001, AJ, 121, 2420
Skillman, E.D., Kennicutt, R. & Hodge, P. 1989, ApJ, 347, 875
Smith, J.A., Tucker, D.L., Kent, S. et al. 2002, AJ, 123, 2121
Staveley-Smith, L., Davies, R.D., & Kinman, T.D. 1992, MNRAS, 258, 334
Taylor, C.L., Brinks, E., & Skillman, E.D. 1993, AJ, 105, 128
Taylor, C.L., Brinks, E., Grashuis, R.M., & Skillman, E.D. 1995, ApJS, 99, 427
Terlevich, R., Melnick, J., Masegosa, J., Moles, M, & Copetti, M. 1991, A&AS,
91, 285
Theureau, G., Bottinelli, L., Coudreau-Durand, N., et al. 1998, A&AS, 130, 333
Thuan, T.X., & Martin, G.E. 1981, ApJ, 247, 823
Thuan, T.X., & Izotov, Y.I. 1997, ApJ, 489, 623.
Thuan, T.X., Lipovetsky, V.A., Martin, J.-M., & Pustilnik, S.A. 1999, A&AS,
139, 1
Ugryumov, A.V., Engels, D., Pustilnik, S.A., et al. 2003, A&A, 397, 463
van Zee, L. 2000, ApJ, 543, L31
van Zee, L., Haynes, M.P., Salzer, J.J., & Broeils, A.H. 1996, AJ, 112, 129
van Zee, L., Haynes, M.P., Salzer, J.J., & Broeils, A.H. 1997, AJ, 113, 1218
van Zee, L., Haynes, M.P., & Salzer, J.J. 1997, AJ, 114, 2479
van Zee, L., Westpfahl, D., Haynes, M.P., & Salzer, J.J. 1998, AJ, 115, 1000
van Zee, L., Skillman, E.D., & Haynes, M.P. 2006, ApJ, 637, 269
Yan, H., Dickinson, M., Stern, D., et al. 2005, ApJ, 634, 109
Gil de Paz, A., Madore, B.F. & Pevunova, O. 2003, ApJS, 147, 29
Granet, C., James, G.L., Pizzani, J. 1997, IEEE Trans. on Antennas and Prop.
45, 1366
Guseva, N.G., Papaderos, P., Izotov, Y.I., et al. 2003a, A&A, 407, 105
Guseva, N.G., Papaderos, P., Izotov, Y.I., et al. 2003b, A&A, 407, 91
Impey, C., & Bothun, G. 1997, ARAA, 35, 267
Izotov, Y.I., Lipovetsky, V.A., Guseva, N.G., Kniazev, A.Y., & Stepanian, J.A.
1990, Nature, 343, 238
Izotov, Y.I., Lipovetsky, V.A., Guseva, N.G., Kniazev, A.Y., & Stepanian, J.A.
1992, A&A, 247, 303
Izotov, Y.I., & Thuan, T.X. 1999, ApJ, 511, 639
Izotov, Y.I., Papaderos, P., Guseva N.G., Fricke, K.J., & Thuan, T.X. 2004a,
A&A, 421, 539
Izotov, Y.I., & Thuan, T.X. 2004b, ApJ, 616, 768
Izotov, Y.I., & Thuan, T.X. 2005, ApJ, 632, 210
Izotov, Y.I., Stasinska, G., Meynet, G., Guseva, N.G., & Thuan, T.X. 2006, 448,
955
Kaldare, R., Colless, M., Raychaudhury, S., & Peterson, B. A. 2003, MNRAS,
339, 625
Karachentseva, V.E. & Karachentsev, I.D. 1998, A&AS, 127, 409.
Karachentsev, I.D., Sharina, M.E., Dolphin, A.E., et al. 2003, A&A, 398, 467
Karachentsev, I.D., Karachentseva, V.E., Huchtmeier, W., & Makarov, D.I. 2004,
AJ, 127, 2031
Kniazev, A.Y., Pustilnik, S.A., & Ugryumov, A.V. 1998, Bulletin SAO, 46, 23
Kniazev, A.Y., Pustilnik, S.A., Pramsky, A.G., Ugryumov, A.V., & Kniazeva,
T.F. 2001, A&A, 371, 404
Kniazev, A.Y., Grebel, E., Hao, L., et al. 2003, ApJ, 593, L73
Kniazev, A.Y., Pustilnik, S.A., Grebel, E., Pramskij, A.G., & Lee, H. 2004a,
ApJS, 153, 429
Kniazev, A.Y., Grebel, E., Pustilnik, S.A., Pramskij, A.G., & Zucker, D. 2005,
AJ, 130, 1558
Kruger, H. 1992, Ph.D. Thesis, Univ. Goettingen
Kunth, D. Lequeux, J., Sargent, W.L.W., Viallefond, F. 1994, A&A, 282, 709
Kunth, D., & Ostlin, G. 2000, A&ARv, 10, 1
Lee, J.C., Salzer, J.J., Impey, C., Thuan, T.X., & Gronwall, C. 2002, AJ, 124,
3088
Lee, J.C., Salzer, J.J., & Melbourne, J. 2004, ApJ, 616, 752
Legrand, F., Kunth, D., Roy, J.-R., Mas-Hesse, J.M., Walsh, J.R. 2000, A&A,
355, 891
Leitherer, C., Schaerer, D., Goldader, J.D., et al. 1999, ApJS, 123, 3
Lipovetsky, V.A., Chaffee, F.H., Izotov, Y.I., et al. 1999, ApJ, 519, 177
Loose, H.-H., & Thuan, T.X. 1986, ApJ, 309, 59
Lu, N.-Y., Hoffman, G.L., Groff, T., Roos, T., & Lamphier, C. 1993, ApJS, 88,
383
Mac Low, M.-M., & Ferrara, A. 1999, ApJ, 513, 142
Martin, J.-M., Gerard, E., Colom, P., Theureau, G., & Cognard, I., in: SF2A, F.
Combes and D. Barret (eds)., p.237, EDP Sciences, 2002
Matteucci F., & Chiosi C., 1983, A&A, 123, 121
Masegosa, J., Moles, M., & Campos-Aguilar, A. 1994, ApJ, 420, 576
Melbourne, J., Phillips, A., Salzer, J.J., Gronwall, C., & Sarajedini, V.L. 2004,
AJ, 127, 686
Momany, Y., Held, E.V., Saviane, I., et al. 2005, A&A, 439, 111
Ostlin, G., & Mouhcine, M. 2005, A&A, 433, 797
Papaderos, P., Loose, H.-H., Thuan, T.X., & Fricke, K. 1996, A&AS, 120, 207
Papaderos, P., Izotov, Y.I., Fricke, K.J., Thuan, T.X., & Guseva, N.G. 1998,
A&A, 338, 43
Papaderos, P., Fricke, K.J., Thuan, T.X., Izotov, Y.I., & Nicklas, H. 1999, A&A,
352, L57
Papaderos, P., Izotov, Y.I., Thuan, T.X. et al. 2002, A&A, 393, 461
Parodi, B.R., Barazza, F.D., & Binggeli, B. 2002, A&A, 388, 29
Pello, R., Richard, J., Schaerer, D., et al. 2005, Conference proc. "The Fabulous
Destiny of Galaxies: Bridging Past and Present", astro-ph/0510180
Pilyugin, L.S. & Ferrini, F. 2000, A&A, 354, 874
Pramskij, A.G., Pustilnik, S.A., Kniazev, A.Y., & Ugryumov, A.V. 2003, SAO
preprint No. 189
Pustilnik, S.A., Lipovetsky, V.A., Izotov Y.I., et al. 1997, Astronomy Letters, 23,
308
Pustilnik, S.A., Engels, D., Ugryumov, A.V., et al. 1999, A&AS, 137, 299
Pustilnik, S.A., Brinks, E., Thuan, T.X., Lipovetsky, V.A., & Izotov, Y.I. 2001a,
AJ, 121, 1413
Pustilnik, S.A., Kniazev, A.Y., Lipovetsky, V.A., & Ugryumov, A.V. 2001b,
A&A, 373, 24
Pustilnik, S.A., Kniazev, A.Y., Masegosa, J., et al. 2002a, A&A, 389, 779
Pustilnik, S.A., Martin, J.-M., Huchtmeier, W., et al. 2002b, A&A, 389, 405
Pustilnik, S.A., Kniazev, A.Y., Ugryumov, A.V., & Pramskij, A.G. 2003a, ApSS,
284, 795
Pustilnik, S.A., Kniazev, A.Y., Pramskij, A.G., Ugryumov, A.V., & Masegosa, J.
2003b, A&A, 409, 917
|
astro-ph/0402417 | 2 | 0402 | 2004-04-20T20:19:57 | The Many Possible Interpretations of Microlensing Event OGLE-2002-BLG-055 | [
"astro-ph"
] | Microlensing event OGLE-2002-BLG-055 is characterized by a smooth, slightly asymmetric single-lens curve with an isolated, secure data point that is ~0.6 magnitudes brighter than neighboring points separated by a few days. It was previously suggested that the single deviant data point and global asymmetry were best explained by a planetary companion to the primary lens with mass ratio log(q)=-3 to -2, and parallax effects induced by the motion of the Earth. We revisit the interpretation of OGLE-2002-BLG-055, and show that the data can be explained by wide variety of models. We find that the deviant data point can be fit by a large number of qualitatively different binary-lens models whose mass ratios range, at the ~3-sigma level, from log(q) ~ -4 to -1. This range is consistent with a planet, brown dwarf, or M-dwarf companion for reasonable primary masses of M> 0.8 M_sun. A subset of these binary-lens fits consist of a family of continuously degenerate models whose mass ratios differ by an order-of-magnitude, but whose light curves differ by <2% for the majority of the perturbation. The deviant data point can also explained by a binary companion to the source with secondary/primary flux ratio of ~1%. This model has the added appeal that the global asymmetry is naturally explained by the acceleration of the primary induced by the secondary. The binary-source model yields a measurement of the Einstein ring radius projected on source plane of \hat r_E=1.87 +/- 0.40 AU. OGLE-2002-BLG-055 is an extreme example that illustrates the difficulties and degeneracies inherent in the interpretation of weakly perturbed and/or poorly sampled microlensing light curves. | astro-ph | astro-ph |
Draft version November 20, 2018
Preprint typeset using LATEX style emulateapj v. 5/14/03
THE MANY POSSIBLE INTERPRETATIONS OF MICROLENSING EVENT OGLE-2002-BLG-055
B. Scott Gaudi1 and Cheongho Han2,3
Draft version November 20, 2018
ABSTRACT
Microlensing event OGLE-2002-BLG-055 is characterized by a smooth, slightly asymmetric single-
lens curve with an isolated, secure data point that is ∼ 0.6 magnitudes brighter than neighboring
points separated by a few days. It was previously suggested that the single deviant data point and
global asymmetry were best explained by a planetary companion to the primary lens with mass ratio
q = 10−3−10−2, and parallax effects induced by the motion of the Earth. We revisit the interpretation
of OGLE-2002-BLG-055, and show that the data can be explained by wide variety of models. We
find that the deviant data point can be fit by a large number of qualitatively different binary-lens
models whose mass ratios range, at the ∼ 3σ level, from q ≃ 10−4 to ≃ 10−1. This range is consistent
with a planet, brown dwarf, or M-dwarf companion for reasonable primary masses of M & 0.8M⊙.
A subset of these binary-lens fits consist of a family of continuously degenerate models whose mass
ratios differ by an order-of-magnitude, but whose light curves differ by . 2% for the majority of the
perturbation. The deviant data point can also be explained by a binary companion to the source with
secondary/primary flux ratio of ∼ 1%. This model has the added appeal that the global asymmetry
is naturally explained by the acceleration of the primary induced by the secondary. The binary-
source model yields a measurement of the Einstein ring radius projected onto the source plane of
rE = 1.87± 0.40 AU. OGLE-2002-BLG-055 is an extreme example that illustrates the difficulties and
degeneracies inherent in the interpretation of weakly perturbed and/or poorly sampled microlensing
light curves.
Subject headings: gravitational lensing -- planetary systems
1. INTRODUCTION
Planetary companions to Galactic disk and bulge mi-
crolens stars can be discovered via the short duration
perturbation they create to the smooth light curve in-
duced by the parent star (Mao & Paczy´nski 1991). The
majority of these perturbations are expected to be rela-
tively simple and grossly characterized by three observ-
ables: the duration, peak time, and magnitude of the
perturbation. In the ideal scenario, these three observ-
ables are simply related to the three parameters describ-
ing the planetary system: the planet/star mass ratio,
the instantaneous projected separation in units of the
angular Einstein ring radius, and the angle of the source
trajectory relative to the planet/star axis (Gould & Loeb
1992). Unfortunately, reality is a bit more complicated,
and a number of degeneracies have been identified which
can hamper the ability to infer these parameters in prac-
tice. Gaudi & Gould (1997) demonstrated that there ex-
ists an ambiguity in the physical mechanism that sets the
width of low-amplitude perturbations which can result in
an order-of-magnitude uncertainty in the inferred mass
ratio. Gaudi (1998) pointed out that a subset of binary
sources with extreme flux ratios can reproduce the dura-
tion and magnitude of a subset of planetary microlensing
perturbations, although Han (2002) demonstrated that
this degeneracy could be resolved with astrometric ob-
servations during the perturbation. Griest & Safizadeh
1 Harvard-Smithsonian Center for Astrophysics, 60 Garden St.,
Cambridge, MA 02138
2 Department of Physics, Institute for Basic Science Research,
Chungbuk National University, Chongju 361-763, Korea
3 Department of Astronomy, Ohio State University, 140 West
18th Ave., Columbus, OH 43210
Electronic address: [email protected],[email protected]
(1998) discuss a two-fold discrete degeneracy in the pro-
jected separation prevalent in high-magnification plane-
tary events. Along with these anticipated degeneracies, a
few have been uncovered in the process of detailed mod-
eling of observed events. Bennett et al. (1999) invoked a
planetary companion to explain a short-duration devia-
tion seen on a close binary-lens light curve. However, it
was later shown by Albrow et al. (2000) that this pertur-
bation could also be fit by one of the secondary caustics
of the close binary lens, when rotation of the binary is
considered. Gaudi et al. (2002) found a weakly asym-
metric event that could be equally well-explained by a
planetary companion, or parallax deviations arising from
the motion of the Earth. Many of these degeneracies are
'accidental,' in the sense that they arise from chance sim-
ilarities between deviations caused by different physical
situations, rather than by intrinsic degeneracies in the
lens equation itself. They are therefore generally only
approximate degeneracies, and can be resolved with ac-
curate, well-sampled light curves. Given the short dura-
tion and unpredictability of planetary deviations, dense,
continuous and accurate light curve coverage is necessary
to both detect and accurately characterize planetary mi-
crolensing perturbations.
A lensing star with a planetary companion is just an
extreme limit of a binary-lens. As discussed by numerous
authors, binary lenses are themselves subject to numer-
ous degeneracies (Albrow et al. 1999; Jaroszy´nski & Mao
2001; Mao & Di Stefano 1995), some of which are rooted
in symmetries in the lens equation itself (Dominik 1999),
and are therefore nearly perfect (Afonso et al. 2000;
Albrow et al. 2002). Binary lenses in which the source
does not cross any caustics can also be confused with bi-
nary sources. This is especially problematic when only
2
MICROLENSING EVENT OGLE-2002-BLG-055
However,
single-band photometry is available. This may partially
account for the fact that, although they were predicted
to be plentiful (Griest & Hu 1992), only one candidate
binary-source lensing event has been identified4, event
OGLE-2003-BLG-095 (Collinge 1994). Indeed, Collinge
(1994) found that the binary-source model for OGLE-
2003-BLG-095 is only preferred over a binary-lens model
at the ∼ 3σ level.
Source and lens binarity are not the only regimes
where degeneracies are plentiful; global deviations from
the fiducial point-source, point-lens, uniform motion (i.e.
Paczy´nski 1986) light curve have also been found to
be subject to degeneracies. Such deviations come in
a variety of
forms. The motion of the Earth pro-
duces departures from uniform relative motion which
can induce observable deviations from the standard
lightcurve form. These deviations can be quite dra-
matic for events with timescales tE of order or larger
than a year (Smith et al. 2002).
in the
more usual case where tE ≪ yr, the effect of the
motion of the Earth can be approximated by a con-
stant acceleration, which results in deviations that can
be symmetric or asymmetric with respect to the peak
of the event (Gould, Miralda-Escud´e, & Bahcall 1994;
Smith, Mao, & Paczy´nski 2003), although asymmetric
deviations are generally easier to recognize. Unfortu-
nately, as demonstrated by Smith, Mao, & Paczy´nski
(2003), any such weak parallax deviation can also be
explained by acceleration of the source, thus mak-
ing the parallax interpretation non-unique for such
short timescale events. Smith, Mao, & Paczy´nski (2003)
furthermore demonstrated that constant-acceleration
events are subject to a two-fold discrete degeneracy be-
tween the magnitude and direction of the acceleration
and the event timescale. Gould (2003) showed that, in
fact, light curve degeneracies extend to even higher or-
der, and are present when one takes into account not only
the acceleration, but also the jerk. This jerk-parallax
degeneracy has been seen in one event toward the bulge
(Park et al. 2004), and has been invoked to resolve the
discrepancy between the photometric and microlensing
mass determinations of the microlens in the Large Mag-
ellanic Cloud event MACHO-LMC-5 (Gould 2003).
Galactic bulge microlensing event OGLE-2002-BLG-
055 exhibits both a global asymmetry and single deviant
data point. Jaroszy´nski & Paczy´nski (2002) first consid-
ered the interpretation of this event; they fit binary-lens
models and included non-uniform motion caused by ei-
ther parallax or arbitrary uniform acceleration. They
argued that, although both uniform acceleration and
parallax explain the global deviations equally well, the
event was most naturally explained by parallax-induced
deviations, and that the single deviant point was best
fit by a binary-lens with mass ratio 10−3 − 10−2. This
mass ratio implies a companion in the Jupiter mass range
for likely primary masses, and they therefore concluded
that OGLE-2002-BLG-055 was a possible planet candi-
date. Here we revisit the interpretation of this event.
We demonstrate that, in fact, there are many possible
interpretations of OGLE-2002-BLG-055. We find that
the short-timescale perturbation could arise from a stel-
4 See Dominik (1998) and Han & Jeong (1998) for additional
discussions of the apparent lack of binary-source events.
Fig. 1. -- The points with errorbars show the light curve of
OGLE-2002-BLG-055. The long-dashed magenta line shows the
best single-lens, single-source, constant-velocity model fit to the
data, with the single high point and two neighboring points re-
moved. The short-dashed blue line shows the best single-lens,
single-source, constant-acceleration model fit to the same dataset.
The dotted green line shows the best single-lens, binary-source,
constant-acceleration model to the entire data set. The solid red
line shows the best single-source, binary-lens, constant-acceleration
model to the entire dataset. The inset shows a close-up near the
deviation.
lar companion to the source, or a stellar, brown-dwarf, or
planetary companion to the lens. The global asymmetry
could arise from parallax deviations, or from accelera-
tion of the source induced by its companion. All of these
interpretations are indistinguishable at the ∼ 3σ level.
Therefore, the correct interpretation of this event is un-
clear, and it serves as a particularly extreme reminder
of the degeneracies involved with poorly sampled and
weakly-perturbed microlensing light curves.
2. DATA
The Galactic bulge microlensing event OGLE-2002-
BLG-055 was observed as part of the third phase of
the OGLE collaboration (Udalski 2003b), and alerted
during the 2002 bulge season with the Early Warn-
ing System (EWS)5. The OGLE data consist of 106 I-
band data points, with 9 points taken during the 2001
season, 47 during the 2002 season, and 50 taken dur-
ing the 2003 season. The data covering the primary
event are shown in Figure 1. All errors are scaled by
a factor of 1.31, as determined from the 2003 baseline
data. This scaling may be somewhat underestimated,
as the best models found here have unreasonably large
values of χ2, although they appear to faithfully repro-
duce all the features of the event. The majority of
the data for OGLE-2002-BLG-055 appear to follow the
usual single-lens form, with the obvious exception of
the data point at HJD′ ≡ HJD − 2450000. = 2424.9,
which has I = 15.449 ± 0.009, highly discrepant from
5 http://www.astrouw.edu.pl/∼ftp/ogle/ogle3/ews/ews.html
GAUDI & HAN
3
the neighboring points at I ∼ 16 separated by 4 − 8
days. Jaroszy´nski & Paczy´nski (2002) report that this
data point is secure, as evidenced by direct inspection of
the raw image.
Unfortunately, there is only I-band data available for
OGLE-2002-BLG-055, which hinders the interpretation
of the model fits to the data for several reasons. First,
without color information during the event, it is impos-
sible to determine separately the color of the source and
blend. Thus it is not possible to compare the derived
source and blend colors and magnitudes to a local color-
magnitude diagram, which provides an important test
of the viability of a given model.
In fact, as this field
was not observed during the second phase of the OGLE
collaboration, there is no V -band data available for the
source field at all. Thus it is also not possible to deter-
mine the local I-band extinction AI via the usual method
of 'clump-calibration' (Stanek 1996; Wo´zniak & Stanek
1996). There are several publicly-available extinction
maps for the Galactic bulge. The maps of Sumi (2004)
are based on OGLE II data, and thus do not cover the
OGLE-2002-BLG-055 field. Popowski, Cook, & Becker
(2003) have derived extinction maps based on MACHO
collaboration data. We estimate the extinction at the
location of the source (R.A. = 17h59m40.s93, Dec. =
−27◦07′18.′′2(J2000); l = 3.14115, b = −1.75827) as the
mean of the two closest points (at distances of 0.12◦ and
0.16◦) in the Popowski, Cook, & Becker (2003) extinc-
tion maps, weighted by the inverse squared distance.
This yields AV = 3.20, or AI = 1.57, where we have
adopted AI /AV = 0.49 (Sumi 2004; Udalski 2003a).
3. MODELING OGLE-2002-BLG-055: TECHNICAL DETAILS
We fit the flux as a function of time to the standard
form,
F (t) = FsA(t) + Fb,
(1)
where Fs is the flux of the source, and Fb is the flux of any
unrelated blended light. Here A(t) is the magnification
as a function of time t.
For a single lens (SL) and single source (SS), the mag-
nification is,
A0[u(t)] =
u2 + 2
u√u2 + 4
(2)
where u(t) is the angular separation between the lens and
source in units of the angular Einstein ring radius,
θE = pκM πrel,
κ =
4G
c2AU
.
(3)
Here πrel is the lens-source relative parallax, and M is
the mass of the lens.
The angular separation between the source and lens
can be written as,
u2(t) = x2(t) + y2(t).
(4)
We consider two different models for the relative lens-
source motion. For constant relative velocity (CV) be-
tween the source, lens, and observer,
x(t) = τ ≡
t − t0
tE
,
y(t) = u0
(5)
Fig. 2. -- In each panel, the points with errorbars show a close-up
of the light curve of OGLE-2002-BLG-055 for ∼ 15 days before and
after the deviant data point. The solid curve in each panel shows a
different single-source, binary-lens, constant-acceleration fit. The
mass ratio and χ2 of each fit are also indicated. The dotted curve
shows the best single-source, single-lens, constant-acceleration fit
with three data points removed, and is the same as the dashed
curve in Figure 1. The bottom two panels show a continuum of
fits, with the range of mass ratios and χ2 values indicated.
(CA), the two components of the angular separation be-
come (Smith, Mao, & Paczy´nski 2003),
x(t) = τ +
1
2Axτ 2,
y(t) = u0 +
1
2Ayτ 2.
(6)
Here Ax and Ay are the two components of the relative
source-lens angular acceleration in units of θE/t2
E. Note
that tE and µrel are now defined at time t0.
For a SL and binary source (BS), the magnification can
be written
ABS(t) = A0[u1(t)] + ǫA0[u2(t)],
(7)
where tE = θE/µrel is the Einstein timescale, µrel is the
relative lens-source proper motion, t0 is the time of clos-
est approach between lens and source, and u0 is the im-
pact parameter in units of θE. For constant acceleration
where ǫ is the flux ratio between the secondary and pri-
mary source. Here u1 and u2 are the angular separations
between the primary and secondary source and the lens,
respectively.
4
MICROLENSING EVENT OGLE-2002-BLG-055
2(t) + y2
Ax
qs
1
2
1
2
Ay
qs
For a bound binary-source, we expand the vector an-
gular separation of the primary and secondary around
the time of maximum magnification of the primary t0,
and keep terms up to the acceleration. This yields u1(t)
given by Eq. 6, with t0 and u0 the time of maximum
magnification and impact parameter of the primary, and
u2
2(t) = x2
2(t), with
τ 2
1 + bx,
x2(t) = τ1 −
y2(t) = u0 −
τ 2
1 + by,
(8)
where τ1 = (t − t0)/tE, and qs is the mass ratio of the
binary-source system, and b = (bx, by) is the vector an-
gular separation from the primary to the secondary in
units of θE, which has the same direction as the accel-
eration of the primary. Unfortunately, this parameteri-
zation is not ideal for the current event because bx and
by are not directly constrained by the data, but rather
the magnification of the secondary source at the time
of the deviant data point, A(td) ≃ ǫ/u2(td). Therefore,
adopting Eq. 8 would lead to correlations between the
parameters Ax,Ay, t0, u0, bx, and by, which would lead
to complications with fitting and error determination.
Instead, we adopt x2(t) = (t − t0,2)/tE and y2(t) = u0,2.
This assumes that the acceleration of the secondary is
negligible, and that the direction of the acceleration of
the primary is not necessarily the same as b. The ac-
celeration of the secondary has two effects. First, the
effective timescale at the time of the perturbation may
be different from tE. Second, the deviation from uni-
form velocity induces a deviation in the light curve. The
former effect is essentially unobservable, because the dif-
ference in tE can be absorbed into ǫ and u0,2 (Gaudi
1998; Gould 1996). For typical parameters, the devia-
tion in the light curve due to acceleration is . 1% while
the magnification of the secondary is significant. This is
of order or smaller than the errors, and thus is negligible.
The alignment between the source separation vector and
acceleration vector as determined from the fit provides
an important test of the binary-source interpretation.
For a binary lens (BL), the magnification is specified
by three parameters: the mass ratio q of the lens, and
instantaneous angular separation d between the two com-
ponents in units of θE, and the angle α between the vector
pointing from primary to secondary and the direction of
µrel. Here we adopt a coordinate system such that the
origin of the binary-lens is at a distance of qd−1(1 + q)−1
from the primary along the axis toward the secondary.
For BL models with small mass ratios, it is also necessary
to consider the effect the finite size of the source star θ∗
on the magnification, parameterized by ρ∗ ≡ θ∗/θE.
We perform all fits in flux rather than magnitude space,
thus allowing the analytic determination of the values of
Fs and Fb for a given model A(t) that minimize the usual
χ2 goodness-of-fit statistic. All other parameters are
minimized using the downhill-simplex routine AMOEBA
(Press, Teukolsky, Vetterling, & Flannery 1992), using
many different initial trial values of the parameters as
seeds. This procedure works well for all models except
the BL model, which generally produces poorly-behaved
χ2 surfaces that are not well explored by downhill-
simplex routines.
In order to approximately minimize BL models, and
to calculate errors on fit parameters for other models, we
use a Markov Chain Monte Carlo (MCMC)6. At each
step in the chain, we vary a single fit parameter by
drawing a random Gaussian deviate with zero mean,
and dispersion set to appropriately sample the likeli-
hood surface. We then calculate the relative likelihood
Lrel = exp(−∆χ2/2) between the new step, and the old
position. If Lrel ≥ 1, the step is accepted. If Lrel < 1,
then we draw a random uniform deviate between 0 and
1. If this deviate is < Lrel, then the new step is accepted,
otherwise it is rejected. When generated in this way, the
resulting distribution of parameters in the Markov Chain
is proportional to the posterior probability distribution,
provided that the chain has converged. Although we do
not rigorously test for convergence, we confirm that the
chains have been run for a sufficient number of steps, by
visually inspecting different chains started at different
points in parameter space, and ensuring that these have
properly mixed.
We define the 1σ errors on a given parameter as the
projection of the ∆χ2 = 1 contour onto that param-
eter axis. We do not attempt to calculate errors for
the BL models, primarily because we find that there are
large, disconnected regions of parameter space that con-
tain continua of degenerate models. Defining confidence
regions in the presence of such degeneracies is difficult
and not very meaningful.
4. MODELING OGLE-2002-BLG-055: RESULTS
Table 1 shows fit parameters and, where appropriate,
1σ errors for the four different classes of models consid-
ered here. The models are labeled by the lens multiplic-
ity (SL=single lens, BL=binary lens), source multiplic-
ity (SS=single source, BS=binary source), and source
motion (CV=constant velocity, CA=constant accelera-
tion). Also indicated is the dataset used; no number
refers to the entire dataset,
'1' indicates the dataset
with the single deviant data point at td = 2424.9 re-
moved, '3' refers the dataset with points at t = 2416.81,
2424.9, and 2428.89 removed. For the constant accelera-
tion fits, there are generally always two degenerate mod-
els (Smith, Mao, & Paczy´nski 2003), these are labeled 'p'
and 'n' for the fits with positive and negative Ay, respec-
tively. The binary-lens fits are labeled as (a,b,c,d,e,f),
which correspond to the panel labels in Figure 2.
Following Jaroszy´nski & Paczy´nski (2002), we first fit
OGLE-2002-BLG-055 to a standard single-lens, single-
source, constant velocity model after removing the sin-
gle deviant data point at td = 2424.9. This model is
clearly a poor fit to the data, yielding χ2 ≃ 180 for 100
degrees of freedom (dof). Removing two additional data
points on either side of the deviant data point improves
the fit by ∆χ2 = 12 for two fewer dof. This implies
that the two data points on either side of the point at td
likely deviate significantly from the smooth underlying
model, supporting the reality of the single data point,
and pointing toward a perturbation that has a timescale
that is of order the interval between the td and neigh-
boring data points, ∼ 4 days. All constant-velocity fits
are quite poor, deviating significantly from the data dur-
ing the rising part of the primary event (see Figure 1),
likely indicating a significant acceleration of the observer,
6 See Verde et al. 2003 for a particularly clear and concise ex-
planation of using MCMC in a practical setting.
GAUDI & HAN
source, or lens.
Including a constant acceleration improves the fits con-
siderably. For the data set with one data point re-
moved, we find χ2 ≃ 149 for 98 dof. When three
data points are removed, the fit improves by ∆χ2 =
15.7 for 2 less degrees of freedom. As anticipated by
Smith, Mao, & Paczy´nski (2003), we find two degenerate
fits for each constant acceleration model, with different
values of tE, Ax, and Ay, but identical values of the other
parameters and χ2. Including the constant acceleration
terms increases the inferred timescale considerably. The
value of χ2 for the best-fit constant acceleration model
with three data points removed is unacceptably high,
despite the fact that the model appears to reproduce
the primary features of dataset quite well (see Figure
1). Inspection of the light curve during the microlensing
event, while the source is significantly magnified, reveals
short timescale (. 1 day) scatter, with amplitude that
is smaller than the typical size of the photometric errors
at baseline. This scatter may be intrinsic to the source.
Regardless of the cause, it appears that the error scaling
derived from the baseline points is probably underesti-
mated.
A binary-source model can produce a large, short-
timescale deviation like that seen in the dataset of
OGLE-2002-BLG-055, provided that ǫ ≪ 1 and u0,2 ≪
1. Figure 2 shows the best-fit binary-source models,
which have χ2 ≃ 142 for 96 degrees-of-freedom. This
is ∆χ2 ≃ 11 larger than the single-source constant-
acceleration fit with three data points removed. This
additional χ2 arises from the inability of the binary-
source model to simultaneously fit the two data points
immediately following td. As with the single-source
models, there are two binary-source models with es-
sentially equal χ2 due to the constant-acceleration de-
generacy. As anticipated, the fits yield small values
In this regime, the rela-
for the flux ratio, ǫ ≃ 1%.
tions between the binary-source parameters ǫ, u0, and
t0 and the salient features of the perturbation are ex-
ceedingly simple (Gaudi 1998). Note that the param-
eters of the primary (t0, tE, u0,Ax,Ay, Fs, Fb), which
are constrained by the data away from the perturba-
tion, are essentially identical to the parameters found for
the constant-acceleration fits with three data points re-
moved. The duration of the perturbation is constrained
by the points neighboring td. Combined with the param-
eters of the primary, this duration yields ǫ. The excess
flux at time td is ∆F (td) ≃ Fsǫ/u2(td), and thus yields
u2(td), which is much better constrained than t0,2 and
u0,2 separately. We find u2(td) = 0.0025 ± 0.0008 and
u2(td) = 0.0018 ± 0.0007 for models SL.BS.CA.n and
SL.BS.CA.p, respectively. Note that, since the relations
between the perturbation observables and the binary-
source parameters ǫ, t0,2, u0,2 depend on tE, which differs
between the two constant-acceleration fits, the binary-
source parameters also differ between the two fits.
Binary-lens models with extreme mass ratios q can also
yield large, short-timescale deviations.
In the limit of
q ≪ 1, and under some simplifying assumptions, the
relation between the gross features of the perturbation
and the binary parameters q, d, α also take on a rela-
tively simple form (Gaudi & Gould 1997; Gould & Loeb
1992). The angle of the trajectory is roughly α ∼
d + 4], where ud = pτ (td)2 + u2
0.
5
tan−1[u0/τ (td)]. The binary separation is d ∼ 0.5[ud +
pu2
Finally, the
mass ratio is q ∼ p∆t/tE. For fits SL.SS.CA.3n and
SL.SS.CA.3p, these expressions yield α ∼ 0.6 − 0.8,
d ∼ 1.1, and q ∼ 10−3. Of course, these formulas are
only very rough approximations, but the implied param-
eters are good starting guesses. We therefore start with
these parameter ranges, and first alter the parameters by
hand until we find a number of promising approximate
fits. These fits are then used as seeds for an automated
point-source binary-lens minimization routine. All viable
point-source fits are then collected, and the parameters
are used as starting guesses for a second minimization,
using a finite source of angular size
θ∗ = 6 µas 10−0.2(I0−14.32).
(9)
Here I0 is the dereddened magnitude of the source
as derived from the point-source fits. Equation (9)
is derived from the color-surface brightness relation of
van Belle (1999), assuming that the (V − I)0 color of
the source is the same as the red clump.
Standard
models of the Galaxy (Han & Gould 1995, 2003) pre-
dict hθEi = 240+190
−190µas
(disk). These values of θE yield source sizes ρ∗ = θ∗/θE
that are too large to reproduce the data for OGLE-2002-
BLG-055 with small mass ratios q. Although this implies
that such fits are somewhat disfavored, it is not actually
possible to rule out these models with such an argument.
We therefore adopt a value of θE = 550µas. Approx-
imately 6% of bulge lenses and 25% of disk lenses are
expected to have values of θE larger than this.
−130µas (bulge), and hθEi = 330+360
Representative finite-source, binary-lens fits are tabu-
lated in Table 1 and shown in Figure 2. We recover the
fits presented by Jaroszy´nski & Paczy´nski (2002), and
find many other additional viable fits, with mass ratios
ranging from q ≃ 5× 10−4 to ∼ 8× 10−2, at the ∆χ2 . 9
level. The best binary-lens (and best overall) fit has
χ2 = 130 for 96 dof. If one forces χ2/dof = 1 for this
fit, then mass ratios in the range q = 10−4 − 10−2 are all
consistent with the data at the ∼ 3σ level.
In the first class the
There two basic classes of fits.
trajectory crosses the planetary caustic, with the deviant
data point at td occurring when the source is interior to,
or near, the caustic. The second class of fits pass out-
side the planetary caustic, with the point at td occurring
when the source is near the ridge of high-magnification on
the planet-star axis. The second class actually represents
a continuous degeneracy in q, d, and α, with mass ratios
in the range 0.01 . q . 0.1 for ∆χ2 . 14. From Figure 2,
we conclude that while many of the fits would have been
distinguishable with a few additional data points during
the perturbation, the family of continuous fits produce
very similar perturbations. The curves in each of the two
bottom panels of Figure 2 deviate from each other by
. 2% for the majority of the perturbation. Distinguish-
ing between these models would therefore require rather
accurate or densely sampled data. The cause of this
uncertainty in q is briefly discussed by (Gaudi & Gould
1997).
5.
IMPLICATIONS AND DISCUSSION
OGLE-2002-BLG-055
can
well-fit
by
large
number
of
be
reasonably
different models.
6
MICROLENSING EVENT OGLE-2002-BLG-055
of
for
that
Which
is most
showed
Jaroszy´nski & Paczy´nski
(2002) demonstrated that
the asymmetry exhibited by the primary light curve
can be explained by a simple uniform acceleration, due
to either parallax deviations arising from the motion
of the Earth, or non-uniform motion of the source due
to the presence of, e.g., a binary companion. We have
demonstrated that the short-timescale deviation can be
explained by a binary companion to the lens with mass
ratio 10−4 . q . 10−1, consistent with a planet, brown
dwarf, or low mass star. The short-timescale deviation
can also be explained by a binary companion to the
source with a flux ratio of ∼ 1%.
likely?
these models
Jaroszy´nski & Paczy´nski
the
(2002)
parallax and constant acceleration fits had essen-
tially identical χ2-values
the same number of
degrees-of-freedom. Thus either are equally viable in
a goodness-of-fit sense. The inferred timescale of the
event spans the range tE = 50− 100 days, which is in the
range where weak parallax effects are neither surprising,
nor necessarily expected. Furthermore, there has been
no estimation of the expected rate of binary-source
events with detectable asymmetries due to acceleration.
Therefore it is difficult to argue which origin for the
observed asymmetry is a priori more likely. In regards
to the deviant data point, a binary-lens model with
q ∼ few × 10−2 provides the best fit to the data, with
the binary-source model disfavored at ∆χ2 = 11 for the
same number of dof, or slightly more than ∼ 3σ level.
However, this difference in χ2 is driven primary by a
couple of data points just after the perturbation, which
may be affected by the short-timescale variability seen in
other parts of the light curve. If one normalizes the error
bars to force χ2/dof = 1 for the best model, then the
binary-lens model is only favored at the ∆χ2 ≃ 8 level.
The inferred blend and source fluxes can also provide
discrimination between models: one expects these fluxes
to trace, on average, the distribution of fluxes of un-
lensed stars in the field near the line-of-sight. The best
binary-lens models have source and blend magnitudes
of Is ∼ 17 and Ib ∼ 19, whereas the binary-source fits
yield Is ∼ 18 and Ib ∼ 17. Unfortunately, without color
information,
it is difficult to definitively distinguish
between these two scenarios. It is also difficult to argue
which model is a priori more likely, since the frequency
of planetary companions to microlens stars is not known,
and defining the appropriate detection criterion for the
current event, which was culled by eye from an ensemble
of alerted microlensing events, is nebulous at best.
The binary-source model does have the advantage of
simplicity: a bound companion to the primary source
can explain both the short-timescale variability and the
acceleration needed to produce the global asymmetry.
However, in order for this scenario to hold, the projected
acceleration vector must be aligned with the projected
binary-source separation vector.
In the particular pa-
rameterization of the binary-source model used here (see
§3), this angle ∆θ between the two vectors is essentially
a free parameter, and thus it can be used to test the
model. We determine the distribution of ∆θ from the
two Markov chains corresponding to the two degenerate
(in the sense of χ2) binary-source models, SL.BS.CA.n
and SL.BS.CA.p. The second model is immediately ruled
out, as the primary's acceleration vector is pointing away
from the companion, with ∆θ > 90◦, wildly inconsistent
with any bound orbit. The first model has ∆θ = 36.2+1.8
−2.0
degrees. At first sight, this may also seem very inconsis-
tent (> 15σ!) with a bound orbit (∆θ = 0), however
there are several reasons why this is not necessarily cor-
rect. First, although the 1σ uncertainty on ∆θ is small,
there is a substantial non-Gaussian tail toward smaller
values. Enforcing ∆θ = 0 in the fit yields χ2 = 156.7, or
∆χ2 = 15.0 for one less dof. Thus, the bound model is
only 'ruled out' at the ∼ 3.5σ level. Furthermore, as we
argue below, the period of a bound binary source with
separation b . 1 is likely to be only a few times longer
than the timescale of the event. Therefore one expects
significant deviations from uniform acceleration that are
not accounted for in the simplified model adopted here.
Thus the inferred value of the ∆θ may be the result of
inadequacies in the model.
It is therefore conceivable
OGLE-2002-BLG-055 could be fit by a bound binary-
source model in which the effects of rotation are included
self-consistently.
Accepting the binary-source model as correct, one
can use the measured parameters of the viable model
SL.BS.CA.n to determine the Einstein ring radius pro-
jected on the source plane rE = DosθE (Han & Gould
1997), where Dos is the distance to the lens. Assum-
ing circular orbits, the acceleration of the primary due
to the secondary is simply a = Gm2/d2, where d is the
semi-major axis of the orbit and m2 is the mass of the
secondary. The projected acceleration a⊥ is therefore,
a⊥ =
Gm2
d2 (1 − sin2 φ sin2 i)1/2
(10)
where i is the inclination of the orbit, and φ is the phase
of the primary orbit referenced to the intersection of the
orbit and the plane of the sky. The constant-acceleration
y, which is given
fit yields the parameter A = qA2
x + A2
by,
a⊥t2
E
rE
.
A =
(11)
The binary-source fit yields t0,2. This can be combined
with the parameters of the primary event to derive the
two components the projected binary-source separation:
bx = 0.5(Ax/qs)∆τ 2 − ∆τ and by = 0.5(Ay/qs)∆τ 2 − u0.
Here ∆τ ≡ (t0,2 − t0)/tE, and we have assumed that
u0,2 ≪ 1. Then b can be related to the semi-major axis
by,
(12)
d = rEb(cos2 φ + cos2 i sin2 φ)−1/2.
Finally, Equations (10), (11), and (12) can be combined
to arrive at an expression for rE in terms of mostly known
quantities,
rE = (cid:20) Gm2t2
b2A
E
(1 − sin2 φ sin2 i)1/2(cos2 φ + cos2 i sin2 φ)(cid:21)1/3
.
(13)
We now use the probability distribution of the binary-
source model fit parameters, together with equation (13),
to construct the probability distribution for rE. Here
the virtue of MCMC becomes clear:
in order to deter-
mine the probability distribution of rE, we simply have
to determine the value of rE at each link in the chain. Be-
cause the density of points in the binary-source MCMC
GAUDI & HAN
7
parameter chain is proportional to the posteriori prob-
ability distribution, the resulting distribution of values
of rE is simply proportional to the desired probability
distribution.
We first account for the parameters in equation (13)
that are not constrained by the binary-source model;
these include the inclination i, the phase φ, and the mass
ratio qs (which enters via b). For each link in the chain,
we randomly choose ten values of φ and cos i, assuming
a uniform distribution of each. We determine qs using
the fluxes of the primary and the secondary. We draw a
source distance Dos from the standard Galactic model of
Han & Gould (1995, 2003). If the source is in the bulge,
we assume that the primary is a giant, and therefore has
a mass of M1 ∼ M⊙. If the source is in the disk, we de-
termine the absolute magnitude of the primary from the
value of Dos, and assuming an extinction of AI = 1.57
(see §2). We determine the absolute magnitude of the
secondary from the flux ratio ǫ, primary flux Fs, Dos,
and AI . We estimate the masses of the primary and sec-
ondary from their absolute magnitudes using an analytic
approximation to the main-sequence mass-luminosity re-
lationship derived from the solar metallicity isochrones
of Bertelli et al. (1994). This then yields the mass ratio
qs. We then determine rE.
In addition, we also determine the radius R∗,2 of the
secondary from its absolute magnitude, also using the
solar-metallicity isochrones of Bertelli et al. (1994). We
require that ρ∗,2 ≡ R∗,2/rE < u2(td), otherwise we dis-
card the inferred value of rE. We note that the resulting
distribution of rE is not very sensitive to the adopted dis-
tribution of source distances, for several reasons. First,
the distance enters through the factor of b−2/3, since b
formally depends on the mass ratio qs. However, inspec-
tion of the expressions for bx and by reveals that qs mod-
ifies the acceleration terms, which themselves are ∝ τ 2
2 ,
where τ2 = (t0,2 − t0)/tE. For the binary-source fits pre-
sented here, τ 2
2 ∼ 10%, and thus these terms are small.
Second, for bulge sources, where qs is affected the most
by the assumed source distance, the dispersion in source
distances is relatively small. Conversely, for disk sources,
where the dispersion in source distances is large, the in-
ferred value of qs is relatively insensitive to the source
distance (since both primary and secondary are assumed
to be on the main sequence). Had we simply assumed
that all sources were in the bulge with Dos = 8 kpc, the
median inferred value of rE would change by . 2%, or
∼ 0.1σ.
Figure 3 shows the resulting distribution of rE. The
median and 68% confidence interval
is rE = 1.87 ±
0.40 AU. Adopting the standard Galactic model of
Han & Gould (1995, 2003), we use the inferred values
of rE and tE to constrain the mass M , distance Dol and
relative proper motion µrel of the lens. The resulting
distribution of M (assuming a uniform prior in linear
mass) is shown in Figure 3. The median and 68% con-
fidence interval is log M/M⊙ = −0.28+0.58
−0.42. Similarly,
we infer Dol = 7.1 ± 0.9 kpc, θE = 240+45
−40µas, and
µrel = 5.45+1.15
−0.95 km s−1 kpc−1. All of these parameters
are typical for disk-bulge or bulge-bulge lensing events,
except for µrel, which is a factor of ∼ 6 smaller than
the median distribution for bulge-bulge events and a fac-
tor of ∼ 8 smaller than the median of disk-bulge events.
Fig. 3. -- In each panel, the curves show the relative likelihood
of the given parameter, as inferred from binary-source, constant
acceleration fit SL.BS.CA.n (see Table 1 and §5). All curves have
been normalized so that the likelihood is unity at the peak. The
solid vertical line shows the median of the distribution, while the
dotted lines show the 68% confidence limits. (a) The relative like-
lihood of rE, the Einstein ring radius of the lens projected on the
source place. (b) The relative likelihood of the mass of the lens, as
inferred from the measured values of rE, the timescale tE, and the
standard Galactic model of Han & Gould (1995, 2003). (c) The rel-
ative likelihood of the period of the binary-source in days. (c) The
relative likelihood of the semi-amplitude K of the radial velocity
of the primary of the binary source, in kilometers per second.
This may be due to the fact that the lens and source
happen to have usually small relative velocities, it may
indicate that the event is due to disk-disk lensing, or it
may simply be because the model is incorrect.
We also infer an Einstein ring radius projected onto
observer plane of rE = 13.3+15.5
−7.0 AU, indicating that, for
the binary-source model, parallax effects are likely to be
negligible.
The probability distributions of several other interest-
ing quantities can also be constructed from the Markov
chain. The semi-major axis of the binary is d = 0.80 ±
0.06 AU. In order to determine the period, we assume
8
MICROLENSING EVENT OGLE-2002-BLG-055
a primary mass of m1 = M⊙, as expected for bulge gi-
ants. The absolute I-band magnitude of the secondary
is MI = 6.60 ± 0.46, which corresponds to a mass of
m2 = 0.67 ± 0.05 M⊙ assuming that it is on the main-
sequence. These corresponds to a late K-dwarf. The
primary has an absolute magnitude of MI = 2.01± 0.49,
making it likely to be a M1 ∼ M⊙ bulge giant. The total
binary mass is Ms ≡ m1 + m2 ≃ 1.7M⊙. The distribu-
tion of the binary period is shown in Figure 3, the me-
dian and 68% confidence interval is P = 200 ± 20 days.
The binary period is only a factor of ∼ 2 larger than
the microlensing event timescale tE. This implies that
the assumption of a uniform acceleration is almost cer-
tainly violated, and that the binary-source model is not
internally self-consistent. The constant acceleration ap-
proximation may not be so bad, as the deviations from
uniform acceleration will only become significant in the
tails of the event, when it is not highly magnified. Never-
theless, the parameters derived from the above analysis
should be interpreted with caution.
It is also possible to predict the semi-amplitude of the
radial velocity of the primary,
K = (cid:20) Gm2
dMs (cid:21)1/2
2
sin i.
(14)
The distribution of K is shown in Figure 3. The median
and 68% confidence interval is 14.0+2.8
−5.6 km s−1. Radial
velocity precisions of a few km s−1 should be attainable
for this source, which has I = 17.9. Thus it may be pos-
sible to directly confirm the binary-source interpretation
of this event.
6. CONCLUSION
Microlensing event OGLE-2002-BLG-055 exhibits a
slightly asymmetric, smooth light curve with a sin-
gle data point that deviates by ∼ 0.6 magnitudes
from neighboring points separated by several days.
Jaroszy´nski & Paczy´nski (2002) concluded that the sim-
plest interpretation of OGLE-2002-BLG-055 was an
event with parallax deviations arising from the motion of
the Earth, and a short-timescale deviation due to a bi-
nary lens with mass ratio q = 10−3− 10−2, thereby mak-
ing OGLE-2002-BLG-055 a candidate planetary event.
Here we demonstrated that OGLE-2002-BLG-055 can
be reasonably well-fit by several different classes of mod-
els, and a wide range of parameters within each model
class. We found that the data can be fit by many dif-
ferent binary-lens models whose mass ratios span three
orders of magnitude, from q = 10−4 to 10−1, thereby
making the secondary consistent with a planet, brown
dwarf, or M-dwarf for reasonable primary masses. A
subset of these binary-lens fits form a family of continu-
ously degenerate models, whose mass ratios differ by an
order of magnitude. Astonishingly, the light curves of
Although the primary goal
these models differ by . 2% for the majority of their du-
ration. A binary-source model is also consistent with the
data, for a secondary/primary flux ratio of ∼ 1%. This
model also naturally explains the global asymmetry of
the lightcurve as due to the acceleration of the primary
induced by the secondary. Under the assumption of a
bound binary-source, this model yields an estimate of
the Einstein ring radius projected on the source plane of
rE = 1.87 ± 0.40 AU.
All of these fits differ by . 3σ, and are essentially
indistinguishable when the scatter due to likely source
variability is considered. Unfortunately, the lack of color
information during the event precludes the discrimina-
tion of models based on the positions of the source and
blend on a color-magnitude diagram.
of
study by
Jaroszy´nski & Paczy´nski (2002) was to affect a modifi-
cation of the OGLE observation strategy to ensure good
sampling of short-duration perturbations, rather than ar-
gue that OGLE-2002-BLG-055 was a bona fide planetary
event, it is still somewhat disturbing that many binary-
lens fits were missed, and the possibility of a binary-
source interpretation was not discussed at all. OGLE-
2002-BLG-055 serves as an extreme reminder of the de-
generacies inherent in microlensing events, and high-
lights the difficulties in interpreting poorly-sampled and
weakly-perturbed events. These difficulties become es-
pecially important when attempting to detect planets
with microlensing. Here observers and modelers need
to be especially vigilant: in order to produce convincing
and reliable planetary detections, it is essential not only
to achieve dense and accurate photometry of planetary
perturbations, but also to acquire as much auxiliary in-
formation as possible, and to perform detailed, careful,
and thorough modeling, in order to ensure that planetary
detections are robust.
the
We would like to thank the anonymous referee for use-
ful comments and suggestions, and the OGLE collabo-
ration for making its microlensing data publically avail-
able. This work was supported by a Menzel Fellowship
from the Harvard College Observatory, by the National
Science Foundation under Grant No. PHY99-07949, by
the Astrophysical Research Center for the Structure and
Evolution of the Cosmos (ARCSEC) of the Korean Sci-
ence & Engineering Foundation (KOSEF), through the
Science Research (SRC) program, and by JPL contract
1226901. B.S.G. would like to acknowledge the hospi-
tality of the staff and participants of the workshop on
"Planet Formation: Terrestrial and Extra Solar," at the
Kavli Institute for Theoretical Physics, University of Cal-
ifornia, Santa Barbara, where the majority of this work
was completed.
REFERENCES
Afonso, C., et al. 2000, ApJ, 532, 340
Albrow, M. D., et al. 1999, ApJ, 522, 1022
Albrow, M. D. et al. 2000, ApJ, 534, 894
Albrow, M. D., et al. 2002, ApJ, 572, 1031
Bennett, D. P. et al. 1999, Nature, 402, 57
Bertelli, G., Bressan, A., Chiosi, C., Fagotto, F., & Nasi, E. 1994,
A&AS, 106, 275
Collinge, M.J. 2004 (astro-ph/0402385)
DePoy, D. L. et al. 2004, ApJ, submitted (astro-ph/0401250)
Dominik, M. 1998, A&A, 333, 893
Dominik, M. 1999, A&A, 349, 108
Gaudi, B. S. 1998, ApJ, 506, 533
Gaudi, B. S. & Gould, A. 1997, ApJ, 486, 85
Gaudi, B. S. et al. 2002, ApJ, 566, 463
Gould, A. 1992, ApJ, 392, 442
Gould, A. 1996, ApJ, 470, 201
GAUDI & HAN
9
Gould, A. 2003, ApJ, submitted (astro-ph/0311548)
Gould, A. & Loeb, A. 1992, ApJ, 396, 104
Gould, A., Miralda-Escud´e, J., & Bahcall, J. N. 1994, ApJ, 423,
L105
Griest, K. & Hu, W. 1992, ApJ, 397, 362
Griest, K., Safizadeh, N. 1998, ApJ, 500, 37
Han, C. 2002, ApJ, 564, 1015
Han, C. & Gould, A. 1995, ApJ, 447, 53
Han, C. & Gould, A. 1997, ApJ, 480, 196
Han, C. & Gould, A. 2003, ApJ, 592, 172
Han, C. & Jeong, Y. 1998, MNRAS, 301, 231
Jaroszy´nski, M. & Mao, S. 2001, MNRAS, 325, 1546
Jaroszy´nski, M. & Paczy´nski, B. 2002, Acta Astronomica, 52, 361
Mao, S. & Di Stefano, R. 1995, ApJ, 440, 22
Mao, S. & Paczy´nski, B. 1991, ApJ, 374, L37
Paczy´nski, B. 1986, ApJ, 304, 1
Popowski, P., Cook, K. H., & Becker, A. C. 2003, AJ, 126, 2910
Press, W. H., Teukolsky, S. A., Vetterling, W. T., & Flannery, B. P.
1992, Numerical Recipes(Cambridge: Cambridge University
Press)
Smith, M. C. et al. 2002, MNRAS, 336, 670
Smith, M. C., Mao, S., & Paczy´nski, B. 2003, MNRAS, 339, 925
Stanek, K. Z. 1996, ApJ, 460, L37
Sumi, T. 2004, MNRAS, 349, 193
Udalski, A. 2003a, ApJ, 590, 284
Udalski, A. 2003b, Acta Astronomica, 53, 291
van Belle, G. T. 1999, PASP, 111, 1515
Verde, L. et al. 2003, ApJS, 148, 195
Wo´zniak, P. R. & Stanek, K. Z. 1996, ApJ, 464, 233
Tab e 1. de Fi aaee
de
A
A
0
0
x
y
E
b
D 245000.
day
S.SS.CV.1
2401.670.22 0.340.03 684 17:28 0:12 17:78 0:22
{
{
{
{
{
{
179.2/100
S.SS.CV.3
2401.610.24 0.330.03 704 17:33 0:13 17:70 0:20
{
{
{
{
{
{
166.5/98
S.SS.CA.1. 2400.570.28 0.290.05 70
17:55
17:45
0:17
0:75
{
{
{
{
148.5/98
13
0:30
0:29
0:04
0:18
8
0:25
0:22
0:05
0:30
20
0:31
0:31
0:12
0:68
S.SS.CA.1. 2400.570.29 0.290.06 92
17:52
17:48
0:40
1:28
{
{
{
{
148.5/98
14
0:25
0:23
0:20
0:42
21
0:40
0:08
0:05
0:18
S.SS.CA.3. 2400.200.33 0.220.04 82
17:86
17:22
0:22
1:04
{
{
{
{
132.8/96
6
0:14
0:17
0:10
0:57
34
0:40
0:20
0:21
1:25
S.SS.CA.3. 2400.200.33 0.220:06 112
17:84
17:24
0:58
1:89
{
{
{
{
132.8/96
21
0:28
0:18
0:35
0:70
S.BS.CA. 2400.000.32 0.230.05 82
17:89
17:22
0:21
0:98
0:011
2425:09
0:001
{
141.7/96
11
0:28
0:60
0:06
0:41
0:004
0:45
0:004
17
0:32
0:19
0:05
0:26
0:005
0:07
0:002
S.BS.CA. 2400.000.35 0.210.05 115
17:97
17:18
0:55
1:93
0:009
2425:00
0:002
{
141.9/96
25
0:38
0:55
0:24
0:77
0:004
0:36
0:004
27
0:28
0:25
0:22
0:95
0:003
0:16
0:001
(cid:15)
{
0;2
0;2
B.SS.CA.a
2401.17
0.21
121.0
17.95
17.18
0.43
2.07
1.19
4:83 10
0.928
4:1 10
136.8/96
B.SS.CA.b
2401.58
0.44
64.9
16.88
19.07
0.19
0.59
1.36
2:77 10
0.942
6:6 10
133.1/96
3
3
B.SS.CA.
2403.99
0.38
56.3
16.98
18.51
0.09
0.54
1.52
3:75 10
0.744
6:3 10
130.9/96
2
3
B.SS.CA.d
2400.92
0.41
54.7
17.00
18.43
0.12
0.44
1.34
1:81 10
0.693
6:3 10
133.5/96
3
3
B.SS.CA.e
2401.22
0.33
82.6
17.32
17.71
0.26
0.97
1.28
1:51 10
0.943
5:4 10
135.3/96
3
3
B.SS.CA.f
2401.24
0.36
59.4
17.20
17.90
0.13
0.52
1.30
2:03 10
0.703
5:7 10
135.1/96
3
3
d
(cid:11)
(cid:26)
4
3
e. S=ig e e B=biay e SS=ig e
e BS=biay
e CV=
a ve
iy CA=
a a
e eai 1=e daa i
eved 3=hee daa i eved =egaive
a a
e eai (cid:12) =iive
a a
e eai (cid:12) ab
def =biay e de hw i
Fig e 2.
1
0
I
M
C
R
O
L
E
N
S
I
N
G
E
V
E
N
T
O
G
L
E
-
2
0
0
2
-
B
L
G
-
0
5
5
|
astro-ph/0409278 | 1 | 0409 | 2004-09-13T08:31:01 | Cosmology with the Square Kilometre Array | [
"astro-ph"
] | We argue that the Square Kilometre Array has the potential to make both redshift (HI) surveys and radio continuum surveys that will revolutionize cosmological studies, provided that it has sufficient instantaneous field-of-view that these surveys can cover a hemisphere in a timescale ~1 yr. Adopting this assumption, we focus on two key experiments which will yield fundamental new measurements in cosmology, characterizing the properties of the mysterious dark energy which dominates the dynamics of today's Universe. Experiment I will map out ~10^9 HI galaxies to redshift z~1.5, providing the premier measurement of the clustering power spectrum of galaxies: accurately delineating the acoustic oscillations and the `turnover'. Experiment II will quantify the cosmic shear distortion of ~10^10 radio continuum sources, determining a precise power spectrum of the dark matter, and its growth as a function of cosmic epoch. We contrast the performance of the SKA in precision cosmology with that of other facilities which will, probably or possibly, be available on a similar timescale. We conclude that data from the SKA will yield transformational science as the direct result of four key features: (i) the immense cosmic volumes probed, exceeding future optical redshift surveys by more than an order of magnitude; (ii) well-controlled systematic effects such as the narrow `k-space window function' for Experiment I and the accurately-known `point-spread function' for Experiment II; (iii) the ability to measure with high precision large-scale modes in the clustering power spectra, for which nuisance effects such as non-linear structure growth, peculiar velocities and `galaxy bias' are minimised; and (iv) different degeneracies between key parameters to those which are inherent in the CMB. | astro-ph | astro-ph | 4 Cosmology with the SKA
0
0
2
C.A. Blakea, F.B. Abdallab, S.L. Bridlec, S. Rawlingsb
p
e
S
3
1
1
v
8
7
2
9
0
4
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
aSchool of Physics, University of New South Wales, Sydney, NSW 2052, Australia
bAstrophysics, Department of Physics, University of Oxford, Keble Road, Oxford, OX1 3RH, UK
cDepartment of Physics and Astronomy, University College London, London, WC1E 6BT, UK
We argue that the Square Kilometre Array has the potential to make both redshift (HI) surveys and radio
continuum surveys that will revolutionize cosmological studies, provided that it has sufficient instantaneous field-
of-view that these surveys can cover a hemisphere (fsky ∼ 0.5) in a timescale ∼ 1 yr. Adopting this assumption,
we focus on two key experiments which will yield fundamental new measurements in cosmology, characterizing
the properties of the mysterious dark energy which dominates the dynamics of today's Universe. Experiment I
will map out ∼ 109(fsky/0.5) HI galaxies to redshift z ≈ 1.5, providing the premier measurement of the clustering
power spectrum of galaxies: accurately delineating the acoustic oscillations and the 'turnover'. Experiment II will
quantify the cosmic shear distortion of ∼ 1010(fsky/0.5) radio continuum sources, determining a precise power
spectrum of the dark matter, and its growth as a function of cosmic epoch. We contrast the performance of
the SKA in precision cosmology with that of other facilities which will, probably or possibly, be available on a
similar timescale. We conclude that data from the SKA will yield transformational science as the direct result
of four key features: (i) the immense cosmic volumes probed, exceeding future optical redshift surveys by more
than an order of magnitude; (ii) well-controlled systematic effects such as the narrow 'k-space window function'
for Experiment I and the accurately-known 'point-spread function' (synthesized beam) for Experiment II; (iii)
the ability to measure with high precision large-scale modes in the clustering power spectra, for which nuisance
effects such as non-linear structure growth, peculiar velocities and 'galaxy bias' are minimised; and (iv) different
degeneracies between key parameters to those which are inherent in the Cosmic Microwave Background.
1. Background
Over the last few years we have entered an
'era of precision cosmology' in which rough esti-
mates of many of the key cosmological parameters
have been replaced by what can reasonably be de-
scribed as 'measurements' (parameters known to
better than ∼ 10 per cent accuracy). Remarkable
progress in observations of the temperature fluc-
tuations in the Cosmic Microwave Background
(CMB) has been central to this scientific transfor-
mation (e.g. the recent results from the WMAP
satellite, Spergel et al. 2003).
The current
standard cosmological model
('ΛCDM') assumes that the Universe is composed
of baryons, cold dark matter (CDM) and Ein-
stein's cosmological constant Λ. The present-day
densities of baryons, (baryons + CDM) and Λ, as
a fraction of the critical density, are denoted by
Ωb, Ωm and ΩΛ, respectively. The local rate of
cosmic expansion is described by the Hubble pa-
rameter h = H0/(100 km s−1 Mpc−1). Structures
in the Universe are assumed to grow by gravita-
tional amplification of a primordial 'power spec-
trum of fluctuations' created by inflation, which
is taken to be a 'scale free' power law Pprim(k) ∝
k nscalar .1
Within this framework, we begin by reviewing
what the CMB tells us about the cosmological
model. The basic CMB observable is the angu-
lar power spectrum of temperature fluctuations,
which consists of a series of 'acoustic peaks'. We
note that the physics of the CMB fluctuations (i.e.
1A power law function is 'scale free' because it contains no
preferred length scale. The power law index nscalar = 1
is described as 'scale invariant' because in this case, the
matter distribution has the same degree of inhomogeneity
on every resolution scale (is 'self-similar').
Inflationary
theories of the early Universe predict that this should be
approximately true.
1
2
the photon-baryon fluid driven by the dark mat-
ter potential wells) depends entirely on the phys-
ical densities of matter and of baryons, Ωmh2 and
Ωbh2.2 However, the projection of the anisotropy
spectrum onto the sky (i.e. the angular locations
of the acoustic peaks) also involves the angular
diameter distance to the last-scattering surface.
• The angular location θA of the first 'acous-
tic peak' of the temperature power spec-
trum turns out to be relatively insensitive
to all parameters apart from the curva-
ture of the Universe. The global geome-
try is close to spatially flat, although pos-
itive spatial curvatures (Ωk ≈ −0.05) and
hence closed-Universe models remain con-
sistent with all available data (Efstathiou
2003).
It is important to note that exact
spatial flatness, typically assumed, remains
a theoretical prior which demands observa-
tional confirmation.
• The physical matter density (Ωmh2 ≈ 0.14)
is set by the ratio of the first acoustic peak
height to the large-scale amplitude of the
temperature power spectrum.
• The physical baryon density (Ωbh2 ≈ 0.024)
is fixed by the relative heights of the first
and second acoustic peaks (and is nicely
in accord with the value derived from
nucleosynthesis-based arguments).
• The values of Ωmh2 and Ωbh2 together de-
termine the sound horizon s at recombina-
tion, the characteristic physical scale fix-
ing the spatial position of the first acous-
tic peak. Using the angular location θA
of this peak, we can derive the angular di-
ameter distance DA = s/θA to the surface
of last scattering (whose redshift z ≈ 1100
is known from thermal physics). Assum-
ing a flat Universe and the ΛCDM model,
DA(z = 1100) depends on a combination
2The factor h2 arises because these physical densities are
proportional to the critical density of the Universe, which
scales as H 2
0 . The existence today of non-zero curvature
or of a cosmological constant Λ is insignificant at recom-
bination.
of Ωm and h. Together with the value of
Ωmh2 known from the first peak height, this
breaks the degeneracy between Ωm and h,
yielding Ωm ≈ 0.3 and h ≈ 0.7.
• The overall normalization of the tempera-
ture power spectrum (i.e. the fluctuation
strength at z = 1100) can be related to
the amplitude of mass fluctuations at red-
shift zero, conventionally defined as the rms
mass perturbation in spheres of radius 8 h−1
Mpc assuming linear evolution3 (and de-
noted σ8). In a ΛCDM model, if the matter
densities and Hubble constant are known,
then this quantity is still degenerate with
the optical depth to the last-scattering sur-
face (τ ). The detection of polarization in
the CMB by the WMAP experiment begins
to break this degeneracy by obtaining the
value τ ≈ 0.17 (hence σ8 ≈ 0.9).
• Changing the slope of the primordial power
spectrum of mass fluctuations (nscalar) in-
duces a broad-band tilt in the CMB temper-
ature power spectrum, that may be quanti-
fied by e.g. the ratio of the first and third
acoustic peaks (note that the quantity Ωbh2
only controls the ratio of odd-to-even peak
heights). Current data favours a roughly
scale-invariant primordial fluctuation spec-
trum (nscalar ≈ 1).
In summary, according to this crude argu-
ment, the CMB yields six independent quanti-
ties (assuming polarization measurements). For a
flat ΛCDM model, the WMAP experiment alone
measured values of (Ωm, Ωb, h, σ8, τ , nscalar) to
5 − 10 per cent accuracy (see Spergel et al. 2003).
Similar values for some of these parameters are
yielded by several independent, non-CMB tech-
niques, for example:
3When the amplitudes of density fluctuations are small
(when the overdensity δ ≪ 1), the equations of struc-
ture formation may be linearized using perturbation the-
ory. Gravitational growth is said to be in the linear regime
and may be solved analytically with relative ease. Eventu-
ally, the fluctuation amplitudes will become too high and
linear theory will break down.
• Observations of Cepheid variables and sec-
ondary distance indicators to measure h
(Freedman et al. 2001).
a different form of dark energy, such as
'quintessence' ? What are the properties of
this dark energy?
3
• Searches for high-redshift supernovae to
measure ΩΛ (Riess et al. 1998; Perlmutter
et al. 1999).
• Measurements of the abundance of rich
clusters to determine σ8 (e.g. Bahcall &
Bode 2003).
• Measurement of the amplitude of cosmic
shear (weak lensing due to large-scale struc-
ture) to constrain σ8 (e.g. Brown et al. 2003;
Hoekstra, Yee & Gladders 2002; Jarvis et al.
2003; Bacon et al. 2003; Pen et al. 2003).
The consistency of these results with those ob-
tained from the CMB demonstrates that -- if
the standard cosmological picture is correct --
these methods are not obviously compromised
by systematic errors at the current level of accu-
racy. The next-generation CMB experiment, the
Planck satellite, is planned to launch in 2007 and
will further refine CMB measurements of these
six parameters and second-order effects via (i)
drastically-improved sensitivity on small scales
and (ii) an enhanced polarization capability.4
However, if we wish to relax any of the assump-
tions of our most basic (flat ΛCDM) model then,
as we will describe, the degeneracies inherent in
the CMB become insuperable and extra informa-
tion is required (Efstathiou & Bond 1999; Bri-
dle et al. 2003). Just such a course of action is
demanded by today's most pressing cosmological
questions:
• Is the current rate of cosmic expansion ac-
celerating, as suggested by recent supernova
measurements?
• If so, does Einstein's cosmological constant
(vacuum energy) exist, despite our inabil-
ity to understand its value theoretically?
Or, is the accelerating expansion driven by
4See http://www.astro.esa.estec.nl/∼Planck.
The
temperature power spectrum measured by WMAP is al-
ready 'cosmic variance limited' (i.e. cannot be improved)
on large scales up to and including the first acoustic peak.
• Can competing models of inflation be dis-
criminated by accurate measurements of
the shape of the primordial power spectrum
(e.g. by a 'running spectral index' instead of
nscalar)?
• Can the tensor (gravitational wave) compo-
nent be isolated?
• Is the Universe exactly flat or does it possess
a small spatial curvature Ωk?
In this paper we focus our discussion upon the
quest to characterize the mysterious dark energy,
which drives the accelerating cosmic expansion
and contributes the majority of the present-day
energy density of the Universe. As an explana-
tion of the dark energy, Einstein's cosmological
constant Λ is consistent with current observa-
tions but poses tremendous theoretical challenges.
The 'natural' (quantum-mechanical) value of Λ
is a factor 10120 higher than that measured
by astronomers. This severe difficulty has mo-
tivated the development of alternative models,
many featuring a dynamic component of dark en-
ergy whose properties evolve with time, such as
quintessence. These competing models are essen-
tially untested and require experiments able to
measure the properties of dark energy as a func-
tion of cosmic epoch. The dark energy model is
commonly characterized by its equation-of-state
w(z) = P/ρ relating its pressure P to its energy
density ρ (in units where the speed-of-light c = 1).
For Einstein's cosmological constant, w(z) = −1.
The resolution of this problem is the current
cosmological frontier, and has been widely identi-
fied as having profound importance for physics as
a whole. Needless to say, if there were a paradigm
shift in the next decade that alleviated the dark
energy problem, then the SKA experiments de-
scribed below would be even more important in
confirming and better understanding the new cos-
mological model.
Can the CMB inform us about dark energy?
Although in the standard (ΛCDM) model dark
4
energy has a negligible effect at recombination,
the CMB is in fact sensitive to the equation-of-
state of dark energy because the angular position-
ing of the acoustic peaks depends on the angular
diameter distance to z = 1100, which is affected
by w(z) (Huey et al. 1999). However, the efficacy
of this probe of w(z) is limited because:
• Using CMB data alone, there is an imme-
diate degeneracy between w(z) and Ωm to
produce the same value of DA(z = 1100).
• Even given accurate external information
about Ωm and h, the quantity DA(z =
1100) can still only constrain a weighted av-
erage of w(z), i.e. one degree of freedom.
• The precision with which this one degree of
freedom can be measured is further limited
because w(z) only influences dynamics at
low redshift, thus its influence upon DA(z =
1100) is small.
• This lack of sensitivity implies that we can
make small adjustments to the value of the
sound horizon s at fixed Ωm by changing
the mix of baryons and CDM (i.e. Ωb) to
leave unchanged the angular locations of
the CMB acoustic peaks, which depend on
s/DA(z = 1100).
This difficulty in measuring a non-Einstein
dark energy equation-of-state w(z) using the
CMB alone, even if assumed to be a constant
wcons 6= −1, is illustrated by Figure 1, in which we
simulate the degeneracy between Ωm and wcons
after Planck data has been analyzed. Even with
the addition of a very accurate independent mea-
surement of h, there remains a ±0.1 scatter in
the value of wcons. We note that changing wcons
from −1 to −1.1 alters the value of DA(z = 1100)
by less than 1%. This is because the co-moving
distance to a given redshift is determined by the
integral
x(z) = Z z
dx
dz ′ dz ′ = Z z
c
H(z ′)
dz ′
0
0
(c/H0) dz ′
pΩm(1 + z ′)3 + (1 − Ωm)(1 + z ′)3(1+wcons)
= Z z
0
to which dark energy only contributes at very low
redshifts (z <
∼ 1). For fixed Ωm and h, the sound
horizon can be changed by a similar fraction by
varying Ωbh2 within its uncertainty range after
the Planck experiment.
The prospects for using the CMB to map out
any variations in the dark energy equation-of-
state with cosmic time, the key observation re-
quired to discriminate quintessence cosmologies,
are even worse. Even if the exact value of Ωm
were known, the CMB primary anisotropies only
measure a weighted average of w(z) (Saini, Pad-
manabhan & Bridle 2003). For example, in a
simple model w(z) = w0 + w1z, we can al-
ways scale w1 with w0 in such a way to keep
DA(z = 1100) fixed and leave the power spec-
trum of the primary anisotropies invariant (al-
though there will be second-order changes, such
as Sunyaev-Zeldovich decrements due to galaxy
clusters, and the late-time Integrated Sachs-Wolfe
effect).
We conclude that, due to degeneracies intrinsic
to the CMB, the outstanding questions of the era
of precision cosmology demand additional experi-
ments of comparable power, with different inher-
ent parameter degeneracies. As discussed above,
the CMB can only weakly constrain the dark en-
ergy model. An accurate measurement of w(z)
requires extremely precise cosmological data at
lower redshifts, where dark energy significantly
affects the dynamics of the Universe; for exam-
ple, per-cent level determinations of the distances
to redshift slices at z ∼ 1.5 What cosmological
probes can provide such precision?
• Observations of distant Type Ia supernovae
as 'standard candles' track the luminosity
distance as a function of redshift and pro-
vide an 'orthogonal constraint' to the CMB.
Supernova surveys will be a powerful probe
of the dark energy model, but we argue that
it is not yet proven that the experimental
systematics can be controlled to sufficient
precision to disentangle the effects of dark
5Beyond z ≈ 1.5, dark energy is expected to have a negli-
gible effect on dynamics. This assertion should be tested;
but if true, distance determinations to z ≈ 3 have lim-
ited power for constraining w(z), and measurements of
the Hubble constant at z ≈ 3 have no such power.
5
−0.4
−0.6
−0.8
w
−1
−1.2
−1.4
−1.6
0.15
0.2
0.25
0.3
0.35
Ω
m
0.4
0.45
0.5
0.55
0.6
Figure 1. Simulated joint constraints (68%, 95%) on the matter density Ωm and a constant dark energy equation-
of-state w(z) = wcons, resulting from Planck observations of the CMB. We use TT, TE and EE simulated Planck
data and only retain points with ℓ > 50 to account for uncertainty in any dark energy perturbations, which
may manifest themselves on very large scales and are strongly dependent on the exact dark energy model. The
strong degeneracy in the (Ωm, wcons) plane arises because the CMB power spectra are approximately invariant
for fixed values of (1) Ωmh2 and (2) the angular diameter distance to the last-scattering surface, DA(z = 1100),
which fixes the angular location of the first acoustic peak. DA(z = 1100) may be held constant by varying Ωm
simultaneously with wcons; the combination Ωmh2 is kept fixed by also varying h. The inclusion of an additional
tight Gaussian prior on the Hubble parameter (h = 0.72 ± 0.02) breaks the main degeneracy, resulting in the
tighter inner contours. However, there remains a secondary degeneracy between the value of wcons and the physical
baryon density Ωbh2. This latter quantity can be varied to change the size of the sound horizon and compensate
for changes in DA(z = 1100) due to wcons, keeping the angular peak locations fixed. The result is that, even if
Ωm and h are known precisely, Planck data cannot determine wcons to a precision better than ±0.1.
6
energy at the required accuracy (see Section
5).
• Redshift surveys of 105 − 106 galaxies in
the local Universe provide an independent
measure of Ωmh via the shape of the galaxy
power spectrum (if nscalar, Ωb, the neutrino
masses and any running spectral index are
known). We argue that the efficacy of this
cosmological probe is fundamentally limited
by uncertainties regarding galaxy bias. In
any case, Figure 1 demonstrates that the
CMB cannot produce a very accurate mea-
surement of dark energy even with the ad-
dition of extra data fixing Ωm and/or h.
In our opinion, the critical additional informa-
tion required to answer the profound questions
posed in this Section will be provided by experi-
ments measuring:
• Acoustic oscillatory features in the cluster-
ing power spectrum (Experiment I, Section
3.2).
• Weak gravitational lensing by large-scale
'cosmic
structure, otherwise known as
shear' (Experiment II, Section 4).
In this paper we argue that radio surveys with
the SKA will provide the ultimate databases for
these experiments, assuming that the telescope is
designed with a wide enough field-of-view.
2. Assumptions of this paper
2.1. SKA capabilities
We adopt in this paper the SKA capabili-
ties given by the current 'Strawman Science Re-
quirements' (Jones 2004) with one key exception:
the instantaneous (1.4 GHz) field-of-view (F OV )
must exceed the Strawman value (F OV ∼ 1 deg2)
by more than an order of magnitude for our pro-
posed experiments.
Abdalla & Rawlings (2004; hereafter AR04)
have considered this requirement in more detail,
concluding that F OV and β (the ratio of the in-
stantaneous bandwidth to the frequency range
0.5 − 1.4 GHz required by Experiment I) must
obey the relation (β/0.25) × F OV > 40 deg2 for
the survey to have a reasonable (∼ 1 yr) duration.
We emphasise that F OV is the field-of-view ob-
tained at 1.4 GHz (HI at z = 0) and that the cal-
culations of AR04 assume that this field-of-view
scales with redshift (∝ ν−2 for the purposes of
this study, where ν is the frequency of redshifted
HI). AR04 also note the necessity for at least half
the collecting area to reside in a core of diame-
ter ∼ 5 km for Experiment I, and that Experi-
ment II requires significant collecting area in the
longer baselines to deliver the sharp angular res-
olution (at 1.4 GHz) necessary for measuring cos-
mic shear. Both of these requirements are already
in the 'Strawman' design.
Using these SKA specifications, and assuming
that the field of view (in units of deg2) scales as
ν−2, the effective sensitivity of a 'tiled' survey6 is
HI ≈ 6 × 109 M⊙ galaxy
sufficient to detect an M ⋆
with a signal-to-noise ratio of 5 out to redshift
z ∼ 1.5 (assuming no evolution in the break of
the HI mass function; see Sec. 2 of Rawlings et
al., this volume).
In this paper we have not attempted to model
those aspects of our proposed surveys which will
be very dependent on the precise SKA design,
such as how the usable F OV scales with wave-
length for Experiment I, or what dynamic range
is achievable for Experiment II; we note that in
practice these limitations might be extremely sig-
nificant. We will also ignore restrictions on 'all-
sky' surveys due to finite processing resources
which -- even given likely technological advances
over the coming decade -- may still prove impor-
tant, given the huge datasets that will be gener-
ated.
2.2. Other assumptions
For the remainder of this paper we assume, un-
less stated otherwise, a fiducial model consisting
of a spatially-flat (Ωk = 0) Universe with cosmo-
logical parameters h = 0.7, Ωm = 0.3, Ωb/Ωm =
0.15, σ8 = 1 and nscalar = 1 (broadly consis-
tent with the recent WMAP results, Spergel et
al. 2003). We adopt a fiducial dark energy model
characterized by an equation-of-state w(z) =
6In which pointings just overlap at 1.4 GHz, and each
receives a 4-hour integration so that the effective exposure
time scales as (ν/1.4 GHz)−2.
w0 + w1z where (w0, w1) = (−0.9, 0), in order
to illustrate the accuracy with which the SKA
will be able to discriminate a general dark energy
model from a cosmological constant model, for
which (w0, w1) = (−1, 0). Unless stated other-
wise, we neglect contributions from tensors (pri-
mordial gravitational waves) and massive neutri-
nos. We assume that structure in the Universe is
seeded by adiabatic Gaussian fluctuations.7
3. Experiment I: HI emission line surveys
The SKA will enable revolutionary progress in
the field of large-scale structure. In particular, it
will map out the cosmic distribution of neutral
hydrogen by detecting the HI 21cm transition at
cosmological distances that are almost entirely in-
accessible to current instrumentation. Once an
HI emission galaxy has been located on the sky,
the observed wavelength of the emission line au-
tomatically provides an accurate redshift, locat-
ing the object's position in the three-dimensional
cosmic web.
We simulated the co-moving space densities of
HI-emitting galaxies as a function of redshift by
extrapolating the observed local HI mass function
using model 'C' of AR04. We note that, although
an integral property of the HI mass function is
constrained by observations of damped-Lyα sys-
tems, considerable uncertainty remains concern-
ing the redshift-evolution of its break. However,
reasonable models indicate that the SKA, if de-
signed with a sufficiently wide field-of-view, can
survey the entire visible sky in a year of opera-
tion, locating ∼ 109 HI emission galaxies over a
volume stretching to redshift z ≈ 1.5 with suf-
ficient number density that clustering statistics
are limited by cosmic variance, not by shot noise.
In this case, the precision with which the galaxy
clustering pattern can be measured is limited en-
tirely by the cosmic volume mapped: the SKA be-
comes the premier large-scale structure machine.
In comparison, planned next-generation optical
redshift surveys (e.g. the proposed KAOS spec-
7An SKA redshift survey would provide a useful probe of
non-Gaussianity by detecting large numbers of superclus-
ters, which correspond to high-peak fluctuations in the pri-
mordial density field evolving into the quasi-linear regime.
7
trograph; http://www.noao.edu/kaos) will only
cover ∼ 1000 deg2 at z ∼ 1, whereas all-sky
photometric redshift surveys (e.g. Pan-STARRS;
http://panstarrs.ifa.hawaii.edu) inevitably
result in a serious degradation of radial power
spectrum modes due to the relatively large red-
shift error.8
3.1. Galaxy power spectrum
The resulting HI redshift survey yields an im-
provement in power spectrum precision of be-
tween one and two orders of magnitude compared
to the current state-of-the-art (Figure 2), delin-
eating features such as the acoustic oscillations
(see Section 3.2). In addition, the SKA clustering
map traces clustering modes on large scales be-
yond the power spectrum turnover (i.e. on scales
k < 0.02 h Mpc−1) that are entirely inaccessi-
ble to local redshift surveys; together with extra
small-scale modes (to k ≈ 0.2 h Mpc−1) that are
hidden by non-linear clustering at redshift zero.
What cosmological information can we glean?
The galaxy clustering pattern encodes a wealth
of data about the underlying constituents of the
Universe and the basic gravitational processes
governing the formation of structure. In partic-
ular, the shape of the power spectrum depends
on a combination of cosmological parameters. In
Figure 3 we demonstrate that parameter con-
straints resulting from the SKA power spectrum
vastly improve upon those obtained using the
Sloan Digital Sky Survey (SDSS). Note in partic-
ular that, since the acoustic oscillations are de-
tected in the SKA power spectrum (see Section
3.2), the baryon fraction is strongly constrained,
whereas the SDSS contours are compatible with
zero baryons.
The precise determination of the power spec-
trum on large scales can be compared with the
predictions of inflationary models, testing for ef-
fects such as departures of the primordial power
spectrum Pprim(k) from a pure power-law. The
narrow window function in k-space (resulting
8Even with a very optimistic redshift error ∆z = 0.02,
a photometric redshift survey must cover ≈ 10 times the
area of a spectroscopic experiment to map out the same
number of Fourier modes in the linear regime (Blake &
Glazebrook 2003).
8
Figure 2. Simulated measurement of the galaxy clustering power spectrum by a 1-year SKA survey, compared
to the 2dFGRS 100k data release (Percival et al. 2001). The measurements have been normalized to the same
underlying power spectrum model, keeping the (1σ) fractional errors in each k-bin the same. For the SKA data,
we assume that the redshift evolution of clustering may be divided out so that results from different redshift slices
can be combined. We note that the 2dFGRS P (k) data points are heavily correlated in this binning, whereas
the SKA survey possesses a much narrower 'k-space window function' owing to the vastly greater cosmic volume
probed, implying that the resulting P (k) measurements are barely correlated. The SKA clustering data delineates
the power spectrum turnover on large scales (k < 0.02 h Mpc−1) and the acoustic oscillations (Section 3.2), as a
direct result of mapping a vastly greater cosmic volume V . The error in a power spectrum measurement in the
cosmic-variance-limited regime scales as 1/√V , and VSKA ≈ 500 V2dF.
9
Figure 3. Joint constraints on the cosmological parameters (Ωmh, fb = Ωb/Ωm) resulting from fitting linear
theory models (calculated using the CAMB software; Lewis, Challinor & Lasenby 2000) to the shape of the SKA
galaxy power spectrum (assuming a linear galaxy biasing scheme). Contours are shown for the analyzed Sloan
Digital Sky Survey (SDSS) data (Pope et al. 2004; 1σ, 2σ, 3σ) and for a simulated SKA HI emission-line survey
(inner contour; 2σ). We only use SKA clustering data in the linear regime (with k < 0.2 h Mpc−1). In order to
match the assumptions of the SDSS contours, we fix nscalar = 1 and marginalize over only the Hubble constant.
Whilst the SKA produces an exquisite delineation of the power spectrum, we argue that this Figure may be wildly
optimistic because the bias model cannot be assumed to be linear at the required level of precision.
10
from the vast cosmic volume probed by the SKA)
implies that any sharp features in Pprim(k) are
not smoothed out and may be detected in a man-
ner not possible using the CMB. It has been sug-
gested that the intriguing outliers in the CMB
temperature power spectrum measured by the
WMAP satellite (at ℓ ∼ 30 and ℓ ∼ 200) could
be due to unknown physics on scales smaller than
the Planck length, which are imprinted during
inflation (e.g. Martin & Ringeval 2004). The
SKA would delineate any such effects. Further-
more, comparison with large-scale CMB modes
can yield information about any tensor contribu-
tion to the CMB anisotropies, arising from grav-
ity waves in the early Universe.9 This is a sec-
ond potential means, together with pulsar obser-
vations, by which the SKA can be used as a grav-
itational wave detector.
However, we wish to argue that the appli-
cations of the SKA large-scale structure mea-
surements discussed above are potentially flawed
(and at best severely complicated) by our lack of
knowledge of the detailed biasing scheme linking
the observed galaxy distribution to the underly-
ing pattern of dark matter fluctuations produced
by theoretical models (e.g. Coles 2004). In gen-
eral, galaxy biasing is almost certainly stochastic,
scale-dependent, non-local and non-linear (Dekel
& Lahav 1999): the assumption of a simple lin-
ear bias, implicit in Figure 3, is likely to be inade-
quate when confronted with sub-percent precision
power spectrum data.
3.2. Acoustic oscillations
Given the likely complexities of galaxy bias, we
seek cosmological experiments that are less sensi-
tive to the biasing model. The power spectrum of
galaxies on large scales ( >
∼ 30 Mpc) should con-
tain a series of small-amplitude acoustic oscilla-
tions of identical physical origin to those seen
in the CMB.10 These features result from oscil-
9Assuming that such a tensor contribution may be disen-
tangled from any large-scale effects due to clustering of
dark energy, galaxy bias, and the potential presence of
'isocurvature' density fluctuations in the early Universe
(in addition to the usual adiabatic perturbations).
10The acoustic oscillations imprinted in the matter power
spectrum possess significantly lower amplitude than those
found in the CMB power spectrum, because the baryonic
lations of the photon-baryon fluid before recom-
bination, and encode a characteristic scale -- the
sound horizon at recombination -- accurately de-
termined by CMB observations as described in
Section 1. This scale can act as a standard ruler
(Eisenstein 2002). Its recovery from a galaxy red-
shift survey depends on the assumed cosmological
parameters, particularly the dark energy model,
and thus constrains that model over a range of
redshifts (Blake & Glazebrook 2003; Seo & Eisen-
stein 2003).
The application of this cosmological test does
not depend on the overall shape of the power
spectrum, which can be divided out using a
smooth fit, but only on the residual oscilla-
tory signature of the acoustic peaks. Hence the
method is insensitive to smooth broad-band tilts
in P (k) induced by such effects as a running spec-
tral index, redshift-space distortions, and com-
plex biasing schemes (e.g. based on haloes; Seljak
2000). It would be very surprising if the biasing
scheme introduced oscillatory features in k-space
liable to obscure the distinctive acoustic peaks
and troughs.
In the absence of major system-
atic effects, constraints on the dark energy model
are limited almost entirely by how much cosmic
volume one can survey, rendering this an ideal
experiment for the SKA, which can map the HI
galaxy distribution out to z ≈ 1.5 (where three
acoustic peaks lie in the linear regime).
In Figure 4 we display simulated power spectra
for a survey covering 20, 000 deg2, divided into
redshift slices. If the observer applies the wrong
dark energy model when converting HI redshifts
to cosmic distances, then the peaks and troughs
in the recovered P (k) lie in the wrong places: the
experiment becomes a powerful probe of dark en-
ergy.
The standard ruler provided by the acoustic
oscillations may be applied separately to the tan-
gential and radial components of the power spec-
trum, resulting in independent measurements of
the co-ordinate distance x(z) to each redshift slice
and its rate of change x′(z) ≡ dx/dz, respec-
component of the matter distribution, which contains the
oscillatory imprint, is sub-dominant to the CDM compo-
nent.
11
Figure 4. Simulated galaxy power spectra for an SKA survey of 20,000 deg2, analyzed in redshift slices of width
∆z = 0.2. Each power spectrum is divided by a smooth polynomial fit, revealing the sinusoidal imprint of acoustic
oscillations, and is shifted along the y-axis for clarity. As redshift increases, the linear regime extends to smaller
physical scales (larger values of k), unveiling more peaks. The acoustic oscillation 'wavelength' may be used as
a standard ruler. This is illustrated by the Figure: the true dark energy model is w = −1, but the observer
has incorrectly assumed w = −0.8 when constructing the power spectra, and consequently the recovered acoustic
scale disagrees with that observed in the CMB (represented by the vertical dotted lines).
12
tively.11 We fitted these simulated measurements
of x(z) and x′(z) over a series of redshift slices,
assuming a dark energy model w(z) = w0 + w1z.
Figure 5 illustrates the accuracy of recovery of
(w0, w1) assuming a fiducial cosmology (−0.9, 0),
demonstrating that in this case we can dismiss
a cosmological constant model (−1, 0) with high
significance. The accuracy of our prior knowl-
edge of Ωm and h affects the tightness of the dark
energy constraints. These parameters must be
known to standard deviations σ(Ωm) = 0.01 and
σ(h) = 0.01 if these uncertainties are not to domi-
nate the overall error in (w0, w1). These demands
are not unrealistic on the SKA timescale: SKA
observations of masers in the vicinity of black
holes will determine h to ∼ 1% accuracy (Green-
hill, these proceedings) and CMB data from the
Planck satellite will fix Ωmh2 to a similar preci-
sion (Balbi et al. 2003).
Given that the SKA will be unavailable until
∼ 2020, what other projects plan to survey the
acoustic peaks? Two leading contenders are the
multi-object near-infrared/optical spectrographs
FMOS and KAOS, due to commence operation
later this decade at the Subaru and Gemini tele-
scopes, respectively.12 These projects may per-
form redshift surveys of a few hundred deg2, de-
riving constraints on (w0, w1) that are several
times poorer than the 20, 000 deg2 SKA survey
presented above (the error bars roughly scale as
1/pfsky). In radio wavebands, SKA prototypes
may be developed such as the HYFAR proposal
(Bunton et al. 2003). The speed of surveying
z = 1 HI emission galaxies for such a prototype
would lag the SKA by roughly an order of magni-
tude; furthermore, the bandwidth may be signif-
icantly smaller than that envisaged for the SKA.
11The actual quantities determined by the experiment are
the ratio of x and x′ to the sound horizon s at recombina-
tion (which determines the characteristic spatial scale of
the acoustic oscillations). The value of s is determined by
CMB observations, via Ωmh2 and Ωbh2, but is affected to
some extent by uncertainities in Ωm and h (Figure 5). The
tangential component of P (k) measured in a survey slice
at redshift z determines x(z) because this latter quantity
governs the tangential scale of the recovered galaxy distri-
bution in that redshift slice: displacements ∆r are deter-
mined by ∆r = x ∆θ. The value of x′(z) fixes the radial
scale of the recovered distribution: ∆r = (dx/dz) ∆z.
12KAOS is still unfunded.
In summary: even if these precursor experi-
ments succeed in detecting a deviation from Ein-
stein's cosmological constant, the SKA will still
be required for a precise delineation of the new
model. The key advantage offered by a radio tele-
scope is that it may be designed with a field of
view exceeding an optical spectrograph by a fac-
tor of 100: the SKA can survey all the available
volume by studying the whole visible sky. More-
over, SKA sensitivity is necessary and sufficient
for detecting tracers of this volume out to z = 1.5.
4. Experiment II: radio continuum surveys
for cosmic shear
Light rays follow geodesics, which bend in the
presence of matter.
It follows that a coherent
shape distortion is imprinted in the distribution
of distant background galaxies by mass fluctu-
ations in the intervening cosmic web:
there is
a small tendency for galaxies to be tangentially
aligned. This pattern of cosmic shear encodes a
vast body of information and, in principle, is a re-
markably powerful cosmological probe. The shear
pattern is directly sensitive to the mass distribu-
tion predicted by the underlying theory, and does
not depend on the complex details of how galaxy
light traces mass. Furthermore, cosmic shear may
be used to measure dark energy: which controls
both the angular diameter distances intrinsic to
the lens equation (e.g. the source-lens distance)
and the redshift evolution of the mass fluctuations
producing the lensing.
A successful cosmic shear survey has three lead-
ing design requirements: very high image quality
for reliable shape measurements, a high source
surface density ( >
∼ 100 arcmin−2) to limit sta-
tistical noise, and a wide survey area to reduce
cosmic variance. Each of these demands is deliv-
ered superbly by an SKA radio continuum survey
(Schneider 1999).
The leading systematic for ground-based op-
tical cosmic shear experiments is the difficulty
in quantifying systematic variations in the point-
spread function, due to inevitable changes in the
atmospheric seeing and telescope properties with
position and time. The psf determination is ac-
complished by observing stars in the field (of
13
Figure 5. Constraints on a dark energy model w(z) = w0 + w1z achievable by an SKA survey of the acoustic
peaks, illustrated by sets of (68%, 95%) contours corresponding to different assumed priors. The solid contours
correspond to perfect knowledge Ωm = 0.3 and h = 0.7. For the dashed contours we impose Gaussian priors
with standard deviations σ(Ωm) = 0.01 and σ(h) = 0.01, which are not unrealistic in the SKA epoch. In each
of these two cases we are able to distinguish a fiducial model (w0, w1) = (−0.9, 0) from a cosmological constant
(marked by the solid circle). Weaker Gaussian priors σ(Ωm) = 0.03 and σ(h) = 0.05 (the dot-dashed contours)
are inadequate for this purpose. Note that there remains a residual degeneracy in the (w0, w1) plane because
at a given redshift, approximately the same cosmology can be obtained by increasing w0 and decreasing w1.
This degeneracy is partially broken by measurements across a range of redshift slices (and by the independent
determinations of x and x′ for each redshift).
14
which there are a limited number density).
In
comparison, the point-spread function of a ra-
dio telescope is well-determined and stable (being
simply derived from the interferometer baseline
distribution, i.e. the synthesized beam). In addi-
tion, the planned SKA angular resolution (≈ 0.05
arcsec at 1.4 GHz) vastly improves upon that ob-
tainable over wide fields from the ground.
cosmic
shear
Space-based optical
surveys
should also exist by the time the SKA is avail-
able. However, the planned area for these surveys
is ≈ 300 deg2 (a SNAP-like experiment; Massey
et al. 2004), which will be dwarfed by an SKA all-
sky continuum survey. The fraction of sky fsky
mapped is directly related to the precision of the
shear power spectrum measurement via a factor
1/pfsky.
An SKA continuum survey will achieve sensi-
tivities of ∼ 30 nJy in a 4-hour pointing, far ex-
ceeding depths attained by contemporary radio
surveys.
In order to simulate the results of an
SKA cosmic shear experiment, we must assume
a model for the radio source populations at these
unprecedented flux densities. We generated this
model from a combination of two populations:
starburst galaxies and quiescent spiral disks (we
ignore for now the Active Galactic Nuclei which
dominate at the highest flux densities but which
have too low surface density to contribute signif-
icantly to the weak lensing signal at SKA sen-
sitivities). We extrapolated local radio luminos-
ity functions for these two populations (Sadler et
al. 2002) as a function of redshift using a Press-
Schechter-based approach, and folded in a pre-
scription for the physical sizes of each population.
Full details of these models will be published else-
where (Abdalla et al., in prep.).
Using these population models, we simulated
radio skies as a function of SKA integration time
and angular resolution. Following the method
of Massey et al. (2004), we then employed stan-
dard shape-estimation software (Kaiser, Squires
& Broadhurst 1995) to recover the shear disper-
sion per galaxy arising from intrinsic shapes and
from measurement errors (σγ ) together with the
surface density of 'usable galaxies' (ng) after cuts
for objects that are unresolved or faint. The val-
ues achieved, σγ ≈ 0.2 and ng ∼ 500 arcmin−2,
are comparable to space-based optical surveys
(Massey et al. 2004) whilst covering a survey area
100 times larger. Figure 6 illustrates the measure-
ment accuracy of the cosmic shear angular power
spectrum using these data.13
The resulting measurements of the dark energy
model can be estimated using the Fisher matrix
methodology (e.g. Refregier et al. 2004). Figure 7
illustrates joint constraints on (w0, w1) using the
shear power spectrum up to ℓ = 105. The anal-
ysis requires models for both the source redshift
distribution (which can be readily inferred from a
deep HI emission-line survey of a sub-area of the
sky) and the linear theory mass power spectrum
(e.g. calculated using CMBFAST; Seljak & Zal-
darriaga 1996) corrected for non-linear effects in
accordance with the prescription of e.g. Smith et
al. (2003).
The power of this experiment may be increased
by splitting the lensed galaxies into two broad
redshift bins (Refregier et al. 2004). This is
readily,
if roughly, accomplished by separating
the radio continuum sources into two groups de-
termined by detection (or absence) in the HI
emission-line survey of Experiment I.
We note that a cosmic shear survey does not
necessarily require exquisitely-measured galaxy
shapes, because there is an irreducible scatter
resulting from the unknown ellipticity of each
galaxy before shear.14
Instead, the 0.05-arcsec
SKA resolution is required to (1) resolve starburst
galaxy disks at high redshift and (2) eliminate
radio source confusion. A potential residual sys-
tematic error may lie in intrinsic alignments of
galaxy disks (owing to structure formation pro-
cesses) which may masquerade as coherent shear
signal. For example, some starburst galaxies may
exist in pairs, forming merging disk systems. Tak-
ing an extremely cautious approach, we assume
for now that starburst galaxies would not be used
in the cosmic shear analysis. This reduces the sur-
face density by a factor of two, but the SKA still
yields a vast improvement in dark energy con-
13We note that it is also possible to analyze a radio cosmic
shear survey directly in the uv-plane (Chang & Refregier
2002).
14In the ideal cosmic shear experiment, the unsheared ob-
jects would all be circular.
15
10−2
10−3
π
2
/
kl
P
)
1
+
l
(
l
10−4
101
102
103
Multipole l
104
105
Figure 6. The cosmic shear angular power spectrum measured from a simulated SKA radio continuum survey.
The upper (red) curve is for a model with Ωm = 0.25 and wcons = −0.9; the lower (blue) curve is for Ωm = 0.3
and wcons = −1; the dotted lines show the ±1σ range for Ωm = 0.3 and wcons = −0.9. We assume that the survey
contains a surface density of usable galaxies ng = 200 arcmin−2 and covers one hemisphere (fsky = 0.5).
16
1
w
0.3
0.2
0.1
0
−0.1
−0.2
−1
−0.95
−0.9
w
0
−0.85
−0.8
Figure 7.
(68%, 95%, 99%) likelihood contours for a dark energy model w(z) = w0 + w1z from an SKA cosmic
shear survey over a hemisphere, assuming 200 usable galaxies per square arcminute. We marginalize over the
other cosmological parameters (Ωm, Ωb, h, ns). These contours are conservative because no redshift information
per galaxy has been included.
straints compared to future space-based surveys.
The shear angular power spectrum is the
most basic lensing experiment; we note that
the bi-spectrum (three-point statistic) promises
comparable, statistically-independent constrain-
ing power on small scales (Takada & Jain 2004).
A third independent approach is to target the
most massive clusters around which the imprint
of cosmic shear is most significant. Deep HI
follow-up can yield full background source red-
shift information, allowing us to cross-correlate
the cosmic shear signal
in redshift slices be-
hind the massive lens, yielding a purely geomet-
ric probe of cosmic distances ('cross-correlation
cosmography', e.g. Jain & Taylor 2003). This
method provides a probe of the angular diame-
ter distance as a function of redshift independent
of the acoustic oscillations in the clustering power
spectrum. However, a potential systematic error
is introduced by the presence of additional pro-
jected masses along the line-of-sight.
5. The SKA as the Dark Energy Machine
The current frontier of cosmological research is
the characterization of the properties of the 'dark
energy' that dominates the energy density of to-
day's Universe and drives its accelerating expan-
sion. In this paper we have focussed on two key
experiments -- acoustic oscillations (Experiment
I) and cosmic shear (Experiment II) -- by which a
wide-field SKA can generate measurements of the
dark energy model that are significantly more ac-
curate than any competition by ∼ 2020. Even if
precursor surveys detect evidence for a deviation
from Einstein's cosmological constant, the SKA
will be required for a precise delineation of the
new model.
In Figure 8 we compare overall SKA measure-
ments of a dark energy model w(z) = w0 + w1z
with (arguably) the most promising competitor:
the (currently unfunded) satellite mission SNAP,
planned to launch in ∼ 2014, whose basic goal is
to locate approximately two thousand supernovae
to z ≈ 1.7 (http://snap.lbl.gov; Aldering et
al. 2004).15
15We note that the planned SNAP mission also includes a
weak lensing component that will yield outstanding space-
17
Observations of distant Type Ia supernovae as
'standard candles' provided the first clear evi-
dence for the accelerating expansion of the Uni-
verse (Riess et al. 1998; Perlmutter et al. 1999).
Future high-redshift supernova searches will be a
powerful probe of the dark energy model. How-
ever, we argue that it is not yet proven that the
dimming effect of cosmic acceleration can be dis-
tinguished from other systematic effects with suf-
ficient precision to render this cosmological probe
as reliable as Experiments I and II above. Such
systematic effects include changes in the intrin-
sic properties of supernovae with galactic envi-
ronment (e.g. metallicity), dust extinction from
both the host galaxy and the Milky Way16, and
the difficulties of photometric calibration (super-
nova K-corrections, together with the inadequacy
of local 'standard stars' such as Vega for provid-
ing sub-percent-level flux calibration).
Many other probes of dark energy have been
suggested in the literature (see e.g. Gurvits, this
volume). We argue that, whilst these probes
will provide important cross-checks of the cosmo-
logical model, none can yield the raw precision
or control of systematic errors afforded by our
Experiments I and II. For example, using SKA
(or other) data we can perform number counts
of haloes as a function of circular velocity and
redshift to infer the cosmological volume element
(Newman & Davis 2000). But this experiment
requires an accurate theoretical prediction of the
number density distribution with circular veloc-
ity (i.e. a precise understanding of baryons) to-
gether with precisely-measured rotation curves at
high redshift. Alternatively, number counts of
galaxy clusters can be utilized (e.g. Weller, Battye
& Kneissl 2002). The main systematic difficulty
here is connecting the cluster observable (e.g. X-
ray luminosity or SZ decrement) to the halo mass
(upon which the cluster abundance is exponen-
based image fidelity. However, as noted in Section 4, an
SKA cosmic shear survey can cover a area ∼ 100 times
greater with comparable resolution.
16The recent measurement of several supernovae at z > 1
by the Hubble Space Telescope (Riess et al. 2004) has as-
suaged this doubt to some extent by detecting a transition
with increasing redshift from accelerating to decelerating
expansion, which would not occur if the inferred accelera-
tion was an artefact of dust extinction.
18
0.5
0.4
0.3
0.2
0.1
1
w
0
−0.1
−0.2
−0.3
−0.4
−0.5
SNAP SN
SNAP cosmic shear
SKA wiggles
SKA cosmic shear
SKA combined
−1.15 −1.1 −1.05
−1
−0.95 −0.9 −0.85 −0.8 −0.75 −0.7
w0
Figure 8. Comparison of constraints on the dark energy equation-of-state, parameterized as w(z) = w0 + w1z,
obtained by the proposed satellite mission SNAP (using both the supernova and cosmic shear experiments) and the
SKA (via acoustic oscillations and cosmic shear). The contours represent '±1σ' confidence intervals. A Gaussian
prior upon the matter density, σ(Ωm) = 0.01, was consistently applied to the supernova and acoustic peaks
analyses. For the latter we also assume a prior σ(h) = 0.01 for the Hubble parameter, which could be obtained
by measuring masers using the SKA. We assumed that the SNAP supernova component produces measurements
of the luminosity distance to an accuracy of 1.4% in each of 50 redshift bins between z = 0.1 and z = 2. The
cosmic shear contours for both SNAP and SKA are conservative because the sample has not been split into two
or more bands in redshift. This analysis method was shown by Refregier et al. (2003) to improve significantly the
constraints from SNAP cosmic shear, and would have a similar effect on the SKA measurements.
tially sensitive). The frequency of strong gravita-
tional lensing (the number of lensing arcs in the
vicinity of massive haloes) is also sensitive to dark
energy (Bartelmann et al. 2003), but in manner
very dependent on accurate simulations (see also
Koopmans, Browne & Jackson, this volume)DW.
We argue that these various experiments will pro-
vide insight into the complex processes of struc-
ture formation, rather than into the underlying
cosmological model.
Finally, if matter does not dominate the dy-
namics of the Universe, then gravitational poten-
tial wells begin to 'decay', inducing a net shift
in the energies of CMB photons passing through
these wells. This phenomenon is known as the
late-time Integrated Sachs Wolfe (ISW) effect.
Its influence on the CMB power spectrum is re-
stricted to large scales (ℓ < 10) where it is clouded
by cosmic variance. However, its presence induces
a correlation between CMB anisotropies and the
low-redshift matter distribution, which has been
detected using current radio surveys (e.g. Boughn
& Crittenden 2002) and could be probed further
with the SKA. Measurement of the late ISW ef-
fect is an important confirmation of the dark en-
ergy model, and is sensitive not just to the evolu-
tion of dark energy, but also to its clustering prop-
erties (e.g. Weller & Lewis 2003; Bean & Dore
2003).
We conclude that, owing to its capacity to
undertake cosmic surveys on an unprecedented
scale, the SKA will become the paramount tool
for answering the key questions of the era of preci-
sion cosmology. The SKA can perform clean and
powerful experiments to characterize dark energy,
mapping the acoustic oscillations in the cluster-
ing power spectrum and the patterns of cosmic
shear distortion in the radio sky. Furthermore,
SKA data can enlighten other pressing cosmolog-
ical questions, pursuing tests of inflationary theo-
ries and measuring spatial flatness with unprece-
dented precision.
REFERENCES
1. Abdalla F., Rawlings S., 2004, MNRAS, in
press
19
(astro-ph/0405232)
3. Bacon D.J., Massey R.J., Refregier A.R., Ellis
R.S., 2003, MNRAS, 344, 673
4. Bahcall N.A., Bode P., 2003, ApJ, 588L, 1
5. Balbi A., Baccigalupi C., Perrotta F., Matar-
rese S., Vittorio N., 2003, ApJ, 588L, 5
6. Bartelmann M., Meneghetti M., Perrotta F.,
Baccigalupi C., Moscardini L., 2003, A&A,
409, 449
7. Bean R., Dore O., 2003, NewAR, 47, 747
8. Blake C.A., Glazebrook K., 2003, ApJ, 594,
665
9. Boughn S.P., Crittenden R.G., 2002, PhRvL,
88b1302
10. Bridle S.L., Lahav O., Ostriker J.P., Stein-
hardt P.J., 2003, Science, 299, 1532
11. Brown M.L., Taylor A.N., Bacon D.J., Gray
M.E., Dye S., Meisenheimer K., Wolf C.,
2003, MNRAS, 341, 100
12. Bunton J.D. et al., 2003, "Panorama of the
Universe: A Cylindrical Reflector SKA" (see
www.skatelescope.org)
13. Chang T.-C., Refregier A., 2002, ApJ, 570,
447
14. Coles P., 2004,
in "The Scientific Promise
of the SKA", SKA Workshop, Oxford, eds
Kramer and Rawlings (astro-ph/0408440)
15. Dekel A., Lahav O., 1999, ApJ, 520, 24
16. Efstathiou G., 2003, MNRAS, 343L, 95
17. Efstathiou G., Bond J.R., 1999, MNRAS,
304, 75
18. Eisenstein D.J., 2002, "Large-scale struc-
ture and future surveys" in "Next Genera-
tion Wide-Field Multi-Object Spectroscopy",
ASP Conference Series vol. 280, ed. M.Brown
& A.Dey (astro-ph/0301623)
19. Freedman W.L., 2001, ApJ, 553, 47
20. Hoekstra H., Yee H.K.C., Gladders M.D.,
2002, ApJ, 577, 595
21. Huey G., Wang L., Dave R., Caldwell R.R.,
Steinhardt P.J., 1999, PhRvD, 59f3005
22. Jain B., Taylor A., 2003, PhRvL, 91n1302
23. Jarvis M., Bernstein G.M., Fischer P., Smith
D., Jain B., Tyson J.A., Wittman D., 2003,
AJ, 125, 1014
24. Jones D.L., 2004, SKA Memo 45
25. Kaiser N., Squires G., Broadhurst T., 1995,
2. Aldering
et
al.,
2004,
submitted
ApJ, 449, 460
20
26. Lewis A., Challinor A., Lasenby A., 2000,
ATNF for financial support.
ApJ, 538, 473
27. Martin J., Ringeval C., 2004, PhRvD, 69,
83515
28. Massey R. et al., 2004, AJ, 127, 3089
29. Newman J.A., Davis M., 2000, ApJ, 534L, 11
30. Pen U.-L., Lu T., van Waerbeke L., Mellier
Y., 2003, MNRAS, 346, 994
31. Percival W.J. et al., 2001, MNRAS, 327, 1297
32. Perlmutter S. et al., 1999, ApJ, 517, 565
33. Pope A. et al., 2004, ApJ, 607, 655
34. Refregier A. et al., 2004, AJ, 127, 3102
35. Riess A.G. et al., 1998, AJ, 116, 1009
36. Riess A.G. et al., 2004, ApJ, 600L, 163
37. Sadler E.M. et al., 2002, MNRAS, 329, 227
38. Saini T.D., Padmanabhan T., Bridle S.L.,
2003, MNRAS, 343, 533
39. Schneider P., 1999, "Weak gravitational lens-
ing with the SKA" in "Perspectives on Ra-
dio Astronomy, Scientific Imperatives at
cm and m wavelengths", Proceedings of a
workshop in Amsterdam, April 7-9 1999
(astro-ph/9907146)
40. Seljak U., 2000, MNRAS, 318, 203
41. Seljak U., Zaldarriaga M., 1996, ApJ, 469,
437
42. Seo H.-J., Eisenstein D.J., 2003, ApJ, 598,
720
43. Smith R.E. et al., 2003, MNRAS, 341, 1311
44. Spergel D.N. et al., 2003, ApJS, 148, 175
45. Takada M., Jain B., 2004, MNRAS, 348, 897
46. Weller J., Lewis A.M., 2003, MNRAS, 346,
987
47. Weller J., Battye R.A., Kneissl R., 2002,
PhRvL, 88w1301
Acknowledgments
CB acknowledges Karl Glazebrook for a valu-
able collaboration developing simulations of
acoustic oscillations. FA and SB thank Phil Mar-
shall and Richard Massey for useful discussions
about cosmic shear. CB thanks the SKA Project
Office for financial support. FA thanks PPARC
for a Gemini Research Studentship and the SKA
Project Office for financial support. SB acknowl-
edges the support of the Royal Society. SR thanks
PPARC for a Senior Research Fellowship and the
|
0808.4157 | 2 | 0808 | 2008-12-09T01:25:08 | No Evidence for Variability of Intervening Absorption Lines toward GRB 060206: Implications for the Mg II Incidence Problem | [
"astro-ph"
] | We examine variability of absorption line strength of intervening systems along the line of sight to GRB 060206 at $z = 4.05$, by the low-resolution optical spectra obtained by the Subaru telescope from six to ten hours after the burst. Strong variabilities of Fe\emissiontype{II} and Mg\emissiontype{II} lines at $z=1.48$ during $t=$5--8 hours have been reported for this GRB \citep{Hao07}, and this has been used to support the idea of clumpy Mg\emissiontype{II} cloudlets that was originally proposed to explain the anomalously high incidence of Mg\emissiontype{II} absorbers in GRB spectra compared with quasars . However, our spectra with higher signal-to-noise ratio do not show any evidence for variability in $t=$6-10 hours. There is a clear discrepancy between our data and Hao et al. data in the overlapping time interval. Furthermore, the line strengths in our data are in good agreement with those observed at $t \sim$ 2 hours by \citet{Tho08}. We also detected Fe\emissiontype{II} and Mg\emissiontype{II} absorption lines for a system at $z = 2.26$, and these lines do not show evidence for variability either. Therefore we conclude that there is no strong evidence for variability of intervening absorption lines toward GRB 060206, significantly weakening the support to the Mg\emissiontype{II} cloudlet hypothe sis by the GRB 060206 data. | astro-ph | astro-ph |
No Evidence for Variability of Intervening Absorption
Lines toward GRB 060206: Implications for the Mg II
Incidence Problem ∗
Kentaro Aoki,1 Tomonori Totani,2 Takashi Hattori,1 Kouji Ohta,2
Koji S. Kawabata,3 Naoto Kobayashi,4 Masanori Iye,5 Ken'ichi Nomoto,6
and
Nobuyuki Kawai7
1Subaru Telescope, National Astronomical Observatory of Japan,
650 North A'ohoku Place, Hilo, HI 96720, U.S.A.
2Department of Astronomy, Kyoto University,
Sakyo-ku, Kyoto 606-8502
3Hiroshima Astrophysical Science Center, Hiroshima University,
1-31-1 Kagamiyama, Higashi-Hiroshima, Hiroshima 739-8526
4Institute of Astronomy, University of Tokyo,
2-21-1, Osawa, Mitaka, Tokyo
5National Astronomical Observatory of Japan,
2-21-1, Osawa, Mitaka, Tokyo
6Institute for the Physics and Mathematics of the Universe,
University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8568
7Department of Physics, Tokyo Institute of Technology,
2-12-1 Ookayama, Meguro-ku, Tokyo 152-8551
(Received 2008 December 1; accepted )
Abstract
We examine variability of absorption line strength of intervening systems along the
line of sight to GRB 060206 at z = 4.05, by the low-resolution optical spectra obtained
by the Subaru telescope from six to ten hours after the burst. Strong variabilities
of Fe II and Mg II lines at z = 1.48 during t =5 -- 8 hours have been reported for this
GRB (Hao et al. 2007), and this has been used to support the idea of clumpy Mg II
cloudlets that was originally proposed to explain the anomalously high incidence of
Mg II absorbers in GRB spectra compared with quasars. However, our spectra with
higher signal-to-noise ratio do not show any evidence for variability in t =6-10 hours.
There is a clear discrepancy between our data and Hao et al. data in the overlapping
time interval. Furthermore, the line strengths in our data are in good agreement with
1
those observed at t ∼ 2 hours by Thone et al. (2008). We also detected Fe II and
Mg II absorption lines for a system at z = 2.26, and these lines do not show evidence
for variability either. Therefore we conclude that there is no strong evidence for
variability of intervening absorption lines toward GRB 060206, significantly weakening
the support to the Mg II cloudlet hypothesis by the GRB 060206 data.
Key words: gamma rays: bursts -- galaxies: ISM
1.
Introduction
The gamma-ray burst (GRB) has opened up a new window to probe high redshift uni-
verse, now reaching comparable redshifts with the studies by galaxies and quasars, and starting
to probe the era of the cosmic reionization (Kawai et al. 2006). The optical GRB afterglow
spectra give us a lot of information about GRB host galaxies (Mirabal et al. 2002; Savaglio,
Fall, and Fiore 2003; Prochaska et al. 2007; Chen et al. 2007), intervening absorption systems
along the line of sight (Prochter et al. 2006; Sudilovsky et al. 2007; Tejos et al. 2007) and the
reionization of the universe (Totani et al. 2006).
GRB 060206 is the 10th highest redshift GRB as of 2008 November. GRB 060206 was
triggered by the Swif t Burst Alert Telescope on 2006 February 6 at 04:46:53 UT. The redshift
of the burst was determined to be 4.048 (Fynbo et al. 2006). The intervening absorption lines
are reported at z = 1.48 (Fynbo et al. 2006; Hao et al. 2007; Thone et al. 2008) and z =2.26
(Aoki et al. 2006; this work). Surprisingly, Hao et al. (2007) claimed strong variation of the
equivalent widths (EWs) of Fe II λ2600.2 and Mg II λ2803.5 at z=1.48; the EW of the Fe II
λ2600.2 system at z = 1.48 showed a more than 80% decrease within 1.5 hours duration. It is
difficult to interpret this system to be located in the GRB host galaxy, because a large redshift
difference requires relativistic motion and hence a large velocity dispersion of lines is expected,
while the absorption line widths are only 14 − 20 km s−1 (Thone et al. 2008). Then, almost
unique possible interpretation is that the intervening absorption system is smaller than the
image size of the GRB afterglow, and the line strength changes with the size evolution of the
GRB afterglow. This requires the size of the absorption system to be less than ∼ 1016 cm (Hao
et al. 2007).
This variability, if real, would change the standard picture of intervening absorption
systems found in bright cosmological objects such as quasars and GRBs. Furthermore, this
result has been used to support the Mg II cloudlet hypothesis that has been proposed to explain
the anomalous Mg II incidence along the GRB sight lines (Frank et al. 2007). Prochter et al.
(2006) reported that the incidence of Mg II absorption lines along GRB sight lines is about four
∗ Based on data collected at Subaru Telescope, which is operated by the National Astronomical Observatory
of Japan.
2
times higher than that for quasars, at a statistical significance greater than 99.9% confidence
limit, and this cannot easily be explained by known selection effects. If the intervening line
variability of GRB 060206 is in fact due to the small size of the Mg II cloudlets, the anomalous
incidence of Mg II systems may also be explained by the larger image size of quasars than GRB
afterglows.
However, GRB 060206 is currently the only one case for the intervening line variability,
and it should be examined carefully. Here we report the analysis of the optical afterglow spectra
of GRB 060206 taken by the Subaru telescope during t = 6.01 -- 9.73 hours after the burst, a
part of which is overlapping with the Hao et al.'s observation (t = 4.13 -- 7.63 hours). We will
examine especially the line variability of the intervening Mg II and Fe II systems. In addition to
the z = 1.48 system, we will give detailed analysis of the z = 2.26 system as well and examine
the variability.
When we completed most of this paper, the revised version of Thone et al.
(2008,
arXiv:0708.3448v2) appeared on the preprint server, in which they also analyzed the same
data from the Subaru archive and reached a similar conclusion of no evidence for variability.
However, details of the Subaru observation and data analysis are not given in their paper.
Here we give more detailed descriptions about the observation logs and the processing of the
Subaru data taken by ourselves, as well as the implications for the Mg II incidence problem.
The z = 2.26 system was originally reported by Aoki et al. (2006) and it is mentioned also in
Fynbo et al. (2006), but Thone et al. (2008) concluded that this is not real. However, we will
show that this system is real, based on the detection of seven Fe II lines and Mg II doublet.
2. Observations and Data Reduction
The afterglow of GRB 060206 was observed on 2006 February 6 (UT) with FOCAS
(Kashikawa et al. 2004) attached to the Subaru 8.2-m telescope (Iye et al. 2004). We obtained
eight spectra of 30 minutes integration for each at 10:47, 11:19, 11:51, 12:23, 12:54, 13:28,
13:59, and 14:31 (UT). All these times are midpoint of exposures. We used the R300 grism
with two different order-cut filters for the observations. The first three and last two spectra
were taken with the O58 filter which covers between 5700 A and 10200 A. The other three
spectra were taken with the Y47 filter which allows an uncontaminated spectrum between 4800
A and 9200 A. The slit width was set to be 0.′′8. The atmospheric dispersion corrector was used
and the slit position angle was 176◦. The spectrophotometric standard star Hz 44 was observed
for sensitivity calibration and removal of atmospheric absorption lines.
It was photometric
condition with good seeing (0.′′6 - 0.′′9 in R−band). The seeing size became smaller as the
GRB was rising, and it was smaller than the slit width at the final epoch of the observations.
Consequently, the resolution of the GRB's spectra was changed from 8.0A to 6.8 A (290 km
s−1 to 250 km s−1 at 8190 A), which is confirmed by measuring absorption lines of a bright star
found in the same slit for the GRB afterglow.
3
The data were reduced using IRAF1 for the procedures of bias subtraction, flat-fielding,
wavelength calibration, and sky subtraction. Wavelength calibration was performed using OH
night sky emission lines, and the rms wavelength calibration error is 0.2 - 0.3 A. The sensi-
tivity calibration was performed as a function of wavelength, and the atmospheric absorption
feature was removed by using the spectrum of Hz 44. Fringe pattern exists beyond 7700 A in
the spectra, and fringe spectra were made from the standard star's spectra normalized by its
continuum. We divided the afterglow spectra by the fringe spectra, and suppressed the fringe
pattern. The initial strength of fringe pattern is 6 -- 11% of the continuum level of the afterglow
spectra, but it is reduced by a factor of ∼2 by this process. The foreground Galactic extinction
of AB = 0.054 mag (Schlegel, Finkbeiner, and Davis 1998) was corrected.
3. Results
Figure 1 displays the optical spectrum of GRB 060206 taken 9.7 hours after the burst.
The absorption feature between 9300 and 9600 A is residuals of the telluric absorption removal.
The spectrum clearly shows the z = 4.05 and 1.48 absorption line systems which have been
reported by Fynbo et al. (2006), Hao et al. (2007), and Thone et al. (2008). Furthermore, the
z = 2.26 absorption line system of Mg II is clearly recognized. We also detected seven Fe II
resonance lines (λ2249.9,2260.8,2344.3,2374.5,2382.8,2586.7, and 2600.2) at z = 2.26. All these
z = 2.26 absorption lines are beyond 7200 A and hence outside of the coverage of the WHT/ISIS
observation (Fynbo et al. 2006; Thone et al. 2008), explaining that Thone et al. (2008) could
not find the absorption system at z = 2.26. We list all detected lines in table 3 in Appendix.
In figure 11 in Appendix, we show the spectrum combined all 4 hours data.
We show Fe II λλ2586.7, 2600.2 absorption line profiles at z = 1.48 in our normalized
spectra in Figure 2. No significant temporal variation is seen between 6.0 and 9.7 hours after
the burst. The absorption lines at 9.73 hours after the burst are narrower than those at other
epochs, but this is due to the change of the spectral resolution; note that seeing size is smaller
than the slit width at this epoch, as mentioned in Section 2. We also show Mg II λλ2796.4,2803.5
absorption lines at z = 1.48 in Figure 3. No significant temporal variation is seen, either. We
measured EW in the observer's frame for these lines by fitting each absorption line with a single
Gaussian. The Gaussian fits are shown in Figure 2 and 3, and the EW results are tabulated in
table 1. Figure 4 presents the time evolution of EW of these lines, and we confirm that there is
no significant variation in the strength of all the four z = 1.48 absorption lines. The difference
between the first and last observations is 0.14 A at maximum (Mg II λλ2796.4), and that is
smaller than 1σ uncertainty. The EW measurements of Mg II λ2803.5, Fe II λ2600.2, and Fe II
λ2586.7 in the same Subaru data performed by Thone et al. (2008) are mostly consistent with
1
IRAF is distributed by the National Optical Astronomy Observatory, which is operated by the Association
of Universities for Research in Astronomy (AURA), Inc., under cooperative agreement with the National
Science Foundation.
4
our measurements. However, we find that the EW of Mg II λ2796.4 in Thone et al. (2008)
is significantly smaller than our measurement. The reason for this discrepancy is not clear,
although the conclusions of no evidence for variability are the same.
The absorption line profiles of Fe II λλ2586.7, 2600.2, Mg II λλ2796.4, 2803.5, Fe II
λλ2249.9, 2260.8, Fe II λ2344.2, and Fe II λλ2374.5, 2382.8, for another intervening absorption
line system at z = 2.26 are shown in Figure 5, 6, 7, 8 and 9, respectively. The absorption line at
8439 A is Al II at z = 4.05, and those at 7710 A and 7748 A are Si II λ1526.7 and Si II* λ1533.4
at z = 4.05, respectively. The equivalent widths are measured in the same way as the system
at z = 1.48, and the results are listed in table 2. We do not find any systematic change of
equivalent width in these all absorption lines at z = 2.26 but one, either (Figure 10). The EW
of Fe II λ2382.8 changed more than 3σ between 6.0542 hours and 7.065 hours. Since the oscil-
lator strength of Fe II λ2374.5 (3.13 × 10−2) is smaller than that of Fe II λ2382.8 (3.20 × 10−1),
Fe II λ2374.5 is less saturated, and thus should be more variable than Fe II λ2382.8. However,
the EW of Fe II λ2374.5 was stable within 1σ. As seen in Figure 9, Fe II λ2382.8 was heavily
blended with Si II* λ1533.4 at z = 4.05. This may cause larger uncertainty in EW, and large
variation in Fe II λ2382.8 is doubtful. Note that Mg II λ2803.5 is systematically stronger than
Mg II λ2796.4 in Figure 7, although the strength of Mg II λ2803.5 should be the same (optically
thick case) as, or weaker (optically thin case) than, Mg II λ2796.4. We noticed Si II λ1808.01 at
z = 4.05 is probably overlapped with Mg II λ2803.5 at 9130.5 A. Using the column density of
Si II (1.7 × 1015) from Thone et al. (2008), we estimated EW of Si II λ1808.01 to be 0.56 A in
observers' frame (optically thin case). We confirm that Si II λ1808.01 absorption line will be
optically thin with the column density, the Doppler parameter b of 20 kms−1 and the small
oscillator strength (2.08 × 10−3). We consider that the contamination of Si II λ1808.01 is likely
the cause of a systematic increase of Mg II λ2803.5 in this result.
4. Discussion and Conclusion
In addition to the Subaru data, the equivalent width measurements of the z = 1.48
system by Hao et al. (2007) and the WHT/ISIS data of Thone et al. (2008) are shown in Figure
4. Our early four spectra are overlapping in time with the last three observations by Hao et
al. (2007). Our EW measurement of z = 1.48 Fe II λ2600.2 at t = 6.011 hours is clearly in
disagreement with the value measured by Hao et al. (2007) at t = 6.15 hours. It should be
noted that Fe II λ2586.7 at z = 1.48 is clearly detected at 6415 A in our spectra (Figure 2)
with an equivalent width of ∼ 0.55 A. On the other hand, it is not mentioned by Hao et al.
(2007), and in fact, this line cannot be clearly seen in the Hao et al.'s spectra. This indicates
that an EW of <
∼ 0.5 A cannot reliably be measured by the Hao et al.'s observation, and hence
the claim of the variability of Fe II λ2600.2 line should suffer from a large uncertainty.
The Hao et al.'s spectra show a strong variability of Mg II lines around t = 5 − 5.5 hours,
and we cannot examine this variability because this epoch is not covered by our data. However,
5
the early measurements of EWs of these lines at z = 1.48 by Thone et al. (2008) at t ∼ 2 hours
are in good agreement with our measurements. It seems rather unlikely that the lines show a
variability only in the time duration of the Hao et al.'s observation and no variability at all
in the earlier or later data with higher signal-to-noise ratio. Furthermore, there is no evidence
for time evolution of EWs in another absorption system at z = 2.26. From these results, we
conclude that the variability reported by Hao et al. (2007) is likely due to some artificial effects
or statistical fluke in low signal-to-noise data.
Our result implies that the data of GRB 060206 no longer provide a strong support
to the Mg II cloudlet hypothesis proposed by Frank et al. (2007) for the anomalously high
incidence of Mg II systems in GRB spectra. Furthermore, several difficulties of this hypothesis
have been pointed out. One such argument is that apparently unsaturated Mg II absorption
lines with the doublet ratio of 1:1 are never found in intervening absorption lines, although
such lines are expected if optically thick Mg II cloudlets partially cover quasar beams (Prochter
et al. 2006; Porciani, Viel, and Lilly 2007; Sudilovsky et al. 2007). Another argument is that
there is no apparent change in Mg II incidence over the continuum and broad emission lines of
quasar spectra, although the broad line regions of quasars are much larger than the continuum
emitting region (Pontzen et al. 2007). It should also be noted that the suggested size of the
Mg II cloudlets is several orders of magnitude smaller than those derived from the observations
of gravitationally lensed quasars (Rauch et al. 2002; Ellison et al. 2004).
Therefore, we consider that the Mg II incidence anomaly is likely to be explained by some
other effects. Although the anomaly has not yet been solved, it could be explained, or at least
the statistical significance of the anomaly is reduced, by a combination of some selection effects
such as dust extinction and gravitational lensing (Porciani, Viel, and Lilly 2007; Sudilovsky et
al. 2007).
We are grateful to the Subaru Telescope staff for their assistance during our observations.
This work was partly supported by the Grant-in-Aid for Scientific Research on Priority Areas
(19047003) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT)
of Japan.
Appendix. Absorption Lines List
We identified absorption lines in the spectrum combined all 4 hrs data (Figure 11), but
1.5 hrs integration between 4800 A and 6000 A, and 2.5 hrs between 8800 A and 10200 A. The
identifications, observed wavelengths in vacuum, and equivalent widths in rest frame are listed
in Table 3. Most absorption lines redward of 7000 A in GRB 060206 are reported here at the
first time
6
References
Aoki, K., Hattori, T., Kawabata, K. S., & Kawai, N. 2006, GCN Circular, 4703
Chen, H.-W., Prochaska, J. X., Ramirez-Ruiz, E., Bloom, J. S., Dessauges-Zavadsky, M., & Foley, R.
2007, ApJ, 663, 420
Ellison, S. L., Ibata, R., Pettini, M., Lewis, G. F., Aracil, B., Petitjean, P. & Srianand, R. 2004, A&A,
414, 79
Frank, S., Bentz, M. C., Stanek, K. Z., Mathur, S., Dietrich, M., Peterson, B. M., & Atlee, D. W.
2007, Ap&SS, 312, 325
Fynbo, J. P. U., et al. 2006, A&A, 451, L47
Hao, H., et al. 2007, ApJ, 659, L99
Iye, M., et al. 2004, PASJ, 56, 381
Kashikawa, N., et al. 2002, PASJ, 54, 819
Kawai, N. et al. 2006, Nature, 440, 184
Mirabal, N. et al. 2002, ApJ, 578, 818
Pontzen, A., Hewett, P., Carswell, R., & Wild, V. 2007, MNRAS, 381, L99
Porciani, C., Viel, M., and Lilly, S. J. 2007, ApJ, 659, 218
Prochaska, J. X., Chen, H-W, Dessauges-Zavadsky, M., & Bloom, J. S. 2007, ApJ, 666, 267
Prochter, G. E. et al. 2006, ApJL, 648, L93
Rauch, M., Sargent, W. L. W., Barlow, T. A., & Simcoe, R. A. 2002, ApJ, 576, 45
Savaglio, S., Fall, S. M., & Fiore, F. 2003, ApJ, 585, 638
Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525
Sudilovsky, V., Savaglio, S., Vreeswijk, P., Ledoux, C., Smette, A., & Greiner, J. 2007, ApJ, 669, 741
Tejos, N., Lopez, S., Prochaska, J. X., Chen, H.-W.,& Dessauges-Zavadsky, M. 2007, ApJ, 671, 622
Thone, C. C., et al. 2008, A&A, 489, 37
Totani, T., Kawai, N., Kosugi, G., Aoki, K., Yamada, T., Iye, M., Ohta, K., & Hattori, T. PASJ, 58,
485
7
Fig. 1. The observed spectrum of GRB 060206 afterglow in vacuum wavelength. This spectrum was taken
at 9.7 hours after the burst. Red, cyan, and blue dashed lines indicate absorption lines at z = 4.05, 1.48
and 2.26, respectively. The hatched region between 9300 and 9600 A indicates the place of water vapor
absorption lines in the atmosphere. The spectrum at the bottom is the noise spectrum.
8
Fig. 2. Spectra of GRB 060206 afterglow around Fe II λλ2586.7, 2600.2 at z = 1.48. All spectra are
normalized by the continuum, and all absorption lines are fitted with a Gaussian.
9
Fig. 3. The same as Fig. 2, but for Mg II λλ2796.4, 2803.5 at z = 1.48.
10
Fig. 4. Equivalent width (observer's frame) of the z = 1.48 absorption lines versus observation time after
the burst. The EWs of Fe II λλ2586.7, 2600.2, and Mg II λλ2796.4, 2803.5 are shown with the symbols
indicated in the figure. Our measurements are indicated by filled red symbols between 6 hours and 10
hours, and Hao et al. (2007) data by open ones between 4 hours and 8 hours. The filled black symbols at
2 hours are from Thone et al. (2008).
11
Fig. 5. The same as Fig. 2, but for Fe II λλ2586.7, 2600.2 at z = 2.26. The absorption line at 8439 A is
Al II at z = 4.05.
12
Fig. 6. The same as Fig. 2, but for Mg II λλ2796.4, 2803.5 at z = 2.26.
13
Fig. 7. The same as Fig. 2, but for Fe II λλ2249.9, 2260.8 at z = 2.26.
14
Fig. 8. The same as Fig. 2, but for Fe II λ2344.2 at z = 2.26.
15
Fig. 9. The same as Fig. 2, but for Fe II λλ2374.5,2382.8 at z = 2.26. The absorption lines at 7710 A and
7748 A are Si II λ1526.7 and Si II* λ1533.4 at z = 4.05, respectively.
16
Fig. 10. Equivalent widths (observer's frame) of the z = 2.26 Mg II λ2796.4 (filled square), Mg II λ2803.5
(filled triangle), Fe II λ2600.2 (filled star), Fe II λ2586.7 (filled circle), Fe II λ2249.9 (open square), Fe II
λ2260.8 (open triangle), Fe II λ2344.2 (open star), Fe II λ2374.5 (open circle), and Fe II λ2382.8 (open
diamond) absorption lines, as functions of time after the burst.
17
Fig. 11. The observed spectrum of GRB 060206 afterglow in vacuum wavelength. This high signal --
to-noise ratio spectrum was made by combining all data. The hatched region between 9280 and 9600
A indicates the place of water vapor absorption lines in the atmosphere. Red, green, blue and cyan
dashed lines indicate absorption lines at z = 4.05, 3.55, 2.26, and 1.48, respectively. The unidentified lines
are indicated by orange dashed lines. See identifications in Table 3.
18
Table 1. Equivalent width (observer's frame) of the z = 1.48 Absorption Lines.
Hour EWFeII2600 EWFeII2586 EWMgII2796 EWMgII2803
6.011
1.30 ± 0.22
0.62 ± 0.23
2.43 ± 0.18
1.88 ± 0.18
6.542
1.29 ± 0.22
0.60 ± 0.20
2.37 ± 0.19
2.04 ± 0.19
7.065
1.09 ± 0.21
0.55 ± 0.24
2.09 ± 0.18
1.86 ± 0.20
7.598
1.24 ± 0.25
0.38 ± 0.20
2.17 ± 0.19
1.90 ± 0.20
8.120
1.27 ± 0.22
0.87 ± 0.24
2.20 ± 0.17
1.83 ± 0.18
8.679
1.01 ± 0.21
0.57 ± 0.20
2.34 ± 0.18
1.90 ± 0.20
9.209
1.11 ± 0.24
0.53 ± 0.20
2.25 ± 0.19
1.87 ± 0.21
9.732
1.18 ± 0.20
0.55 ± 0.20
2.29 ± 0.17
2.00 ± 0.19
19
Table 2. Equivalent width (observer's frame) of the z = 2.26 Absorption Lines.
Hour EWFeII2600 EWFeII2586 EWMgII2796 EWMgII2803
6.011
3.09 ± 0.25
3.15 ± 0.28
4.72 ± 0.23
5.11 ± 0.22
6.542
3.28 ± 0.26
2.72 ± 0.24
4.73 ± 0.22
5.07 ± 0.22
7.065
3.24 ± 0.28
3.13 ± 0.30
4.99 ± 0.26
5.47 ± 0.24
7.598
3.24 ± 0.29
2.99 ± 0.27
4.80 ± 0.25
5.17 ± 0.24
8.120
3.06 ± 0.27
2.70 ± 0.23
4.84 ± 0.23
4.85 ± 0.21
8.679
3.28 ± 0.28
2.73 ± 0.25
4.83 ± 0.25
4.84 ± 0.22
9.209
3.15 ± 0.30
2.84 ± 0.28
4.74 ± 0.26
5.04 ± 0.24
9.732
2.99 ± 0.27
2.46 ± 0.22
4.64 ± 0.23
4.92 ± 0.21
Hour EWFeII2249 EWFeII2260
EWFeII2344
EWFeII2374
EWFeII2382
6.011
0.44 ± 0.24
0.74 ± 0.25
2.0 ± 0.14
1.9 ± 0.28
2.9 ± 0.27
6.542
0.37 ± 0.17
0.48 ± 0.18
2.1 ± 0.14
1.7 ± 0.24
2.7 ± 0.21
7.065
0.45 ± 0.27
0.57 ± 0.26
1.9 ± 0.14
1.6 ± 0.30
1.9 ± 0.32
7.598
0.41 ± 0.19
0.34 ± 0.23
2.2 ± 0.16
1.8 ± 0.21
2.0 ± 0.20
8.120
0.40 ± 0.15
0.52 ± 0.24
2.1 ± 0.13
1.6 ± 0.24
2.1 ± 0.22
8.679
0.20 ± 0.11
0.43 ± 0.16
2.3 ± 0.14
2.0 ± 0.22
2.6 ± 0.27
9.209
0.24 ± 0.16
0.60 ± 0.20
2.4 ± 0.10
1.8 ± 0.24
2.4 ± 0.24
9.732
0.36 ± 0.19
0.36 ± 0.14
2.3 ± 0.13
1.9 ± 0.23
2.5 ± 0.26
20
Wavelength (A)
5527.8
6138.1
6259.9
6280.4
6336.5
6364.3
6388.1
6415.1
6447.8
6576.5
6589.0
6614.0
6694.8
6712.3
6740.8
6935.0
6952.4
7040.8
7055.6
7078.2
7087.7
7105.7
7325.0
7360.7
7632.3
7710.3
7730.6
7748.0
7758.0
7823.3
7835.8
8123.2
8389.5
8421.4
Table 3. Line Identification
Line
H I 1215.67
H I 1215.67
N V 1238.82
N V 1242.80
Si IV 1393.76
Si II 1260.42
z
Wr (A)
Ref.
3.5471
4.0492
--
--
4.0531
0.093 ± 0.015
4.0534
0.063 ± 0.017
3.5463
0.46 ± 0.048
4.0494
1.6 ± 0.017
Si II* 1264.74, 1265.00
4.0504
1.1 ± 0.016
1
2
3
3
1
3
3
3
3
3
3
3
Fe II 2586.65
Fe II 2600.17
O I 1302.17
1.4801
0.20 ± 0.027
1.4798
0.48 ± 0.033
1, 3
4.0504
1.5 ± 0.023
O I* 1304.86, 1306.03, Si II 1304.37
4.0487
1.1 ± 0.028
Si II* 1309.28
4.0516
0.20 ± 0.012
?
?
--
--
--
--
C II 1334.53, C II* 1335.66, 1335.71
4.0482
2.2 ± 0.019
Mg II 2796.35
Mg II 2803.53
Si IV 1393.76
C IV 1550.77
?
1.4800
0.93 ± 0.031
1.4799
0.77 ± 0.031
1, 3
1, 3
4.0517
0.79 ± 0.017
3.5497
0.092 ± 0.011
--
--
Si IV 1402.77
4.0527
0.34 ± 0.051
?
Fe II 2249.88
Fe II 2260.78
Fe II 2344.21
Si II 1526.71
Fe II 2374.46
--
--
2.2557
0.10 ± 0.0089
2.2558
0.14 ± 0.012
2.2558
0.68 ± 0.023
4.0503
1.2 ± 0.024
2.2557
0.55 ± 0.029
Si II* 1533.43
4.0527
0.63 ± 0.047
Fe II 2382.77
C IV 1548.20
C IV 1550.77
Fe II 1608.45
Mn II 2576.88
Fe II 2586.65
2.2559
0.74 ± 0.061
4.0532
0.69 ± 0.15
4.0528
0.51 ± 0.016
4.0503
0.59 ± 0.012
2.2557
0.17 ± 0.024
2.2557
0.89 ± 0.023
1: Hao et al. 2007, 2: Fynbo et al. 2006, 3: Thone et al. 2008
21
Wavelength (A)
8438.6
8464.9
8486.8
9103.2
9127.0
Table 3. (Continued.)
Line
Al II 1670.79
Fe II 2600.17
Mn II 2606.46
Mg II 2796.35
Mg II 2803.53
z
Wr (A)
Ref.
4.0507
1.4 ± 0.017
2.2555
0.98 ± 0.020
2.2561
0.083 ± 0.018
2.2554
1.5 ± 0.029
2.2555
1.6 ± 0.029
22
|
astro-ph/0305213 | 1 | 0305 | 2003-05-12T22:39:30 | Analysis of the Interaction Effects in the Southern Galaxy Pair Tol1238-364 and ESO381-G009 | [
"astro-ph"
] | In the context of the connection among galaxy-galaxy interaction, starbursts and nuclear activity, we present and discuss a quantitative morphological analysis based on BVR images and a detailed spectroscopic investigation of two interacting galaxies, the Seyfert 2 Tol1238-364 (IC 3639) and its companion ESO381-G009, forming a triple system with ESO381-G006. Broad-band optical photometry is complemented by Halpha imaging, which provides information about the distribution of star forming regions across the galaxies. Long-slit spectroscopic data obtained at different position angles of the slit are employed to determine the physical conditions of circumnuclear and extranuclear regions. A mixture of thermal and non-thermal ionizing radiation is found in the surroundings of the nucleus of Tol1238-364 and the energy budget supports the presence of a circumnuclear starburst.Several regions in both the galaxies show anomalous line ratios: additional ionization by shock-heating and low ionization of some extranuclear HII regions are suggested as possible explanations. An analysis of the emission-line profiles reveals the presence of a broad Halpha component in the nuclear region of Tol1238-364. Independent estimates of the star formation rates (SFR) were obtained through flux-calibrated Halpha-images and FIR emission in the four IRAS bands. Overall SFR densities have been compared with the SFR densities derived from Halpha emission in the individual regions of the galaxies sampled by long-slit spectra. In both galaxies an enhancement of the star formation activity with respect to isolated galaxies is revealed. The prevalence of starburst or nuclear activity has been examined through FIR color indices. The interaction scenario is discussed on the basis of the observed galaxy properties. | astro-ph | astro-ph |
Analysis of the Interaction Effects in the Southern Galaxy Pair Tol1238-364
and ESO381-G0091
S. Temporin
Institut fur Astrophysik, Leopold-Franzens-Universitat Innsbruck, Technikerstrasse 25, A-6020
Innsbruck, Austria
[email protected]
S. Ciroi and P. Rafanelli
Dipartimento di Astronomia, Vicolo dell'Osservatorio 2, I-35122 Padova, Italy
[email protected], [email protected]
M. Radovich
Astronomical Observatory of Capodimonte, Via Moiariello 16, I-80131 Napoli, Italy
[email protected]
J. Vennik
Tartu Observatory, 61602 Toravere, Tartumaa, Estonia
[email protected]
G. M. Richter
Astrophysikalisches Institut Potsdam, An der Sternwarte 16, D-14482 Potsdam, Germany
[email protected]
and
K. Birkle
Max-Planck-Institut fur Astronomie, Konigstuhl 17, D-69117 Heidelberg, Germany
[email protected]
ABSTRACT
In the context of the connection among galaxy-galaxy interaction, starbursts and
nuclear activity, we present and discuss a quantitative morphological analysis based
on BV R images and a detailed spectroscopic investigation of two interacting galaxies,
-- 2 --
the Seyfert 2 Tol1238-364 (IC 3639) and its companion ESO381-G009, forming a triple
system with ESO381-G006. Broad-band optical photometry is complemented by Hα
imaging, which provides information about the distribution of star forming regions
across the galaxies. Long-slit spectroscopic data obtained at different position angles
of the slit are employed to determine the physical conditions of circumnuclear and
extranuclear regions. A mixture of thermal and non-thermal ionizing radiation is found
in the surroundings of the nucleus of Tol1238-364 and the energy budget supports
the presence of a circumnuclear starburst. Several regions in both the galaxies show
anomalous line ratios: additional ionization by shock-heating and low ionization of
some extranuclear H II regions are suggested as possible explanations. An analysis of
the emission-line profiles reveals the presence of a broad Hα component in the nuclear
region of Tol1238-364. Independent estimates of the star formation rates (SFR) were
obtained through flux-calibrated Hα-images and FIR emission in the four IRAS bands.
Overall SFR densities have been compared with the SFR densities derived from Hα
emission in the individual regions of the galaxies sampled by long-slit spectra. In both
the galaxies an enhancement of the star formation activity with respect to isolated
galaxies is revealed. The prevalence of starburst or nuclear activity has been examined
through FIR color indices. The interaction scenario is discussed on the basis of the
observed galaxy properties.
Subject headings: galaxies: individual (Tol1238-364, ESO381-G009) -- galaxies: inter-
actions -- galaxies: Seyfert -- galaxies: starburst
1.
INTRODUCTION
The importance of the role played by interaction in triggering nuclear bursts of star formation
and activity in galaxies has been clearly pointed out in the last decades. N-body simulations have
shown that interactions drive large gas flows toward the center of galaxies (e.g. Barnes & Hernquist
1991). Disk instabilities in colliding galaxies lead rapidly to the formation of strong bars and to gas
inflows fueling early starbursts and/or AGN activity before the galaxy merging (Mihos & Hernquist
1996; Barnes & Hernquist 1996; Mihos 1999). The role of the orbital geometry in triggering inflow
and activity has been considered: prograde encounters favor bar instabilities (Barnes & Hernquist
1996).
Other clues to the connection between interaction and starburst and AGN activity come from
several statistical studies: an excess of both starburst and Seyfert galaxies among galaxy pairs and
an excess of physical companions in samples of starburst galaxies (SBGs) and Seyferts have been
1Based on data collected at the ESO-MPIA 2.2 m telescope in La Silla, Chile
-- 3 --
observed (e.g. Keel & Van Soest 1992; Rafanelli, Violato, & Baruffolo 1995; Rafanelli, Temporin,
& Baruffolo 1997).
Detailed studies of interacting systems can provide a mean to evaluate the validity of this
scenario (e.g. Rifatto et al. 2001). In this context we performed a photometric and spectroscopic
analysis of the Seyfert 2 Tol1238-364 and of its companion ESO381-G009, as well as an evaluation of
far-infrared (FIR) emission of each galaxy of the pair in order to determine the physical properties of
circumnuclear and extranuclear regions and to obtain independent estimates of the star formation
rates.
Tol1238-364 is a Seyfert 2 galaxy morphologically classified as SB(rs)bc; its near infrared bar
is aligned at P. A. = 150◦ (Mulchaey, Regan, & Kundu 1997; Gonz´alez Delgado et al. 1998). After
the recent detection of a broad Hα component in polarized light (Heisler, Lumsden, & Bailey 1997),
the galaxy is listed among the Seyfert 2s hosting a hidden broad line region (HBLR). The column
density, as measured by means of the photoelectric absorption cutoff in the hard X-ray spectrum,
assumes values larger than 1025 cm−2 (Risaliti, Maiolino, & Salvati 1999). It is also classified as
Luminous Infrared Galaxy (LF IR = 4.36×1010 L⊙, Lutz 1992) and is located in the "radio-bright"
side of the FIR/radio correlation (Helou, Soifer, & Rowan-Robinson 1985), as shown by Bransford
et al. (1998). It forms a triple system along with the almost face-on SB(r?) ESO381-G009, located
at 1′.8 arcmin North-East and an edge-on galaxy 2′.6 arcmin to North-West (e.g. Karachentseva
& Karachentsev 2000). Even if the optical images do not reveal strongly distorted morphologies,
traces of plumes and bridges of neutral gas have been detected in the H I 21 cm line (Babic, Price,
& Jones 2000; Barnes & Webster 2001). However, the neutral hydrogen appears still bound to the
individual galaxies. Its mass, based on Australia Telescope Compact Array observations in the 21
cm line (Barnes & Webster 2001), is of 3.6, 4.0, and 1.5 ×109h−2
75 M⊙ for Tol1238-364, ESO381-
G009, and ESO381-G006, respectively. The same data show that the radial velocities, listed in
Table 1 together with the coordinates and B-band absolute magnitudes of the three galaxies, are
concordant.
The Seyfert 2 and its NE-companion are characterized by strongly enhanced star formation also
in the nuclear region. The starburst nature of Tol1238-364 is pointed out by UV (Gonz´alez Delgado
et al. 1998; Gonz´alez Delgado & Heckman 1999), FIR (Lutz 1992) and radio (Bransford et al. 1998)
observations. ROSAT HRI data show soft X-ray emission extended up to a radius of 6.8 kpc best
fitted with a two-component model of a power law and thermal emission (Levenson, Weaver, &
Heckman 2001). ESO381-G009 presents enhanced FIR emission, even if its FIR luminosity is not
sufficient to classify it as Infrared Luminous Galaxy (LF IR = 7.42×109 L⊙, Lutz 1992).
The present paper is organized as follows: Observations and data reduction are described in
§ 2. The main properties of the galaxies are derived through the analysis of the photometric and
spectroscopic data and the comparison with photoionization models in § 3. Comments on the
FIR emission of each of the two galaxies obtained applying Maximum Entropy procedures to the
IRAS data, are given in § 4. The results are summarized and extensively discussed in § 5, where
-- 4 --
considerations on the environment of the galaxies and the evolutionary history of the triplet are
expressed, as well.
2. OBSERVATIONS AND DATA REDUCTION
Our data were obtained under good seeing conditions (. 1′′) during an observing run in April
1995 at the ESO-MPIA 2.2 m telescope in La Silla, equipped with the Faint Object Spectrograph
and Camera EFOSC2. The detector was a 1k×1k CCD with a pixel size of 19 µm and a scale of
0.336 arcsec/pixel. Two images were taken during the same nights in redshifted, narrow band Hα
(λc = 6651.7 A, FWHM = 61.3 A) and continuum (λc = 6521.2 A, FWHM = 74.3 A) filters. The
spectrophotometric standard star Kopff 27 was also observed with the same filters. Additionally
broad-band BV R images of both galaxies, together with standard star fields, were taken. Obser-
vation details concerning both imaging and spectroscopy, including exposure times, slit position
angles (P.A.), and limit surface brightnesses and magnitudes are summarized in Table 2.
Hα images were reduced in a standard way: they were bias-subtracted, flat-fielded, cleaned
from cosmic rays, and background subtracted. The tabulated spectrum of Kopff 27 was convoluted
with the system (filter + CCD + telescope) transmission curve and normalized to the bandwidth
of the Hα filter. From the comparison with the counts measured on the images, corrected for
atmospheric extinction, we computed the zero points to be applied for flux calibration.
The flux calibrated images were corrected for atmospheric extinction. The continuum image
was aligned to the Hα image using field stars close to the galaxy and then subtracted from it; a
check that the scaling factor derived from the photometric calibration was correct was given by the
disappearance of non saturated stars in the continuum-subtracted image. A further check consisted
in comparing the Hα fluxes of the nucleus and the extranuclear regions extracted from the spectra
(see below) with those of the relevant regions in the calibrated image. The fluxes were found in
good agreement (they differ by less than 20%, which is within the errors of measurement).
After the subtraction of the Hα-continuum component, the images were processed by means
of an adaptive smooth filtering procedure developed at the Astrophysikalisches Institut Potsdam
(Richter et al. 1991; Lorenz et al. 1993) in order to reduce the noise and enhance faint details of the
internal structure of the galaxies (Fig. 1). BV R images were also reduced in a standard way and
registered. The non-saturated foreground stars were subtracted by fitting them with a model of
the point spread function (PSF) obtained with IRAF2/DAOPHOT routines. Saturated stars were
masked with the task IMEDIT. The photometric calibration constants have been derived by use
of the package PHOTCAL applied to the standard stars, with the usual transformation equations.
A B −R color map (Fig. 2) was obtained to analyze the color gradient across the galaxies and to
2IRAF is distributed by the National Optical Astronomy Observatories, which are operated by the Association of
Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation
-- 5 --
have an insight into the spatial distribution of young and old stellar populations.
A 2′′-wide slit was used to obtain long-slit spectra covering the spectral ranges 3200 -- 6030
A (grism #3) and 3850 -- 7950 A (grism #6) with dispersions 1.9 A pixel−1 and 2.6 A pixel−1
respectively. The resulting spectral resolution estimated from the FWHM of faint comparison lines
is of the order of ∼10 A, while the scale along the slit is 0′′.336 pixel−1.
Tol1238-364 was observed at three different slit position angles: P.A. = 90◦, 146◦, and 150◦.
The spectra at P.A. = 150◦ and 146◦ were taken across the center of the galaxy (the first one along
the galaxy's bar and centered on the nucleus), while the spectrum at P.A. = 90◦ covers the region
∼ 3′′ South of the nucleus, where many knots of emission are visible in the Hα image (Fig. 3, top).
The spectra of ESO381-G009 were taken at P.A. 90◦ and 125◦ through the nucleus (Fig. 3, bottom).
This last orientation of the slit allowed to observe both the nucleus and the emission knot on its
NW side (see Fig. 3, bottom). Spectra of the standard stars Kopff 27 and Feige 56 were taken
immediately before or after the spectra of the galaxies, to allow their flux calibration.
The two-dimensional spectra were reduced with standard IRAF packages: after bias subtrac-
tion, flat-fielding and a careful subtraction of cosmic rays, they were wavelength calibrated. The
IRAF tasks IDENTIFY, REIDENTIFY, FITCOORDS, and TRANSFORM were used in sequence
to determine the dispersion solution and linearize the two-dimensional spectra. An estimate of the
wavelength calibration error was obtained by evaluating the rms of the mean difference between
measured and predicted sky-line wavelengths. This rms value is ∼ 1.3 A, equivalent to 80 km s−1 at
λ = 5000 A. The spectra were corrected for atmospheric extinction and calibrated (CALIBRATE)
to a flux scale using sensitivity functions obtained with the tasks STANDARD and SENSFUNC
applied to the observed spectrophotometric standard stars. Finally the sky-background was fitted
by using two column samples at both sides of the galaxy spectrum and subtracted with the task
BACKGROUND.
Contour maps of the extended Hα + [N II] λ6548,6583 emission lines and continuum-subtracted
Hα profiles along the slit were used to identify and extract the spectra of the nucleus and of several
extranuclear emitting regions. Such regions are labeled on the slit images overlaid to the contour
maps of the galaxies in Fig. 3: labels without apex are used for regions at P.A. = 150◦ in Tol1238-364
and P.A. = 90◦ in ESO381-G009, an apostrophe indicates the regions at P.A. = 146◦ in Tol1238-364
and P.A. = 125◦ in ESO381-G009, while a quotation mark indicates the regions at P.A. = 90◦ of
Tol1238-364.
A comparison with the Hα images revealed a clear one-to-one correspondence among the so-
selected emitting regions and the emission knots enhanced by the adaptive smooth filter (see Fig. 3).
1-D spectra of these regions (Figs. 4 and 5) were extracted and the corresponding blue and red
parts were combined together. The central regions N and N' of Tol1238-364 were further divided
into & 1.0′′-wide subregions (n1, n2, n3, n4, n5, n1', n2', n3', n4', n5') in order to analyze the
circumnuclear region in more detail. Their spectra are shown in Fig. 6. Regions n3 and n3' are
centered on the Hα emission peaks along the slit. The signal-to-noise ratio (SNR) of the spectra
-- 6 --
ranges from 5 to 15 in the blue (λ ∼ 4000 A) and from 35 to 55 in the red (λ ∼ 6500 A) in the
central regions of Tol1238-364 (n1, ..., n5, n1', ..., n5', C"). In the outermost regions the SNR in
the continuum is significantly lower, although it is still sufficiently good in the prominent emission
lines. The regions with the lowest values are the outermost ones at P.A. = 150◦ and have SNR ∼
2 in the blue, around λ 4000 A, ∼ 5 around λ 5000 A, and ∼ 15 in the red. For ESO381-G009 the
SNR in the center is ∼ 8 in the blue and ∼ 40 in the red, while in the outer regions SNR ∼ 2 in
the blue and ∼ 8 in the red.
A correction for Galactic extinction (AV = 0.17 mag, as derived from the AB value given
by Burstein & Heiles (1982) following Cardelli, Clayton, & Mathis (1989) and assuming a visual
selective extinction RV = 3.1) was applied3.
Before measuring the emission-line fluxes, a template spectrum -- conveniently diluted to match
the absorption features following the method outlined by Ho, Filippenko, & Sargent (1993) -- was
used to correct the spectra for the effects of the underlying stellar population, particularly affecting
Hβ. The Hβ absorption was detected in all but one of the extracted 1-D spectra of Tol1238-3634
and in the central regions of ESO381-G009.
A careful deblending of Hα and the [NII] satellite-lines was performed using a multi-gaussian
fit procedure. In the nuclear region of the Seyfert galaxy a better fit of the blend was obtained
assuming the presence of a broad Hα component (Fig. 7), possibly indicating that this galaxy is
a Seyfert of type 1.9. The mean full-width at half-maximum (FWHM) of the broad component
found in the nuclear regions at P.A. = 150◦ and 146◦ is 1735 ± 60 km s−1, in agreement with the
FWHM ∼ 1860 km s−1 found in polarized light by Heisler et al. (1997). The mean FWHM of the
narrow line is ∼ 300 km s−1.
Measured fluxes of the narrow lines were corrected for internal extinction assuming as intrinsic
value of the Balmer decrement Hα/Hβ=2.85 (Osterbrock 1989). The Galactic interstellar reddening
curve parametrized by Miller & Mathews (1972) has been used. Hereafter we indicate with the
term "intensities" these extinction-corrected fluxes. The internal extinction derived in the above
way can be considered an upper limit in the case of the nucleus of the Seyfert galaxy, since the
intrinsic ratio Hα/Hβ=3.1 might be more appropriate according to some authors (e.g. Veilleux &
Osterbrock 1987). The observed emission-line fluxes and intensities relative to Hβ, the absolute
fluxes of Hα and Hβ, and the values of internal extinction for the individual regions are listed in the
Appendix (Tables A1 through A6). The distances of the extremes of every region to the intensity
peak along the slit, expressed in arcseconds, are indicated, as well.
3The use of the Galactic extinction value derived from the maps of Schlegel, Finkbeiner, & Davis (1998), AV =
0.229 mag, would imply a 5% difference in the measured fluxes.
-- 7 --
3. MORPHOLOGICAL, PHOTOMETRIC, AND SPECTROSCOPIC
PROPERTIES
3.1. Morphology
The morphology of the two galaxies was investigated through the analysis of the broad-band
optical images. In order to study the radial trend of the ellipticity (e = 1 − b
a with a and b semiaxes
of the ellipse) and position angle (P.A.), their isophotes were fitted with ellipses with fixed center.
In the outermost isophotes of Tol1238-364 the P.A. had to be fixed as well. The e and P.A. versus
a plots in Fig. 8 reveal in Tol1238-364 the existence of a bar extended until a ∼ 5′′ with e ≃ 0.5
and P.A. ∼ 145◦, spiral arms mostly visible in B and extended until a ∼ 16′′, and a clear twist
of the isophotes (P.A. ∼ 80◦), which become nearly circular beyond a ∼ 30′′. Also in ESO381-
G009 (Fig. 9) we observe a twist of the isophotes, whose P.A. changes from ∼ 120◦ to ∼ 30◦ and
whose ellipticity increases beyond a ∼ 30′′, giving the impression that they are stretching towards
Tol1238-364. A strongly elliptical bar (e ∼ 0.8) is located at P.A. ∼ 125◦ and seems to extend its
isophotes up to a ∼ 20′′, after which we observe a sudden transition to the spiral arms arranged
into a ring-like shape.
We repeated the isophote fitting with free ellipses in order to investigate the change in the
position of the isophote center. We estimated a decentering degree, as defined by M´arquez & Moles
(1999), of 7.5% for Tol1238-364 and 3.4% for ESO381-G009. This parameter gives an indication of
the degree of asymmetry of the galaxy disks.
A bulge-disk decomposition of the surface brightness profiles of both galaxies in the three
passbands was performed without a PSF-deconvolution, assuming an exponential law for the two
components (Andredakis 1998; Andredakis & Sanders 1994):
Σbulge(r) = 5.36Be exp(cid:20)−1.68
hD(cid:19) ,
Σdisk(r) = D0 exp(cid:18) r
r
re(cid:21)
(1)
(2)
where Σbulge(r) and Σdisk(r) are the surface intensities of the bulge and disk components (in mag
arcsec−2) at a radius r, Be is the surface intensity at the effective radius re (i.e. the radius containing
half of the bulge light), D0 is the central surface intensity of the disk and hD is the disk scale-length.
The structural parameters averaged on the three passbands, the bulge-to-total light ratios B/T ,
and the total magnitudes integrated to infinity are listed in Table3. Magnitudes corrected for
Galactic extinction are reported as well. The B and V magnitudes measured for Tol1238-364 are
in agreement with Hunt et al. (1999). The B/T ratios are typical of disk-dominated galaxies. Let
us note that the existence of a real bulge in ESO381-G009 is uncertain. Actually the central part
of the galaxy consists of a structure composed by two knots, the brightest of which was chosen as
the nucleus, although it is not located in the geometrical center and we lack information about the
kinematics of the system. In order to investigate the color gradient B −R, we built a new B surface
-- 8 --
brightness profile by applying the same ellipses fitted to the R-band isophotes. The B −R profile of
Tol1238-364 (Fig. 10) shows a red color for the bar, which is probably dominated by an old stellar
population, blue spiral arms, and becomes red again in the outermost part of the galaxy, where
the disk might be affected by dust extinction. In ESO381-G009, instead, the inner part of the bar
exhibits a blue color (Fig. 11), probably as an effect of the presence of central star-forming regions
in agreement with the bright knots visible in the Hα image. The profile reddens in the outer part
of the bar and becomes blue again in correspondence of the spiral arms.
The continuum-subtracted Hα image shows the presence of bright emitting regions distributed
all over Tol1238-364, but more concentrated toward its nucleus and in the north-eastern spiral arm
facing the companion galaxy (Figs. 1 and 3), and confirms the structures already revealed by the
analysis of the broad-band images for ESO381-G009. Additionally, a sort of plume departing from
the northern side of ESO381-G009 is visible in all images, as well as an extension of star forming
regions (better visible in the Hα image) on the south, outside the "ring", especially in direction
of the companion galaxy. All the above features are also emphasized as blue regions in the B −R
image (Fig. 2).
3.2. Spectroscopic Classification of the Emission-Line Regions
We analyze in this section the spectroscopic characteristics of the two galaxies separately:
Tol1238-364. The trend of the ionization degree along the direction of the slits is shown in
Fig. 12, where the emission-line ratios [O III] λ5007/Hβ, and [N II] λ6583/Hα are plotted against
the radial distance to the nucleus for the regions identified as explained in § 2. All the diagrams
show a peak in correspondence of the nucleus as expected in the case of photoionization by a source
with a power-law spectrum, and a decreasing ionization degree moving away from it. Nevertheless
the outermost regions exhibit an increasing trend of the [O III]/Hβ ratio and an exactly opposite
trend of the [N II]/Hα ratio. This symmetric behavior is typical of H II-like regions, however we
stress that this trend is observed in only two bins per slit position. The rise of the ionization degree
is more evident at P.A. 90◦ for the regions at the East side of the nucleus and could be partly
caused by an underestimate of Hβ, difficult to measure in spectra with relatively low SNR, and
partly an effect of the presence of a hot interstellar medium.
In order to investigate the nature of the emission line regions and of their ionizing sources we
used the intensities listed in the Appendix (Tables A1 through A6) in order to build the classic
diagnostic diagrams (Figs. 13 and 14) [O III] λ5007/Hβ vs [O I] λ6300/Hα, [N II] λ6583/Hα,
[S II] λ6716+6731/Hα (Veilleux & Osterbrock 1987, VO diagrams), and [O I] λ6300/Hα, [O I]
λ6300/[O III] λ5007, [N II] λ6583/Hα, [O III] λ5007/Hβ vs [O II] λ3727/[O III] λ5007 (Baldwin,
Phillips, & Terlevich 1981; Shields & Filippenko 1990; Halpern & Steiner 1983).
A careful examination of the diagnostic diagrams allowed us to investigate in some detail the
nature of the circumnuclear and extranuclear regions (throughout this paper we call "circumnuclear
-- 9 --
regions" the regions within 1 kpc from the nucleus, assuming H0 = 75 km s−1 Mpc−1). The nuclear
and circumnuclear regions n2, n3, n4, n2', n3', and n4' extracted from the central portions of the
bidimensional spectra have line-ratios characteristic of Seyfert galaxies, while the adjacent regions
n1, n5, n1', and n5', as well as region C" at P.A. = 90◦, gradually move towards the LINER area
of the diagrams, probably indicating a transition from a dominating non-thermal photoionizing
source to a dominating thermal source.
Along P.A. = 150◦ a progressive transition from Seyfert-like to H II-like properties is evident
when going out from the nucleus. The region A1 on the west side of the nucleus shows typical
H II-like features, while B1 on the opposite side has transition properties between LINERs and
H II regions. At P.A. = 146◦ the extranuclear regions all fall in the H II-like area of the diagrams,
although A2' approaches the LINER region in the [O I]/Hα diagram and enters it in the [S II]/Hα
one. A similar behavior is observed in the regions B2" and A3" at P.A. = 90◦; the region B3" has
an anomalously high [O III]/Hβ ratio, which locates it at the border between Seyferts and LINERs.
However, this could be a consequence of an underestimate4 of Hβ and the region is most probably
to be classified as LINER type. In contrast with Bransford et al. (1998) we do not observe any
marked E/W asymmetry in the distribution of the emission line ratios. The discrepancy among
different diagnostic diagrams could be caused by an enhancement of [O I] and [S II] lines consequent
to ionization by starburst-driven shocks, as suggested by Bransford et al. (1998). However, some
of the anomalous regions are not well centered on any knot of emission in the Hα image (Fig. 3),
and have low density of star formation rate (see last column of Table 4; details on the content of
the table are given in §3.4). This could indicate that these regions are weakly ionized by a low
number of young stars. Actually, "faint emission" regions are found to have relatively high values
of [S II]/Hα (∼ 0.3 -- 0.5) and [O I]/Hα (∼ 0.04 -- 0.1) because of their low ionization degree (Smith
et al. 1993).
A qualitative examination of the spectra of the individual extranuclear and circumnuclear
regions (Figs. 4 and 6) gives further insights into their nature.
In particular, we notice that
the spectra of the outermost regions tend to be dominated by young (or sometimes very young)
stellar populations, as suggested by the weakness of metal absorption lines, the strength of Balmer
absorption lines, the high strength of the CaH+Hǫ doublet (2.3 A . EW . 9.5 A) relative to the
CaK absorption line (0.5 A . EW . 5.2 A), and the blue continuum usually peaked around 4000
A (e.g. region A2' in Fig.15). These spectra appear to be dominated by a population of A-type
(or F-type) stars. In two of the outermost regions, B3 and B2', we observe a very blue continuum
without detectable metal absorption lines and very weak Balmer absorption lines, indicative of a
dominating population of O-B stars. Toward inner radii, the extranuclear regions show a slightly
redder continuum, peaked around 4500 A, progressively deeper metal absorption lines, but still
very strong Balmer absorption lines indicating important contributions to the spectrum from both
4The correction for underlying stellar absorption was particularly difficult for this spectrum, therefore the measured
Hβ flux is uncertain.
-- 10 --
a relatively young and an old (G to K-type) stellar population (e.g. region B1 in Fig.15). The
most important contribution from old stellar populations is found in the circumnuclear regions,
which exhibit a red continuum, strong metal lines and comparably strong Balmer absorption lines
(see e.g. regions n5, n2' in Fig.15). The CaK (3.5 A . EW . 7.7 A) absorption line becomes
stronger than the CaH+Hǫ doublet (1.6 A . EW . 4.5 A), as typical of G-type or later stellar
populations (e.g. Rose 1984; Leonardi & Rose 1996), but not in all the regions within the central
kiloparsec.
In particular this is not the case for regions n1', C", and especially the nucleus n3
(EW(CaK) ∼ 3 A, EW(CaH+Hǫ) ∼ 5 A), where an A-type population probably gives a non-
negligible contribution to the spectrum, and region n2, where a red continuum is overlapped with a
component rising toward blue wavelengths (Fig.15). These last spectra give evidence of the presence
of (relatively) recent nuclear and circumnuclear starbursts. The general trend described above, with
old population contributions becoming progressively more important with decreasing radius and
young populations dominating the outermost regions (at radii of about 3-5 kpc), is in agreement
with the stellar population gradients observed by Raimann et al. (2002). However, these authors
(their Fig. 3) find that in the nucleus the CaK absorption line has equivalent width significantly
larger than CaH+Hǫ. This discrepancy with our results could be a consequence of a different spatial
sampling and size of the extracted regions. While the spatial scale of their spectra is 0′′.7 pixel−1
and they extracted one-dimensional spectra in windows of 2′′.1 in the nuclear regions, our spatial
scale is 0′′.336 pixel−1 and the good seeing conditions allowed us to extract nuclear regions ∼ 1′′
in size along the slit. Rebinning our spectra to match the spatial sampling of Raimann et al. and
extracting regions of similar size, we verified that a 1-pixel offcenter (i.e. 0′′.7) is sufficient to obtain
a spectrum of the nucleus dominated by an old stellar population, and thus reproduce the findings
of Raimann et al. (2002).
Indeed, the distribution of stellar populations in the circumnuclear
regions appears very inhomogeneous with strong variations within a few arcseconds, which makes
the analysis very sensitive to the extraction method.
Since in H II regions the equivalent widths of Balmer emission lines are proportional to the
ratio of ionizing photons to visible continuum photons from the embedded stars (Kennicutt, Keel,
& Blaha 1989), we can obtain additional information on the population of the H II regions from the
observed values of EW(Hα) listed in Table 4. The Hα EWs of the extranuclear regions are at (or
even below) the lower limit of the distribution found by Kennicutt, Keel, & Blaha (1989) for disk
H II regions (defined as the regions located at a distance > 1 kpc from the nucleus) of spiral and
irregular galaxies. These low emission-line equivalent widths could indicate that the contribution
of the young stellar associations to the optical continuum is rather small and might suggest a
significant contribution to the continuum by older stellar populations or a prolonged period of star
formation over the disk.
ESO381-G009. The ionization-sensitive line ratios at several positions along the slits and the
diagnostic diagrams have been plotted for the companion of Tol1238-364 as well (emission line
fluxes are given in Tables A6 and A7). As expected, the [O III]/Hβ and [N II]/Hα (Fig. 16) have
symmetric trends typical of H II-like regions, except for the regions A3', A4', A5', which exhibit
-- 11 --
the same trend in both the line ratios. Actually, also for this galaxy not all the line ratios can be
explained by pure photoionization from thermal radiation: the shift of several points toward the
LINER area in the [S II]/Hα diagnostic diagram (Fig. 17) could be caused in some cases by the
faintness of the emission regions (A2, B1), but in other cases (A1, A1', A2', A3') they are probably
an indication of shock heating as additional ionization mechanism (Dopita & Sutherland 1995).
The spectra of all the regions show typical properties of young stellar populations with a blue
continuum. Metal absorption lines are detectable only in the nucleus (N and N') together with
deep Balmer absorptions. However, the Hα equivalent width measured in the spectra are rather
low (see Table 5), ranging from ∼ 15 to ∼ 83 A.
3.3. Photoionization Models
The code CLOUDY 90 (Ferland et al. 1998) has been used to calculate photoionization mod-
els, which have been compared to observational emission-line flux ratios, in order to evaluate the
physical parameters of extranuclear and circumnuclear regions. CLOUDY is designed to simulate
emission-line regions ranging from intergalactic medium to the BLR of quasars. Assuming diluted
gas, heated and ionized by the radiation field of a central object, and simultaneously solving the
equations of statistical and thermal equilibrium, CLOUDY can predict the physical conditions of
the gas and its resulting emission line spectrum.
Simple models can be obtained by specifying as input parameters the continuum emitted by
the source, a set of assumed chemical abundances, the total hydrogen density NH, and the ionization
parameter U = QH/(4πr2NHc), where QH is the number of ionizing photons and r is the distance
between the source and the inner side of the gas cloud assumed to have a plane parallel geometry.
We derived a first estimate of the physical parameters and abundances of the ionized gas in
the extranuclear regions by comparison with empirical diagrams (McGaugh 1991; Denicol´o et al.
2002). In particular the R23 ratio, defined as log[( [O II] λ 3727 + [O III] λ 4959,5007)/Hβ], was
compared with the model grid in the plane [O III]/[O II] vs R23 in Fig. 10 of McGaugh (1991).
From this comparison the ionization parameter was found in the range −3.5 . logU . −3.0. The
metal abundances were derived by means of the N2 = [N II] λ 6584/Hα calibrator (Denicol´o et
al. 2002, their Fig. 1) by referring to the model track for logU = −3. They were found in the
ranges5 8.2 . 12+ log(O/H) . 8.6 and 8.2 . 12+ log(O/H) . 8.45 for Tol1238-364 and ESO381-
G009, respectively. The choice not to directly calculate the abundances was motivated by the
fact that the metallicity is very sensitive to the gas temperature, whose determination requires the
measurement of the [O III] λ4363 line. Some authors (Raimann et al. 2000) found a ratio [O III] λ
4363/Hβ . 0.02 in H II and starburst galaxies, which implies electronic temperatures lower than
104 K. Unfortunately, the low SNR of our spectra in the blue range prevented the detection of this
5[12+ log(O/H)]⊙ = 8.9
-- 12 --
line at intensity levels lower than 0.1 relative to Hβ in all the extranuclear regions.
Models were obtained assuming as continuum Mihalas non-LTE stellar atmospheres (Mihalas
1972) with T = 40000 K and as NH a typical value derived from the [S II]6716/6731 ratios (NH ∼ 50
cm−3), while the ionization parameter was left free to vary in the range −4 < logU < −3 with step
0.25 dex, and the metal abundances were varied in the range 0.1Z⊙ ≤ Z ≤ 0.4Z⊙ with step 0.1. The
resulting grid of models (dotted lines) is plotted onto the [N II]/Hα diagnostic diagram in Fig. 19
overlapped with the observational points. The models that appear to better reproduce the observed
line ratios for most of the extranuclear regions of Tol1238-364 are those with −3.50. logU. −3.25
and abundances slightly changing around 0.2Z⊙. The same models seem to be in agreement also
with ESO381-G009 flux ratios, but with a larger spread in metallicity between 0.1Z⊙ and 0.2Z⊙
and with higher ionization parameters, distributed around −3.25.
In the case of the circumnuclear regions of Tol1238-364, which are dominated by the non-
thermal ionization (§3.2), the models were calculated by assuming as continuum a power law with
spectral index α = −1.8 (Fν ∝ ν α), NH = 103 cm−3 (as derived from the [S II]6716/6731 ratios),
and −3 < logU < −2. Models with different power law index (−1.5 and −2.0) have also been
attempted but they significantly deviate from the observed values. In these regions the electronic
temperature of the gas can be estimate since [O III] λ4363 is sufficiently strong to be measured.
Nevertheless, the usual formulas (e.g. Pagel et al. 1992; Izotov, Thuan, & Lipovetsky 1994) for
the direct calculation of metal abundances are calibrated with typical H II regions and not with
active nuclei, therefore they cannot be applied in this case. Trying to model the N2 calibrator as
a function of metallicity and ionization parameter for a power-law ionizing continuum with index
−1.5 we have obtained theoretical tracks, which are far below the observed [N II]/Hα ratios. This
is an indication that there is a significant overabundance of nitrogen. Indeed, theoretical tracks
obtained assuming a triple nitrogen abundance approach much more the data in the range of the
expected ionization parameter for an AGN. For the above reason we constructed a grid of models
with metallicities Z = 0.5, 1.0, and 1.5Z⊙ and nitrogen abundances higher by a factor of 3. The
models with −2.75 < logU < −2.50 and solar or even supersolar metallicity (Fig. 19, dash-dot
lines) reproduce in a reasonable way the nuclear and circumnuclear regions within a radius of ∼
0.5 kpc from the nucleus. The comparison of these models with the observed [S II] emission-line
fluxes reveals also a possible overabundance (by a factor of ∼ 2) of sulfur in the circumnuclear
regions of Tol1238-36.4. Similar cases of nitrogen and sulfur overabundances have been frequently
observed in Seyfert 2 galaxies (Storchi-Bergmann & Pastoriza 1990) and are probably a consequence
of circumnuclear starburst activity. However, our observed flux-ratios show also a good agreement
with the two-component models by Moy & Rocca-Volmerange (2002), which combine the effects
of shocks with AGN photoionization.
In particular, the strong [O III] λ4363 and [O I] λ6300
measured in the circumnuclear regions of Tol1238-364 are well reproduced by these models, whereas
pure photoionization models predict too low values for these emission lines. Therefore we cannot
exclude as alternative explanation to our flux-ratios, the presence of shocks as an additional source
of ionization.
-- 13 --
Regions n1, n5, n1', n5', and C" are not reproduced by any of the above models. As we
have already pointed out in § 3.2, these regions are located in a transition area of the diagnostic
diagrams. Their properties can be explained with hybrid models involving different proportions of
mixed thermal and non-thermal ionization, as explained in Radovich, Hasinger, & Rafanelli (1998).
3.4. Star Formation and Energy Budget
The Hα line intensity of all emitting regions extracted from the spectra of Tol1238-364 and
ESO381-G009 were converted into luminosities (Table 4 and 5) using the distances 47.9 Mpc and
44.3 Mpc, respectively, derived from the mean redshift values measured along the slits for each of
the two galaxies. These luminosities were used to calculate the corresponding star formation rates
(except for the pure Seyfert-like regions) adopting the relation of Hunter & Gallagher (1986):
SFR = 7.07 × 10−42LHα M⊙ yr−1
(3)
The values shown in Tables 4 and 5 for Tol1238-364 and ESO381-G009 range from about 0.001
to 0.15 M⊙ yr−1, but much more significant are the SFR densities, which are 10 to 100 times higher
than in normal spirals (Elmegreen 1998). The mean value of the SFR density derived considering
all H II-like regions identified along the slits up to a distance of 28′′ from the nucleus is 1.60 × 10−7
M⊙ yr−1 pc−2 in Tol1238-364.
The total flux measured on the calibrated Hα image within a radius of 37′′ is F(Hα+[N II]) =
3.83×10−12 ergs s−1 cm−2. The percentage of [NII] emission detected by the Hα interference filter
has been evaluated taking the average of the [NII]/Hα ratios measured in the extranuclear spectra
(<[NII]6548+6583/Hα> = 0.48), although in the central kpc the [NII] lines contribute to the flux
in the Hα image by a higher percentage. Galactic and internal extinction corrections were applied.
For the last, the mean extinction value derived from the spectra, E(B-V) = 0.4, was used. No
attempt was made to correct for the contribution from the active nucleus. The resulting corrected
Hα luminosity is LHα = 1.44×1042 ergs s−1, which yields a total SFR = 10.22 M⊙ yr−1 and a SFR
density of 4.41 × 10−8 M⊙ yr−1 pc−2.
In a similar way, mean SFR densities of 1.1 × 10−7 and 1.62 × 10−8 M⊙ yr−1 pc−2 were
obtained from the spectra and the Hα image of ESO381-G009. The total luminosity and SFR
evaluated from the image within a radius of 41′′ are 5.58×1041 ergs s−1 and 3.95 M⊙ yr−1, respec-
tively. In this case the average observed ratio [N II]6548+6583/Hα = 0.33 and the mean internal
extinction value E(B-V) = 0.59 were used to correct the flux F(Hα+[N II]) = 1.0×10−12 ergs s−1
cm−2 measured in the image. In both cases the SFR densities derived from the spectra are consid-
erably larger than the corresponding value derived from the Hα images. This apparent discrepancy
is a natural consequence of the orientation of the slit, preferentially located along alignments of
knotty structures on the galaxies. A comparison with the work of Bushouse (1987) shows that both
-- 14 --
galaxies have SFR densities approaching the upper value found for a sample of interacting spiral
galaxies (6.1×10−11 . SFRD . 7.6×10−8 M⊙ yr−1 pc−2) and clearly higher than values found
in a sample of isolated spirals (2×10−10 . SFRD . 2×10−8 M⊙ yr−1 pc−2). Low values of H I
depletion time scales, log(MHI)/SFR) = 8.5 - 9 yr, are also observed. This is expected in case of
interacting systems, according to the distribution shown in Figure 11 of Bushouse (1987).
As already noticed in § 3.1 the Hα emitting regions of Tol1238-364 appear more concentrated
in the spiral arm facing the companion galaxy. Also in ESO381-G009 a concentration of H II
regions seem to be present in direction of Tol1238-364 (Fig. 2). In order to better investigate the
distribution of the emitting regions in the two galaxies, we measured the Hα fluxes in 18 circular
sectors with aperture 20◦ and radii 2, 4, and ∼ 10 kpc, excluding the nuclei. In each galaxy, the
fluxes were normalized to the total galaxy flux (excluding the nucleus). The angular distribution
of the normalized fluxes is represented with bar histograms in Fig. 20. Bars with horizontal dashes
indicate fluxes within 2 kpc, bars with inclined dashes indicate fluxes between 2 and 4 kpc, and
empty bars indicate fluxes at radii > 4kpc. The range of angles in which each galaxy faces the
companion are marked with horizontal bars. The enhancement of Hα flux at these positions is
evident, especially at the outer radii for ESO381-G009 and both at intermediate and outer radii
for the Seyfert galaxy. A second peak in the angular flux distribution of ESO381-G009 is found
around P.A. 300◦ in correspondence of the NW end of its bar.
SFRs in the active galaxy have been calculated for all the extranuclear regions and those
circumnuclear regions whose diagnostic emission-line ratios suggest a mixture of thermal and non-
thermal ionizing radiation. In the latter case the SFR values quoted in Table 4 are upper limits
(marked with a colon), since the fractional contribution of the active nucleus to the Hα luminosity
is unknown.
However, some simple energy budget considerations could be done in order to verify the capa-
bility of the Seyfert's non-thermal source to ionize the regions outside the nucleus. The number of
ionizing photons necessary to produce the observed Hα luminosity,
Qion = 7.3 × 1011 LHα photons s−1
(4)
(Kennicutt 1998), was evaluated in every region and compared (Table 4) with the number of nuclear
ionizing photons which in principle could reach the considered region (Q′
nuc).
Q′
nuc is actually a fraction of the total number of ionizing photons produced per second by the
central source (Qnuc). In fact it is diluted by the covering factor Ω/4π, which depends on the size
and distance of each region to the nucleus (Ω is the solid angle under which the considered region
is seen from the nucleus). Obviously a precise measure of this factor cannot be achieved, since we
see only the projected sizes and distances of the regions.
The value of Q′
nuc is given by
Q′
nuc = Qnuc
Ω
4π
.
(5)
-- 15 --
Because of the impossibility to estimate the covering factor of the very central region n3, the
value of Qnuc has been evaluated in indirect way by selecting among all the circumnuclear regions
those exhibiting a clear Seyfert nature and similar line ratios according to the diagnostic diagrams,
namely n2, n4, n2', n3' and n4'. The Qion of such regions was multiplied by 4π/Ω and the average
of the resulting values was assumed to be the actual number of nuclear ionizing photons
Qnuc =< Qion
4π
Ω
>,
(6)
included in Table 4 in correspondence of the central subregion n3. This Qnuc was then diluted to
obtain Q′
nuc of every region.
In the region-by-region comparison of Qion and Q′
nuc there was no need to take into account
the filling factor, that is the fraction of the total volume occupied by the gas. This factor is believed
to be very low, typically few 0.01 (Durret 1990), and is generally assumed to affect in the same
measure both the Narrow Line Region and the extranuclear regions.
This comparison revealed that Qion >Q′
nuc for all the regions located at distances R&1 kpc
from n3 indicating that their ionization is not dominated by the active nucleus. This result is in
agreement with the diagnostic diagrams, which show a clear thermal nature of the ionizing source
in these regions. Within the central kiloparsec Qion.Q′
nuc, except for the region n2. In fact the
nuclear ionization is clearly dominating in the regions n4, n2', n3' and n4', which lie inside the
Seyfert area of the diagnostic diagrams. Instead it is likely mixed in different percentages to the
ionization from thermal sources in the regions n5, n5' and C", which lie in the Seyfert area, but
slightly displaced toward the Liners, and in the regions n1' and n1, which occupy the Seyfert-
Liner transition region and the Liner area, respectively. According to the diagnostic diagrams
(Fig. 13), also the circumnuclear region n2 is dominated by the nuclear non-thermal radiation,
however it exhibits an opposite behavior (Qion >Q′
nuc) with respect to the other circumnuclear
regions. In particular its high observed Qion would require a number of nuclear ionizing photons
a factor 3.4 higher than the estimated one. One could argue that Qnuc, as determined above, is
still underestimated due to dust absorption and that only in direction of n2 the real amount of
nuclear ionizing radiation can be estimated. In such a circumstance we would expect to observe
a similarly high Qion also in the immediately adjacent n1 region. Since this is not the case, we
rule out this hypothesis and confirm instead the presence of a circumnuclear starburst inside n2 as
already suggested in § 3.2 on the basis of the analysis of its spectral properties. This idea is also
supported by the estimated value of internal extinction in n2, AV ≃ 3.3 mag, much higher than in
the surrounding zones.
The number of ionizing photons produced in every region was calculated also for ESO381-
G009 (Table 5), and expressed in equivalent number of O5 stars, N(O5), assuming that each O5
star emits ∼ 5 × 1049 ionizing photons s−1 (Osterbrock 1989). The obtained values range from the
upper limit for normal H II regions ionized by clusters or associations of OB stars to values found
for "giant" and "supergiant" H II regions (Kennicutt, Edgar, & Hodge 1989). The starbursts with
the highest values of SFR are located in the nucleus of the galaxy.
-- 16 --
4. FIR EMISSION
The connection between the enhanced far-infrared emission and galaxy interactions has been
established in many works (e.g. Telesco, Wolstencroft, & Done 1988; Borne et al. 2000). In order
to study the infrared activity of each member of the pair, position and flux calibrated raw IRAS
data were extracted from the IRAS database server of the Space Research Organisation of the
Netherlands (SRON). The program GIPSY (Groningen Image Processing System, Assendorp et
al. 1995), was used to create low resolution co-added maps at 12, 25, 60 and 100 µm centered on
Tol1238-364.
The overlap of these maps on the corresponding Digitized Sky Survey optical image allowed to
show the fainter emission of ESO381-G009 at 12 and 25 µm, compared with the active galaxy. At
60 and 100 µm both galaxies seemed to be embedded in the same emission, because of the larger
size of the detectors at those wavelengths (Assendorp et al. 1995). To remove the "confusion" and
evaluate the contributions of the two objects, a higher resolution (by a factor of ∼5) was achieved
(Fig. 21) by applying the program HIRAS, which drives the MEMSYS5 maximum entropy imaging
algorithm (Bontekoe, Koper, & Kester 1994).
The result showed ESO381-G009 as an emitting source clearly separated from Tol1238-364 in
all the four bands. The sum of the flux densities of the two galaxies, measured by means of the
GIPSY task FLUX (Table 6), was compared with previously published values, for which only a
unique emitting source was considered. In particular our results are in good agreement both with
the Point Source Catalog, apart from a slightly higher emission at 100 µm in our measurement,
and with the values of Rush, Malkan, & Spinoglio (1993) obtained using the ADDSCAN procedure
of the Infrared Processing and Analysis Center (IPAC).
Our flux ratios were compared with the infrared color-color diagrams log(F25/F12) vs log(F60/F25)
and log(F100/F25) vs log(F60/F12), which can indicate whether the FIR emission is dominated by
a dust-extinguished active nucleus, a pure starburst or a mixture of these two components (Do-
pita et al. 1998). We found that the FIR emission of Tol1238-364 is dominated by a moderately
obscured AGN, while ESO381-G009 has an expected "warm" starburst nature (Fig. 22). Further
confirmations came from the analysis of the spectral indices α100,60 and α60,25, and mostly of the
60µm curvature (α60,25 − α100,60), whose values −0.425 and −1.545 are typical of a Seyfert 2 and
an H II galaxy respectively (Miley, Neugebauer, & Soifer 1985).
The total fluxes between 42.5 and 122.5 µm have been calculated following Helou et al. (1985):
FIR = 1.26 × 10−11(2.58F60 + F100)
erg cm−2 sec−1
(7)
and converted into luminosities, using the distances given in § 3.4. The resulting values (Ta-
ble 6) indicate that Tol1238-364 is ∼ 10 times brighter than ESO381-G009, a clear effect of the
In fact the dusty torus is expected to contribute notably to the
presence of an active nucleus.
-- 17 --
far-IR luminosity (Storchi-Bergmann, Mulchaey, & Wilson 1992; Genzel et al. 1998; Lutz et al.
1998), which is generally produced by interstellar dust heated by the UV radiation field of young
and hot stars. Considering a single temperature component and a λ−1 emissivity law (Young et
al. 1989), a major content of warm dust was revealed in Tol1238-364 (Table 6), likely concentrated
into the bar and in correspondence of the giant H II regions populating its spiral arms. However,
the star formation rate, derived from the FIR luminosities following the relation given by Hunter et
al. (1986), is high in both galaxies as expected to occur in an interacting system. The overall SFR
estimated from LF IR (SFRLF IR ∼ 15.9 M⊙ yr−1) appears somewhat higher than that estimated
from LHα (SFRLHα ∼ 10.2 M⊙ yr−1). In fact, the far infrared luminosity generally includes the
contribution of the cirrus component, diffuse dust heated by the starlight radiation field, whose
effect is to give an overestimate of the star formation rate, up to a factor 2 (Elmegreen 1998).
Furthermore, in the case of the Seyfert galaxy the dust heated by the nuclear non-thermal radia-
tion contributes to the total FIR emission in a percentage which cannot be determined. On the
opposite, for ESO381-G009, SFRLHα is a factor ∼ 2 lower than SFRLF IR. A possible reason could
be the error related to the "deblending" of the two IRAS sources.
5. DISCUSSION AND CONCLUSIONS
We have analyzed the physical properties of the galaxy pair Tol1238-364 and ESO381-G009,
which belongs to a triple system together with ESO381-G006, on the basis of optical imaging and
long-slit spectroscopy in order to point out possible effects of interaction.
The triple system was analyzed by Barnes & Webster (2001) as well, who suggested that also
a fourth galaxy, ESO381-G014, might belong to this small galaxy group, because of its detection in
radio observations at a radial velocity of 3304 km s−1. For the supposed four-members group they
reported a median radial velocity of 3285 km s−1, a median projected velocity dispersion of 11 km
s−1, a median projected galaxy-galaxy separation of 110 kpc, and a median crossing time of order
10 Gyr. Since they did not detect strong signs of galaxy interactions, like a common envelope of
neutral gas or prominent tidal tails, they speculated that the triple system might be much looser
than the projected density implies. ESO381-G014 has been indicated as a member of the group
also in the catalog of groups of nearby optical galaxies (NOGG)6 compiled by Giuricin et al. (2000).
However, its projected separation from the triplet (we assume as average coordinates of the triplet
α(J2000) = 12h40m50s.7, δ(J2000) = −36◦44′32′′) is ∼ 42′, corresponding to ∼ 0.5 Mpc, assuming
H0 = 75 km s−1 Mpc−1), i.e. ∼33 times the diameter of Tol1238-364 as estimated at the 25 mag
arcsec−2 B isophote. Instead, the centers of the galaxies in the triplet are all encompassed by a
minimum circle of radius 1′.87 (25 kpc) and have a median projected separation of 2′.63 (i.e. ∼
6In the NOGG catalog, which contains only galaxies brighter than B = 14 mag, the group is erroneously indicated
as a quartet, because the galaxy ESO381-G009 is counted twice, with the alternative names PGC 042519 and PGC
097487.
-- 18 --
35 kpc). While we have no pieces of evidence other than the concordant radial velocity to support
either exclude the membership of ESO381-G014 to the group, we consider it unlikely for this galaxy
to have a strong gravitational influence on the members of the triplet.
The radial velocity of the galaxies pose them at the same distance of the Centaurus cluster
(Abell 3526, α(J2000.0) = 12h48m48s.7, δ(J2000.0) = -41◦18′44′′). However, they are at a projected
distance of ≈ 3.5 Mpc from the cluster center and well separated from the main cluster galaxy
concentration, therefore it is unclear whether they can be considered cluster members. In any case,
at such a separation the influence of the Centaurus cluster on the triplet, if any, is most likely
negligible with respect to the mutual influence of the galaxies within the triplet itself, provided this
is really a tight system. We have actually observed properties in Tol1238-364 and ESO381-G009
(the tightest pair within the triple system, at least in projection on the sky) that are consistent
with an interaction scenario, as we discuss in the following.
Both galaxies show a prominent bar, and ESO381-G009 also a ring of H II regions. Theoretical
N-body simulations show that close encounters between galaxies can lead to bar formation (e.g.
Noguchi 1987; Gerin, Combes, & Athanassoula 1990), although bars can form in isolated galaxies,
as well, as a consequence of disk instabilities.
Indeed, studies of the relative frequency of bars
among various Hubble types and in different environments have demonstrated that galaxies in close
binary systems and some groups have a greater occurrence of bars than field galaxies (Elmegreen,
Elmegreen, & Bellin 1990). This result is somewhat controversial, in fact it was recently found (van
den Bergh 2002) that the bar frequency does not appear to depend on the galaxy environment.
Nevertheless, we think that the presence of a bar in at least two out of three members of a galaxy
system, like in our case (since ESO381-G006 is edge-on we cannot establish whether it is barred
or not), strongly favors the hypothesis that the mutual interaction has played a role in the bar
formation. Further support to this hypothesis is given by the detection of star formation along the
bars, which indicate that the bars are still relatively young and actively funneling gas toward the
center of the galaxies.
The two galaxies show evidences of only slight morphological distortions (see §3.1). Specifically,
we observe a twist of the inner isophotes, most likely caused by the bars, and an elongation and
change of the P.A. of the outermost isophotes of both galaxies toward one another. Additionally,
the decentering degree of the most external isophotes with respect to the luminosity center of
Tol1238-364 is higher than the maximum value (5%) found by M´arquez & Moles (1999) for isolated
galaxies, indicating a significant asymmetry of the outskirts of the disk likely caused by interaction
processes. On the contrary, ESO381-G009 does not exhibit such an asymmetry of the outermost
isophotes, but its Hα image shows an extension of H II regions, also detectable in the broad band
images, on the southern side of the disk, outside the ring-like structure, facing the Seyfert galaxy.
Additionally, the optical images show a plume on its northern side. This plume appears more
extended in the contour maps of the H I 21 cm line (Babic, Price, & Jones 2000; Barnes & Webster
2001). No other tidal features are visible in the optical images, but there are signs of an H I bridge
between Tol1238-364 and ESO381-G006 (Babic, Price, & Jones 2000). However, since the detection
-- 19 --
of this bridge is limited to one velocity channel, the possibility that this is a spurious feature cannot
be excluded. Most of the neutral hydrogen remains located in the discs of the galaxies.
We note that the formation of prominent tidal tails is not ubiquitous in interacting galaxies,
but depends on the geometry of the encounters, being particularly favored in prograde interactions,
as shown by numerical simulations (see, e.g., the seminal work of Toomre & Toomre 1972). In a
recent work, Barton Gillespie, Geller, & Kenyon (2003) identified within a sample of close galaxy
pairs a number of galaxies exhibiting triggered star formation but not long tidal tails. The color
profiles of these galaxies show blue dips in their centers, in analogy with the B −R profile we
obtained for the non-active galaxy ESO381-G009.
In fact, we could not identify a real bulge structure in this galaxy, but two knots with high
SFR, the brightest of which we adopted as nucleus. This kind of structure appears in agreement
with the gas infall and centrally concentrated star formation predicted by the numerical simulations
of Mihos & Hernquist (1996) for bulgeless galaxies in the early stages of interactions.
Tol1238-364 hosts a typical Seyfert 2 nucleus, with a hidden BLR. The non-thermal ionization
produced by the central power-law source is confined within the first kiloparsec and no evidence of
ionization cones is found along the three studied directions. Considerations on the energy budget
(§ 3.4) reveal the presence of at least one circumnuclear starburst, a hypothesis that finds support
in the observed spectral features. Besides the central starbursts, a large number of H II regions are
found in both galaxies with a major concentration in the zone of their minimal mutual distance,
further supporting the interaction scenario. In ESO381-G009 the H II regions are mainly located
along the bar and in the ring, while in Tol1238-364 they are diffused all over the disk. The SFR
and the density of SFR are strongly enhanced with respect to normal and/or isolated spirals, and
higher in Tol1238-364 than in its companion (see § 3.4). This enhancement could be a combined
effect of the mutual interactions of the galaxies, as shown by the comparison between the star
formation properties of interacting and isolated galaxies (see e.g. Bushouse 1987; Kennicutt et al.
1987; Keel & Van Soest 1992), and of the perturbation of the whole galactic disk by the stellar
bar (Aguerri 1999), although the relation between the global star formation and the presence of
a bar in a galaxy is still debated. Also, numerical simulations demonstrated that both galaxy
interactions (e.g. Barnes & Hernquist 1991, 1996; Mihos & Hernquist 1996) and bars or, in general,
non-axisymmetric features (as summarized e.g.
in Combes 2001) can produce torques and gas
radial inflows toward the center, inducing nuclear starbursts and/or fueling a central AGN.
The H II regions of both galaxies,whose excitation is in general the effect of photoionization
from hot, young stars although in some cases the contribution of starburst driven shocks cannot
be excluded, exhibit subsolar metallicities. However, the Seyfert nucleus shows higher (solar or
even supersolar) metallicity, with traces of overabundance of nitrogen and sulphur, which could be
related to the presence of circumnuclear starburst.
The IRAS infrared colors confirmed the dual starburst - AGN nature of Tol1238-364 and
the simple starburst nature of ESO381-G009, but the infrared emission of the Seyfert appears
-- 20 --
dominated by the active nucleus and shows higher FIR luminosity and dust content.
Based on the result of their analysis, Barton Gillespie et al. (2003) suggested that blue central
colors, moderate EW(Hα), and small velocity separations in galaxy pairs, are indicative of galaxies
that have undergone a close pass and than moved apart, while their triggered burst of star formation
ages. Accordingly, numerical simulations by Mihos & Hernquist (1994), showed that two interacting
disk/halo galaxies remain relatively unperturbed and show no increase of star forming activity until
they reach the perigalacticon. Furthermore, an enhanced star formation activity is also induced
in the disks of the galaxies before they reach the widest separation after their first encounter.
The observational properties we have outlined in the present work suggest that this stage of the
evolution, i.e. the phase subsequent to a first close passage, might be the case for the galaxy pair
under study. The lack of really prominent morphological distortions might indicate that the galaxy
separation is actually larger than appearing in projection on the sky.
In conclusion, although some of the observed properties, taken individually, can be found
also in isolated galaxies, when considered all together they provide, in our opinion, a significant
indication of an undergoing interaction between two gas-rich galaxies. From our data we cannot
establish which role is played by ESO381-G006 in the interaction. On the basis of a comparison
with the findings in other interacting pairs (e.g. Barton Gillespie et al. 2003) and the results of
galaxy encounter simulations (Mihos & Hernquist 1994, 1996), we suggest that the galaxy pair has
already undergone a first close passage, which determined the onset of bursts of star formation both
in the disks and in the central regions, caused moderate morphological distortions, and, possibly,
disk instabilities leading to bar formation. Nevertheless, a clear connection of the Seyfert activity
with the interaction cannot be demonstrated and remains at a speculative level.
We are very grateful to the anonymous referee for precious comments which highly improved
the presentation of the paper. ST acknowledges support from the University of Padova through
a grant which has funded part of this work. JV acknowledges support provided by the Estonian
Science Foundation, grant 4702.
A. Measured Emission Line Fluxes of Tol1238-364 and ESO381-G009
REFERENCES
Aguerri, J. A. L. 1999, A&A, 351, 43
Andredakis, Y. C. 1998, MNRAS, 295, 725
Andredakis, Y. C. & Sanders, R. H. 1994, MNRAS, 267, 283
Assendorp, R., Bontekoe, Tj. R., de Jonge, A. R. W., Kester, D. J. M., Roelfsema, P. R. & Vesselius,
P. R. 1995, A&AS, 110, 395
-- 21 --
Babic, B., Price, R. M. & Jones, K. 2000, in ASP Conf. Ser. 209, Small Galaxy Groups, ed. M. J.
Valtonen & C. Flynn (San Francisco: ASP), 132
Baldwin, J. A., Phillips, M. M., Terlevich, R. J. 1981, PASP, 93, 5
Barnes, J. E. & Hernquist, L. E. 1991, ApJ, 370, L65
Barnes, J. & Hernquist, L. 1996, ApJ, 471, 115
Barnes, D. G. & Webster, R. L. 2001, MNRAS, 324, 859
Barton Gillespie, E., Geller, M. J., & Kenyon, S. J. 2003, ApJ, 582, 668
Bransford, M. A., Appleton P. N., Heisler C. A., Norris, R. P., Marston, A. P. 1998, ApJ, 497, 133
Bontekoe,Tj. R., Koper, E., & Kester, D. J. M. 1994, A&A, 284, 1037
Borne K. D., Bushouse, H., Lucas, R. A. & Colina, L. 2000, ApJ, 529, L77
Burstein D. & Heiles C. 1982 AJ, 87, 1165
Bushouse, H. A. 1987, ApJ, 320, 49
Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245
Combes, F. 2001, Advanced Lectures on the Starburst-AGN Connection, Singapore: World Scien-
tific, p. 223
Denicol´o, G., Terlevich, R., & Terlevich, E. 2002, MNRAS, 330, 69
Dopita, M. A., Heisler, C., Lumsden, S., Bailey, J. 1998, ApJ, 498, 570
Dopita, M. A. & Sutherland, R. S. 1995, ApJ, 455, 468
Durret, F. 1990, A&A, 229, 351
Elmegreen, D. M. 1998, Galaxies and Galactic Structure, Prentice-Hall Inc., Upper Saddle River,
NJ 07458
Elmegreen, D. M., Elmegreen, B. G., & Bellin, A. D. 1990, ApJ, 364, 415
Ferland, G. J., Korista, K. T., Verner, D. A., Ferguson, J. W., Kingdon, J. B. & Verner, E. M.
1998, PASP, 110, 761
Genzel, R., Lutz, D., Sturm, E., Egami, E., Kunze, D., Moorwood, A. F. M., Rigopoulou, D.,
Spoon, H. W. W., Sternberg, A., Tacconi-Garman, L. E., Tacconi, L., Thatte, N. 1998,
ApJ, 498, 579
Gerin, M., Combes, F., & Athanassoula, E., A&A, 230, 37
-- 22 --
Giuricin, G., Marinoni, C., Ceriani, L., & Pisani, A. 2000, ApJ, 543, 178
Gonz´alez Delgado, R. & Heckman, T. 1999, Ap&SS, 266, 187
Gonz´alez Delgado, R. M., Heckman, T., Leitherer, C., Meurer, G., Krolik, J., Wilson, A. S., Kinney,
A., Koratkar, A. , 1998, ApJ, 505, 174
Halpern, J. P. & Steiner, J. E. 1983, ApJ, 269, L37
Heisler, C. A., Lumsden, S. L., & Bailey J. A. 1997, Nature, 385, 700
Helou, G., Soifer, B. T. & Rowan-Robinson, M. 1985, ApJ, 298, L7
Ho, L. C., Filippenko, A. V. & Sargent, W. L. W. 1993, ApJ, 417, 63
Hunt, L. K., Malkan, M. A., Rush, B., Bicay, M. D., Nelson, B. O., Stanga, R. M., & Webb, W.
1999, ApJS, 125, 349
Hunter, D. A. & Gallagher, J. S. III 1986, PASP, 98, 5
Hunter, D. A., Gillet, F. C., Gallagher, J. S., Rice, W. L., Low, F. J. 1986, ApJ, 303, 171
Izotov, Y. I., Thuan, T. X. & Lipovetsky, V. A. 1994, ApJ, 435, 647
Karachentseva, V. E. & Karachentsev, I. D., 2000, Astronomy Reports, 44, 501
Keel, W. C. & Van Soest, E. T. M. 1992, A&AS, 94, 553
Kennicutt, R. C. Jr. 1998, ARA&A, 36, 189
Kennicutt, R. C. Jr., Edgar, B. K. & Hodge, P. W. 1989, ApJ, 337, 761
Kennicutt, R. C. Jr., Keel, W. C., & Blaha, C. A., AJ, 97, 1022
Kennicutt, R. C. Jr., Roettiger, K. A., Keel, W. C., van der Hulst, J. M., & Hummel, E. 1987, AJ,
93, 1011
Levenson, N. A., Weaver, K. A., & Heckman, T. M. 2001, ApJS, 133, 269
Leonardi, A. J. & Rose, J. A. 1996, AJ, 111, 182
Lorenz, H., Richter, G. M., Capaccioli. M., Longo, G. 1993, A&A, 277, 321
Lutz, D. 1992, A&A, 259, 462
Lutz, D., Spoon, H. W. W., Rigopoulou, D., Moorwood, A. F. M. & Genzel, R. 1998, ApJ, 505,
L103
M´arquez, I. & Moles, M. 1999, A&A, 344, 421
-- 23 --
McGaugh, S. S. 1991, ApJ, 380, 140
Mihalas, D. 1972, Non-LTE Model Atmospheres for B and O stars (Boulder: NCAR) (NCAR-
TN/STR-76)
Mihos, C. 1999, Ap&SS, 266, 195
Mihos, C. & Hernquist, L. 1994, ApJ, 431, L9
Mihos, C. & Hernquist, L. 1996, ApJ, 464, 641
Miley, G. K., Neugebauer, G., & Soifer, B. T. 1985, ApJ, 293, L11
Miller, J. S. & Mathews, W. G. 1972, ApJ, 172, 593
Moy, E. & Rocca-Volmerange, B. 2002, A&A, 383, 46
Mulchaey, J. S., Regan, M. W., & Kundu, A. 1997, ApJS, 110, 299
Noguchi, M. 1987, MNRAS, 228, 635
Osterbrock, D. E. 1989, Astrophysics of Gaseous Nebulae and Active Galactic Nuclei (Mill Valley,
CA: Univ. Science Books)
Pagel, B. E. J., Simonson, E. A., Terlevich, R. J. & Edmunds, M. G. 1992, MNRAS, 255, 325
Radovich, M., Hasinger, G., & Rafanelli, P. 1998, Astron. Nachr., 319, 325
Rafanelli, P., Violato, M. & Baruffolo, A. 1995, AJ, 109, 1546
Rafanelli, P., Temporin, S. & Baruffolo, A. 1997, Astron. Nachr., 318, 249
Raimann, D., Storchi-Bergmann, T., Bica, E., Melnick, J., & Schmitt, H. 2000 MNRAS, 316, 559
Raimann, D., Storchi-Bergmann, T., Gonz´alez Delgado, R. M., Cid Fernandes, R., Heckman, T.,
Leitherer, C., and Schmitt, H. 2002, MNRAS, submitted (preprint astro-ph/0210635)
Richter, G. M., Lorenz, H., Bohm, P., Priebe, A., & Capaccioli, M. 1991, Astron. Nachr., 312, 345
Rifatto, A., Rafanelli, P., Ciroi, S., Radovich, M., Vennik, J., Richter, G. M., and Birkle, K. 2001,
AJ, 122, 2301
Risaliti, G., Maiolino, R., & Salvati, M. 1999, ApJ, 522, 157
Rose, J. A. 1984, AJ, 89, 1238
Rush, B., Malkan, M. A., & Spinoglio, L. 1993, ApJS, 89, 1
Schlegel, D. J., Finkbeiner, D. P., Davis, M. 1998, ApJ, 500, 525
-- 24 --
Shields, J. C. & Filippenko, A. V. 1990, AJ, 100, 1034
Smith, R. C., Kirshner, R. P., Blair, W. P., Long, K. S., & Winkler, P. F. 1993, ApJ, 407, 564
Storchi-Bergmann, T., Mulchaey, J. S., & Wilson, A. S. 1992, ApJ, 395, 73
Storchi-Bergmann, T. & Pastoriza, M. G., 1990, PASP, 102, 1359
Telesco, C. M., Wolstencroft, R. D., & Done, C. 1988, ApJ, 329, 174
Toomre A. & Toomre J. 1972, ApJ, 178, 623
van den Bergh, S. 2002, AJ, 124, 782
Veilleux, S. & Osterbrock, D. E. 1987, ApJS, 63, 295
Young, J. S., Xie, S., Kenney, J. D. P. & Rice, W. L. 1989, ApJS, 70, 699
This preprint was prepared with the AAS LATEX macros v5.0.
-- 25 --
Fig. 1. -- Hα images of ESO381-G009 (left) and Tol1238-364 (right), after application of an adaptive
smooth filter.
-- 26 --
Fig. 2. -- B −R image of ESO381-G009 (upper left) and Tol1238-364 (lower right). Darker regions
are bluer
-- 27 --
Fig. 3. -- Contour maps of the Hα adaptive-smoothed images of Tol1238-364 (top) and ESO381-
G009 (bottom) with superimposed the traces of the slit positions and the regions extracted from
the spectra. Regions are labeled according to the explanations in the text.
-- 28 --
4000
5000
6000
7000
4000
5000
6000
7000
Fig. 4. -- Nuclear and extranuclear spectra of Tol1238-364, labeled according to Fig. 3.
-- 29 --
4000
5000
6000
7000
Fig. 5. -- Spectra of the regions extracted from ESO381-G009, labeled according to Fig. 3.
-- 30 --
4000
5000
6000
7000
Fig. 6. -- Spectra of the nuclear and circumnuclear regions extracted from the central portions N
and N' of Tol1238-364. The regions are labeled with increasing numbers from NW to SE.
.
-- 31 --
Fig. 7. -- Multi-gaussian fit of the blend Hα + [NII] λ6548,6563 for the nuclear region of Tol1238-
364. Fluxes in erg cm−2 s−1 A−1 are plotted versus the wavelength in A. The fit of the blend is
drawn with a solid line, while the individual fit components are drawn with dashed lines. A broad
Hα component was required to match the observed profile.
-- 32 --
TOL 1238-364
10
20
30
40
50
0.6
0.4
0.2
0
150
100
50
0
0
Fig. 8. -- Trend of the isophotal ellipticity (top) and position angle (bottom) with increasing radius
for Tol1238-364 in B (full circles), V (stars), and R (squares).
-- 33 --
ESO 381-G009
20
40
60
0.8
0.6
0.4
0.2
0
150
100
50
0
0
Fig. 9. -- Trend of the isophotal ellipticity (top) and position angle (bottom) with increasing radius
for ESO381-G009 in B (full circles), V (stars), and R (suqares).
-- 34 --
TOL 1238-364
15
20
25
1
0
-1
1.5
1
0.5
0
10
20
30
40
50
Fig. 10. -- From top to bottom: fit to the R-band surface brightness profile of Tol1238-364 with an
exponential bulge (dashed line) and an exponential disk (dash-dotted line) component; fit residuals,
and B −R radial profile.
-- 35 --
ESO 381-G009
20
40
60
15
20
25
1
0
-1
2
1.5
1
0.5
0
0
Fig. 11. -- From top to bottom: fit to the R-band surface brightness profile of ESO381-G009
with an exponential bulge (dashed line) and an exponential disk (dash-dotted line) component; fit
residuals, and B −R radial profile.
-- 36 --
n2
n1
A2 A1
n3
n4
n5
B1
B2
B3
A2'
n2'
n1'
n3'
n4'
n5'
A1'
B1'
C''
B3''
A3''
B2'
A2''
A1''
B1''
B2''
Fig. 12. -- Trend of the ionization degree with the radial distance, represented by the emission-line
ratios [O III] λ5007/Hβ and [N II] λ6583/Hα, for Tol1238-364. Positive radii indicate regions East
to the nucleus.
n2
N, n3
n4
n5
n1
-- 37 --
n2, n3
N
n4
n5
n2, n3
N
n4
n5
n1
B1
B3
A2
A1
B2
n1
B1
A1
B3
A2
B2
A1
B2
n1'
A2'
B2'
A1'
B1'
C''
B3''
A2''
A1''
B2''
N'
n2'
n5'
n3'
n4'
n3'
n4'
N', n2'
n1'
n5'
n3'
n4'
n2', N'
n5'
A2'
B2'
A1'
B1'
C''
B3''
B1''
B2''
A3''
A2''
A1''
n1'
A2'
B1'
B2'
A1'
C''
B3''
B1''
A3''
A2''
A1''
B2''
Fig. 13. -- Veilleux - Osterbrock (VO) diagnostic diagrams for all the regions extracted from the
spectra of Tol1238-364. S, L, and H indicate Seyfert, Liner, and H II areas of the diagrams; the
solid line sets the empirical separation between AGN and H II-like regions, while the dashed line
divides Seyferts from Liners. Diamonds indicate the extranuclear regions and the central regions at
the three P.A.; triangles indicate the nucleus and circumnuclear regions extracted from the central
kiloparsec.
-- 38 --
n4'
N'
n3'
n2'
n5'
n1'
A2'
B1'
A1'
B2'
B2'
B1'
A2'
A1'
n5'
n1'
n5'
n1'
A2'
A1'
B1'
B2'
n5'
n1'
A2'
B2'
A1'
B1'
n4'
N'
n3'
n2'
n3'n2'
n4'
N'
n3'
n2'
N'
n4'
N
n3 n4
n1
n5
A1
B2
n1
n5
N
N
n4
n5
n3 n4
n3
n3
N
n4
n5
B2
A1
n1
B1
A1
B2
B3
n1
B1
B3
A1
B2
C''
A2''
B3''
B2''
A1''
A2''
A1''
B2''
B3''
C''
C''
B1''
A1''
A3''
C''
B2''
B3''
A2''
B3''
B1''
A3''
A1''
A2''
B2''
Fig. 14. -- [OII] λ3727 - based diagnostic diagrams for all the regions of Tol1238-364. Symbols and
S, L, and H are as in Fig. 13; the corresponding areas of the diagrams are separated by dashed
lines.
-- 39 --
3800
4000
4200
4400
Fig. 15. -- Enlargement of the blue range of a few circumnuclear regions and two extranuclear
regions of Tol1238-364 showing different behaviors of metal and Balmer absorption lines, which
suggest different underlying stellar populations (see text for details).
-- 40 --
A2'
A3'
N'
A1'
B1
N
A1
A2
Fig. 16. -- Same as Fig. 12, but for ESO381-G009.
-- 41 --
N'
A1'
A3'
A2'
B1
N
A2
A1
N'
A1'
A2'
B
A3' A1
A2
A1'
N
N'
Fig. 17. -- VO diagnostic diagrams for ESO381-G009. Diamonds and triangles indicate regions
extracted from the spectra at P.A. = 125◦ and P.A. = 90◦, respectively.
-- 42 --
A1'
N'
A1'
N'
A2'
A3'
A1'
N'
A3'
N'
A2'
A1'
Fig. 18. -- Same as Fig. 14, but for ESO381-G009 at P. A. = 125◦.
-- 43 --
Fig. 19. -- Grid of photoionization models for H II regions calculated with the code CLOUDY using
the input parameters described in the text and four different values of metallicity (dotted lines) Z
= 0.1, 0.2, 0.3, and 0.4 Z⊙. The grid is represented onto VO-diagnostic diagrams together with
the observed logarithmic line ratios for the H II regions of Tol1238-364 (left) and ESO381-G009
(right). The photoionization parameter logU varies from −3.75 to −3.0 with a step of 0.25 from
bottom to top along the model lines. The observational points are mainly concentrated around the
line at Z = 0.2 Z⊙ for Tol1238-364 and between 0.1 and 0.2 Z⊙ for ESO381-G009. Circumnuclear
regions of Tol1238-364 within a radius of 0.5 kpc are compared with power-law ionization models
(dot-dashed lines) for three values of metallicity, 0.5, 1.0, and 1.5 Z⊙. logU changes from −3.00 to
−2.00 with a step of 0.25. The observational points appear in agreement with solar or supersolar
metallicities.
-- 44 --
0.14
0.12
0.1
0.08
0.06
0.04
0.02
ESO 381-G009
TOL 1238-364
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
0
50
100
150
200
Angle (degrees)
250
300
350
0
0
50
100
150
200
Angle (degrees)
250
300
350
Fig. 20. -- Angular and radial distribution of Hα fluxes normalized to the total Hα flux (excluding
the nucleus) of Tol1238-364 (left) and ESO381-G009 (right). Fluxes are summed within circular
sectors with aperture 20◦. Horizontal dashed bars indicate fluxes within a radius of 2 kpc, inclined
dashed bars indicate fluxes between 2 and 4 kpc, and empty bars indicate fluxes at radii > 4 kpc.
The horizontal bars indicate the range of angles in which each galaxy faces the companion.
-- 45 --
Fig. 21. -- IRAS co-added contour-maps of the galaxy pair at 60 and 100 µm (top) and the
corresponding high resolution HIRAS contour-maps (bottom), overimposed to the Digitized Sky
Survey optical image; 1 and 2σ (dashed lines), and 3, 4, 6, 10, 30, 50, 100, 200σ (solid lines) contour
levels are drawn. Right Ascension and Declination (1950.0) are displayed along x and y axes.
-- 46 --
Fig. 22. -- Infrared color-color diagrams, similar to the ones in Dopita et al. (1998), but with the
two galaxies represented as distinct sources: ESO381-G009 (square) clearly appears as a "warm"
starburst, while Tol1238-364 (star ) is dominated by a moderately obscured AGN.
-- 47 --
Table 1. Galaxy Properties
Object
α(J2000.0)
( h m s)
(◦ ′ ′′)
δ(J2000.0) Radial Velocitya MB
(mag)
(km s−1)
b
Tol1238-364
ESO 381-G009
ESO 381-G006
12 40 52.9 −36 45 22
12 40 58.4 −36 43 55
12 40 40.8 −36 44 20
3282
3288
3101
−20.2
−19.3
−17.2
Note. -- Units of right ascension are hours, minutes, and seconds, and units of declination are
degrees, arcminutes, and arcseconds.
aObtained from the H I 21 cm line Barnes & Webster (2001).
bConverted from apparent magnitudes given in the NASA Extragalactic Database.
Table 2. Observation Summary
Object
Date
U.T.
Exp. Time P.A.
Spectral Range
Filter
(hh:mm)
(sec)
(A)
a
µlim
(mag arcsec−2)
b
mlim
(mag)
Tol1238-364
"
"
"
"
"
ESO381-G009
"
"
1995 Apr. 1
1995 Apr. 1
1995 Apr. 3
1995 Apr. 3
1995 Apr. 4
1995 Apr. 4
1995 Apr. 3
1995 Apr. 3
1995 Apr. 3
1995 Apr. 1
1995 Apr. 1
1995 Apr. 1
1995 Apr. 1
1995 Apr. 1
09:23
09:47
05:55
06:47
06:28
07:04
06:23
07:59
08:16
07:49
08:02
08:10
07:05
07:23
Spectroscopy
150◦
"
90◦
"
146◦
"
90◦
125◦
"
Imaging
· · ·
· · ·
· · ·
· · ·
· · ·
3850 - 7950
3200 - 6030
3850 - 7950
3200 - 6030
3850 - 7950
3200 - 6030
3850 - 7950
3850 - 7950
3200 - 6030
· · ·
· · ·
· · ·
· · ·
· · ·
900
1200
900
900
1800
2400
900
900
900
600
300
300
900
900
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
B
V
R
Hαb
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
23.79
23.96
24.11
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
22.79
23.13
23.12
3.56×10−16
8.47×10−16
Hα-cont.
· · ·
· · ·
aLimit surface brightnesses are evaluated at a 3σ level above the background, and the limit magnitudes are evaluated assuming
a 3σ signal within the PSF radius corresponding to 80% of the total flux.
bFor the Hα image we give the brightness and flux limit in units of ergs cm−2 s−1.
--
4
8
--
-- 49 --
Table 3. Photometric Results
Object
B
V
R
(mag)
(mag)
(mag)
< re > < hD > < B/T >
(arcsec)
(arcsec)
Tol1238-364
ESO381-G009
12.92
12.69
13.71
13.48
12.45
12.28
13.22
13.05
11.98
11.84
12.74
12.60
2.2
3.0
7.5
0.2
12.5
0.12
Note. -- For each galaxy, observed magnitudes are listed on the first row
and magnitudes corrected for Galactic extinction on the second row. Extinction
values in V and R were derived from AB Burstein & Heiles (1982) following
Cardelli, Clayton, & Mathis (1989) and assuming a visual selective extinction
RV = 3.1.
-- 50 --
Table 4. Tol1238-364: Ionizing Photons and SFR
Id.
Ra
(kpc)
L(Hα)
Qion
Q′
nuc
EW(Hα)
SFR
SFR/pc2
(×1040 erg s−1)
(phot. s−1)
(phot. s−1)
(A)
(M⊙ yr−1)
(M⊙ yr−1 pc−2)
A2
A1
n1
n2
n3
n4
n5
B1
B2
B3
−3.90
−1.60
−0.70
−0.30
0.0
0.26
0.51
0.93
1.83
3.43
A2' −3.71
A1' −1.53
−0.81
n1'
−0.39
n2'
0.26
n3'
n4'
0.37
0.67
n5'
1.74
B1'
B2'
4.13
A3" −5.10
A2" −2.95
A1" −1.37
C"
0.66
0.97
B1"
1.83
B2"
B3"
3.90
1.57
2.31
0.68
21.21
19.45
0.89
0.22
1.39
1.45
1.14
0.92
1.35
0.97
2.43
2.69
0.64
0.77
2.05
0.86
0.38
1.61
2.19
0.83
2.15
0.76
1.07
1.15E+52
1.69E+52
4.98E+51
1.55E+53
1.42E+53
6.49E+51
1.59E+51
1.01E+52
1.06E+52
8.32E+51
6.73E+51
9.85E+51
7.06E+51
1.77E+52
1.96E+52
4.69E+51
5.61E+51
1.50E+52
6.27E+51
2.76E+51
1.17E+52
1.60E+52
6.05E+51
1.57E+52
5.58E+51
7.80E+51
P.A. = 150◦
2.20E+51
5.74E+51
8.28E+51
4.53E+52
2.82E+53
3.61E+52
1.22E+52
6.59E+51
3.91E+51
1.75E+51
P.A. = 146◦
2.48E+51
4.84E+51
7.40E+51
2.68E+52
3.61E+52
2.36E+52
9.04E+51
6.19E+51
1.86E+51
P.A. = 90◦
1.07E+51
2.06E+51
9.12E+51
1.88E+52
7.80E+51
3.88E+51
2.11E+51
94.93
31.31
9.39
53.96
106.62
50.57
12.77
16.70
46.95
87.16
36.28
43.68
9.42
25.38
49.09
17.83
9.86
35.73
98.28
64.92
94.16
57.72
11.71
24.21
27.01
28.00
0.111
0.164
0.048:
· · ·
· · ·
· · ·
0.015:
0.098
0.103
0.081
0.065
0.095
0.068:
· · ·
· · ·
· · ·
0.054:
0.145
0.061
0.027
0.114
0.155
0.059:
0.152
0.054
0.076
7.47E−8
2.50E−7
2.65E−7:
· · ·
· · ·
· · ·
1.06E−7:
3.86E−7
1.77E−7
8.87E−8
4.27E−8
1.88E−7
3.13E−7:
· · ·
· · ·
· · ·
2.99E−7:
1.73E−7
4.29E−8
2.16E−8
1.42E−7
2.03E−7
1.61E−7:
4.65E−7
9.30E−8
5.33E−8
aDistance in kpc to the center of the galaxy. Negative distances indicate regions West or North-West of the
nucleus.
Note. -- Q′
nuc values are calculated taking as reference the value given in bold-face in correspondence of the
central region n3, and obtained as explained in the text. SFR and SFRD values marked with a colon are upper
limits.
-- 51 --
Table 5. ESO381-G009: Ionizing Photons and SFR
Id.
Ra
(kpc)
L(Hα)
Qion
N(O5)
EW(Hα)
SFR
SFR/pc2
(×1040 erg s−1)
(phot. s−1)
(A)
(M⊙ yr−1)
(M⊙ yr−1 pc−2)
A2 −3.10
A1 −1.94
N
B1
0.0
1.74
A3' −5.93
A2' −3.40
A1' −1.89
N'
B1'
19.35
0.0
0.23
1.43
5.52
0.37
0.44
0.40
1.28
3.53
0.09
P.A. = 90◦
33
208
805
54
P.A. = 125◦
65
58
187
516
14
1.65E+51
1.04E+52
4.03E+52
2.72E+51
3.24E+51
2.91E+51
9.35E+51
2.58E+52
6.80E+50
28.41
15.47
76.07
32.60
82.61
45.92
58.24
66.08
16.83
0.016
0.101
0.390
0.026
0.031
0.028
0.091
0.250
0.007
9.36E-9
1.01E-7
4.49E-7
1.76E-8
2.77E-8
3.60E-8
1.09E-7
2.68E-7
2.59E-9
aDistance in kpc to the center of the galaxy. Negative distances indicate regions West of the nucleus.
-- 52 --
Table 6. FIR Data and Dust Content
Quantity
Tol1238-364
ESO381-G009
F12 (Jy)
F25 (Jy)
F60 (Jy)
F100(Jy)
0.580 ± 0.065
2.198 ± 0.140
7.942 ± 0.130
13.591 ± 0.431
0.125 ± 0.035
0.204 ± 0.085
1.122 ± 0.102
1.378 ± 0.296
log(F25/F12)
log(F60/F25)
log(F100/F25)
log(F60/F12)
0.578 ± 0.076
0.558 ± 0.034
0.791 ± 0.041
1.136 ± 0.055
0.212 ± 0.302
0.740 ± 0.220
0.829 ± 0.274
0.953 ± 0.161
α60,25
α100,60
-1.476 ± 0.090
-1.051 ± 0.094
-1.947 ± 0.580
-0.402 ± 0.598
Tdust (K)
Mdust (M⊙)
39.5
5.5×106
LFIR (L⊙)
SFR (M⊙ yr−1)
3.04×1010
15.86
45.1
3.0×105
3.24×109
1.69
-- 53 --
Table A1. Tol1238-364 - P.A. = 150◦: Flux Ratios
Line
λ
A1
A2
N
B1
B2
B3
−9.7 -- −3.7
−23.5 -- −9.7
−3.7 -- 3.0
3.0 -- 5.4
5.4 -- 10.8
10.8 -- 19.2
[OII]
[NeIII]
Hγ
[OIII]
HeII
[OIII]
[OIII]
[FeVII]
[FeVI]
[NI]
[CaV]
HeI
[OI]
[OI]
[NII]
Hα
[NII]
[SII]
[SII]
[SII]
3727
3869
4340
4363
4686
4959
5007
5158
5177
5199
5309
5876
6300
6364
6548
6563
6583
6717
6724
6731
Hbeta 4861 = 1.00
F(Hα) (10−15 erg/s/cm2)
3.05 ± 0.98
4.86
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
0.55 ± 0.10
2.35 ± 0.96
2.00 ± 0.70
4.14 ± 1.12
0.84
4.50
0.39 ± 0.14
2.16 ± 1.36
0.56
0.12 ± 0.02
0.15
0.11 ± 0.04
0.14
0.13 ± 0.03
0.14
3.81
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
2.69
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
5.22
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
0.24 ± 0.06
2.41 ± 0.07
0.39 ± 0.23
0.22 ± 0.09
0.38 ± 0.11
0.24
2.34
0.37
0.22
0.37
0.71 ± 0.20
0.81 ± 0.11
7.38 ± 0.15
1.55 ± 0.45
0.51 ± 0.12
0.95 ± 0.15
0.67
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
0.79
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
7.03
0.08 ± 0.03
0.07
0.11 ± 0.03
0.10
0.18 ± 0.03
0.16
0.13 ± 0.03
0.11
1.44
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
0.49
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
0.93
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
0.16 ± 0.07
0.16 ± 0.04
0.18 ± 0.02
0.37 ± 0.16
0.09 ± 0.04
0.38 ± 0.08
0.11
0.17 ± 0.06
0.10
· · ·
· · ·
0.13
· · ·
· · ·
· · ·
· · ·
0.13
0.81 ± 0.03
0.52
0.32 ± 0.03
0.20
0.22
· · ·
· · ·
· · ·
· · ·
0.07
0.12 ± 0.07
0.09
· · ·
· · ·
0.32
· · ·
· · ·
· · ·
· · ·
0.55 ± 0.12
0.27 ± 0.05
1.27 ± 0.04
1.43 ± 0.29
0.30 ± 0.06
0.30 ± 0.07
0.32
0.20
0.77
0.66
0.21
0.23
4.97 ± 0.70
3.89 ± 0.27
4.72 ± 0.09
6.20 ± 0.93
4.06 ± 0.37
3.76 ± 0.34
2.85
2.85
2.85
2.85
2.85
2.85
1.90 ± 0.30
1.14 ± 0.10
4.16 ± 0.08
2.98 ± 0.51
1.45 ± 0.15
1.18 ± 0.14
1.08
0.83
2.50
1.36
1.01
0.89
1.06 ± 0.19
0.61 ± 0.07
1.17 ± 0.04
1.57 ± 0.36
0.85 ± 0.10
0.67 ± 0.10
0.59
0.44
0.68
0.69
0.58
0.50
1.73 ± 0.33
1.01 ± 0.12
2.48 ± 0.07
2.61 ± 0.63
1.42 ± 0.18
1.01 ± 0.16
0.95
0.72
1.45
1.14
0.97
0.75
0.67 ± 0.14
0.40 ± 0.06
1.31 ± 0.04
1.04 ± 0.28
0.57 ± 0.08
0.35 ± 0.06
0.37
0.29
0.76
0.45
0.39
0.26
27.2 ± 0.3
30.4 ± 0.3
238.8 ± 0.0
10.4 ± 0.1
25.7 ± 0.3
23.6 ± 0.2
84.0
57.2
664.8
50.4
52.8
41.4
F(Hβ) (10−15 erg/s/cm2)
5.5 ± 0.7
7.8 ± 0.5
50.6 ± 1.0
1.7 ± 0.2
6.3 ± 0.5
6.3 ± 0.5
ca
E(B-V)
aExtinction coefficient
29.5
0.73
0.51
20.1
0.41
0.28
233.2
0.66
0.46
17.7
1.02
0.71
18.5
0.47
0.32
14.5
0.36
0.25
Note. -- In this and the following tables, bold face numbers indicate flux ratios and absolute fluxes corrected for
internal reddening. Below the label of every region, the distance in arcsec of the region's extremes to the intensity peak
along the slit are indicated; positive distances are in East direction.
-- 54 --
Table A2. Tol1238-364 - P.A. = 146◦: Flux Ratios
Line
λ
A1'
A2'
N'
B1'
B2'
−8.7 -- −4.0 −22.8 -- −8.7
−4.0 -- 3.7
3.7 -- 11.4
11.4 -- 24.5
[OII]
[NeIII]
Hγ
[OIII]
HeII
[OIII]
[OIII]
[NI]
HeI
[OI]
[OI]
[NII]
Hα
[NII]
[SII]
[SII]
[SII]
3727
3869
4340
4363
4686
4959
5007
5199
5876
6300
6364
6548
6563
6583
6717
6724
6731
Hbeta 4861 = 1.00
F(Hα) (10−15 erg/s/cm2)
3.74 ± 0.30
4.14 ± 0.87
1.61 ± 0.16
2.74 ± 0.36
3.62 ± 0.40
4.89
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
5.92
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
2.58
0.82 ± 0.26
1.24
· · ·
· · ·
0.17 ± 0.07
0.21
· · ·
· · ·
3.75
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
4.03
· · ·
· · ·
0.48 ± 0.09
0.51
· · ·
· · ·
· · ·
· · ·
0.22 ± 0.04
0.37 ± 0.12
2.27 ± 0.14
0.23 ± 0.06
0.30 ± 0.04
0.22
0.36
2.20
0.22
0.30
0.69 ± 0.06
1.10 ± 0.19
7.19 ± 0.36
0.64 ± 0.08
0.87 ± 0.06
0.67
· · ·
· · ·
0.16 ± 0.03
0.13
1.06
· · ·
· · ·
· · ·
· · ·
6.81
0.15 ± 0.05
0.13
0.62
· · ·
· · ·
0.86
· · ·
· · ·
0.25 ± 0.04
0.15 ± 0.03
0.12 ± 0.03
0.17
0.12
0.11
0.10 ± 0.03
0.21 ± 0.06
0.88 ± 0.07
0.18 ± 0.04
0.13 ± 0.04
0.08
· · ·
· · ·
0.14
· · ·
· · ·
0.54
0.30 ± 0.05
0.18
0.13
· · ·
· · ·
0.12
· · ·
· · ·
0.48 ± 0.04
0.40 ± 0.10
1.51 ± 0.09
0.53 ± 0.06
0.37 ± 0.04
0.35
0.26
0.86
0.36
0.33
3.93 ± 0.16
4.37 ± 0.44
5.00 ± 0.25
4.15 ± 0.25
3.24 ± 0.13
2.85
2.85
2.85
2.85
2.85
1.43 ± 0.07
1.60 ± 0.19
4.52 ± 0.23
1.51 ± 0.11
0.95 ± 0.05
1.03
1.04
2.56
1.03
0.83
0.56 ± 0.05
1.09 ± 0.13
1.48 ± 0.09
0.85 ± 0.07
0.61 ± 0.05
0.40
0.69
0.81
0.57
0.53
1.05 ± 0.09
1.76 ± 0.23
2.97 ± 0.18
1.41 ± 0.11
0.97 ± 0.09
0.74
1.11
1.63
0.94
0.85
0.49 ± 0.05
0.68 ± 0.10
1.49 ± 0.09
0.56 ± 0.05
0.37 ± 0.04
0.35
0.43
0.82
0.37
0.32
25.5 ± 0.3
14.1 ± 0.3
89.9 ± 0.9
34.7 ± 0.3
24.1 ± 0.2
49.0
33.5
281.2
74.4
31.2
F(Hβ) (10−15 erg/s/cm2)
6.5 ± 0.2
3.2 ± 0.3
18.0 ± 0.7
8.4 ± 0.4
7.4 ± 0.2
ca
E(B−V)
aExtinction coefficient.
17.2
0.42
0.29
11.7
0.56
0.39
98.7
0.74
0.51
26.1
0.49
0.34
11.0
0.17
0.12
-- 55 --
Table A3. Tol1238-364 - P.A. = 90◦: Flux Ratios
Line
λ
A1"
A2"
A3"
C"
B1"
B2"
B3"a
−9.4 -- −2.3 −16.8 -- −9.4 −28.2 -- −16.8
−2.3 -- 1.1
1.1 -- 4.1
4.1 -- 9.5
9.5 -- 22.6
[OII]
[NeIII]
Hγ
[OIII]
HeII
[OIII]
[OIII]
[NI]
HeI
[OI]
[OI]
[NII]
Hα
[NII]
[SII]
[SII]
[SII]
3727
3869
4340
4363
4686
4959
5007
5199
5876
6300
6364
6548
6563
6583
6717
6724
6731
Hbeta 4861 = 1.00
F(Hα) (10−15 erg s−1 cm−2)
2.15 ± 0.34
3.01 ± 0.57
3.08 ± 1.48
4.05 ± 1.34
1.73 ± 0.54
3.40 ± 0.92
9.37 ± 4.69
3.06
0.64 ± 0.35
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
4.43
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
3.49
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
6.19
1.83 ± 1.15
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
2.88
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
4.72
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
17.60
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
0.25 ± 0.05
0.34 ± 0.06
0.48 ± 0.17
1.66 ± 0.46
0.46 ± 0.13
0.24 ± 0.04
0.99 ± 0.50
0.24
0.33
0.48
1.61
0.44
0.23
0.95
0.80 ± 0.08
1.02 ± 0.09
1.14 ± 0.22
5.38 ± 1.08
1.41 ± 0.20
0.78 ± 0.20
3.48 ± 1.15
0.77
· · ·
· · ·
0.98
· · ·
· · ·
0.14 ± 0.04
0.15 ± 0.05
0.11
0.11
0.09 ± 0.02
0.17 ± 0.04
0.06
· · ·
· · ·
0.11
· · ·
· · ·
1.12
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
5.12
· · ·
· · ·
1.33
· · ·
· · ·
0.40 ± 0.24
0.22 ± 0.05
0.29
0.46 ± 0.20
0.29
· · ·
· · ·
0.15
· · ·
· · ·
· · ·
· · ·
0.75
· · ·
· · ·
· · ·
· · ·
3.24
· · ·
· · ·
· · ·
· · ·
0.21 ± 0.08
0.48 ± 0.24
0.15
· · ·
· · ·
0.25
· · ·
· · ·
0.34 ± 0.04
0.37 ± 0.04
0.30 ± 0.09
1.76 ± 0.40
0.57 ± 0.09
0.32 ± 0.10
0.33 ± 0.19
0.22
0.23
0.26
1.06
0.31
0.22
0.16
4.34 ± 0.22
4.52 ± 0.23
3.31 ± 0.46
4.73 ± 0.90
5.25 ± 0.42
4.22 ± 0.51
6.06 ± 1.70
2.85
2.85
2.85
2.85
2.85
2.85
2.85
1.58 ± 0.09
1.38 ± 0.08
0.99 ± 0.18
3.85 ± 0.77
2.41 ± 0.22
1.73 ± 0.24
2.33 ± 0.77
1.03
0.87
0.85
2.31
1.30
1.16
1.09
0.79 ± 0.06
0.74 ± 0.06
0.77 ± 0.17
1.33 ± 0.33
0.98 ± 0.12
0.94 ± 0.18
1.79 ± 0.63
0.50
0.45
0.66
0.77
0.51
0.62
0.80
1.29 ± 0.12
1.30 ± 0.10
1.31 ± 0.30
2.58 ± 0.65
1.70 ± 0.22
1.60 ± 0.34
3.05 ± 1.13
0.82
0.79
1.12
1.50
0.88
1.05
1.36
0.50 ± 0.06
0.56 ± 0.05
0.54 ± 0.14
1.25 ± 0.31
0.72 ± 0.10
0.66 ± 0.15
1.26 ± 0.49
0.32
0.34
0.46
0.73
0.37
0.43
0.56
33.8 ± 0.3
22.9 ± 0.2
10.1 ± 0.1
10.8 ± 0.2
22.6 ± 0.2
12.5 ± 0.3
8.4 ± 0.3
79.5
58.4
13.7
30.1
78.2
27.8
38.8
F(Hβ) (10−15 erg s−1 cm−2)
7.8 ± 0.3
5.1 ± 0.2
3.1 ± 0.4
2.3 ± 0.4
4.3 ± 0.3
3.0 ± 0.3
1.4 ± 0.3
cb
E(B−V)
27.9
0.55
0.39
20.5
0.61
0.42
4.8
0.20
0.14
10.6
0.67
0.46
27.5
0.80
0.56
9.8
0.52
0.36
13.6
0.99
0.69
aThe emission-line ratios for region B3" are somewhat uncertain because of difficulties in appropriately correcting Hβ for the underlying
stellar absorption. This additional source of uncertainty is not contained in the errors quoted in the table.
bExtinction coefficient.
Table A4. Tol1238-364: Flux Ratios of the Subnuclear Regions
Line
λ
n1
n2
P.A. = 150◦
n3
n4
n5
n1'
n2'
n3'
n4'
n5'
P.A. = 146◦
−3.7 -- −2.0
−2.0 -- −0.3
−0.3 -- 0.7
0.7 -- 1.7
1.7 -- 3.0
−4.0 -- −2.0
−2.0 -- −0.3
−0.3 -- 0.7
0.7 -- 2.0
2.0 -- 3.7
[OII]
[NeIII]
Hγ
[OIII]
HeII
[OIII]
[OIII]
[FeVII]
[FeVI]
[NI]
[CaV]
[NII]
HeI
[FeVII]
[OI]
[OI]
[NII]
Hα
[NII]
[SII]
[SII]
[SII]
3727
3869
4340
4363
4686
4959
5007
5158
5177
5199
5309
5755
5876
6086
6300
6364
6548
6563
6583
6717
6724
6731
4.00 ± 3.56
6.87
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
0.08 ± 0.02
0.51 ± 0.10
4.18 ± 1.00
6.14 ± 2.27
1.05 ± 0.18
0.68 ± 0.07
2.15 ± 0.32
5.58 ± 1.79
0.15
0.51
4.28
13.20
1.79
0.96
2.71
10.89
0.22 ± 0.13
0.52 ± 0.13
1.59 ± 0.48
0.38
0.52
0.16 ± 0.02
0.30 ± 0.05
0.22
0.30
0.11 ± 0.03
0.12 ± 0.03
0.19 ± 0.05
0.12 ± 0.04
0.14
0.12
1.62
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
1.18
· · ·
· · ·
1.27 ± 0.61
0.22 ± 0.13
· · ·
· · ·
· · ·
· · ·
0.74 ± 0.19
0.44 ±0.08
1.14 ± 0.35
0.60
1.40
0.21 ± 0.03
0.12 ± 0.04
0.25
0.13
0.12 ±0.03
0.14 ± 0.04
0.25 ± 0.11
0.13 ±0.06
0.15
0.14
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
1.07 ± 0.59
2.57 ± 0.26
2.62 ± 0.05
1.96 ± 0.14
1.58 ± 0.28
0.75 ± 0.41
2.30 ± 0.16
2.39 ± 0.07
2.06 ± 0.16
1.19 ± 0.45
1.03
2.40
2.51
1.96
1.58
0.71
2.21
2.33
2.03
1.13
2.69 ± 0.89
8.38 ± 0.67
8.05 ± 0.08
5.83 ± 0.29
4.61 ± 0.60
3.32 ± 1.00
7.14 ± 0.43
7.49 ± 0.15
6.29 ± 0.38
3.77 ± 0.90
2.53
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
7.50
· · ·
· · ·
· · ·
· · ·
7.49
5.83
0.06 ± 0.01
0.08 ± 0.03
0.05
· · ·
· · ·
0.08
0.08 ± 0.03
0.08
0.26 ± 0.08
0.15 ± 0.02
0.14 ± 0.03
0.20
· · ·
· · ·
· · ·
· · ·
0.13
0.14
0.06 ± 0.02
0.07 ± 0.02
0.05
0.07
0.08 ± 0.02
0.09 ± 0.02
0.05
0.09
4.60
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
3.04
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
6.71
· · ·
· · ·
· · ·
· · ·
7.20
0.05 ± 0.02
0.05
· · ·
· · ·
6.12
· · ·
· · ·
· · ·
· · ·
0.15 ± 0.04
0.16 ±0.02
0.12 ± 0.03
0.13
· · ·
· · ·
· · ·
· · ·
0.14
· · ·
· · ·
· · ·
· · ·
0.11
· · ·
· · ·
· · ·
· · ·
0.26 ± 0.08
0.21 ± 0.02
0.10 ± 0.01
0.16 ± 0.08
0.56 ± 0.63
0.23 ±0.05
0.22 ± 0.02
0.20 ± 0.06
0.12
· · ·
· · ·
0.13
0.11 ± 0.03
0.06
0.10
· · ·
· · ·
0.16
· · ·
· · ·
0.31
· · ·
· · ·
0.15
· · ·
· · ·
0.17
0.09 ± 0.02
0.07
0.17
· · ·
· · ·
3.49
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
--
5
6
--
0.77 ± 0.42
1.15 ± 0.15
1.01 ± 0.03
0.37 ± 0.03
0.33 ± 0.12
0.50 ± 0.32
0.81 ± 0.08
0.88 ± 0.04
0.80 ± 0.10
0.61 ± 0.27
0.44
· · ·
· · ·
0.42
0.52
0.37
0.40 ± 0.10
0.37 ± 0.02
0.15 ± 0.03
0.14
0.19
0.15
0.32
· · ·
· · ·
0.22
· · ·
· · ·
0.46
0.61
0.63
0.29 ± 0.06
0.29 ± 0.03
0.25 ± 0.07
0.16
0.20
0.19
0.30
...
· · ·
1.06 ± 0.39
2.53 ± 0.23
1.65 ± 0.03
0.51 ± 0.04
0.81 ± 0.16
1.36 ± 0.50
1.58 ± 0.13
1.30 ± 0.04
1.10 ± 0.10
0.94 ± 0.36
0.56
0.81
0.78
0.51
0.79
0.55
0.84
0.86
0.84
0.42
5.44 ± 1.36
9.00 ± 0.63
6.06 ± 0.06
1.83 ± 0.09
2.93 ± 0.38
7.12 ± 1.71
5.40 ± 0.32
4.32 ± 0.09
3.76 ± 0.23
6.34 ± 1.33
2.85
2.85
2.85
1.83
2.85
2.85
2.85
2.85
2.85
2.85
3.10 ± 0.84
7.22 ± 0.58
5.06 ± 0.05
1.59 ± 0.08
2.24 ± 0.31
4.13 ± 1.07
5.00 ± 0.30
4.19 ± 0.08
3.61 ± 0.22
4.25 ± 0.98
1.61
2.26
2.36
1.59
2.18
1.64
2.62
2.75
2.73
1.90
1.34 ± 0.48
1.92 ± 0.17
1.27 ± 0.03
0.47 ± 0.03
0.92 ± 0.17
1.91 ± 0.63
1.55 ± 0.11
1.30 ± 0.04
1.35 ± 0.11
1.66 ± 0.46
0.67
0.56
0.57
0.47
0.89
0.72
0.78
0.83
1.00
0.71
2.49 ± 0.92
4.04 ± 0.36
2.77 ± 0.06
0.95 ± 0.07
1.66 ± 0.33
3.40 ± 1.16
3.05 ± 0.21
2.70 ± 0.08
2.67 ± 0.21
2.81 ± 0.84
1.25
1.18
1.24
0.95
1.61
1.28
1.54
1.73
1.99
1.19
1.15 ± 0.44
2.12 ± 0.19
1.49 ± 0.03
0.48 ± 0.03
0.75 ± 0.16
1.49 ± 0.54
1.51 ± 0.11
1.40 ± 0.04
1.32 ± 0.11
1.14 ± 0.36
0.57
0.62
0.66
0.48
0.73
0.56
0.76
0.90
0.98
0.48
Hbeta 4861 = 1.00
F(Hα) (10−16 erg/s/cm2 )
F(Hβ) (10−16 erg/s/cm2 )
ca
66.6 ± 2.0
746.5 ± 0.0
1528.0 ± 0.0
322.7 ± 3.2
74.7 ± 2.2
54.7 ± 1.6
240.9 ± 2.4
419.5 ± 0.0
132.9 ± 1.3
55.1 ± 1.7
247.5
7702.0
7065.0
322.7
79.0
351.0
881.4
975.8
233.2
279.1
12.3 ± 2.7
83.0 ± 5.8
252.2 ± 2.5
176.8 ± 7.1
25.5 ± 2.5
7.7 ± 1.6
44.6 ± 2.2
97.2 ± 1.9
35.3 ± 1.8
8.7 ± 1.6
86.9
0.85
2703.0
1.51
2479.0
0.99
176.8
0.00
27.7
0.04
123.1
1.20
309.3
0.84
342.6
0.55
81.8
0.36
97.9
1.05
Table A4 -- Continued
Line
λ
n1
n2
n3
n4
n5
n1'
P.A. = 150◦
P.A. = 146◦
n3'
n2'
n4'
n5'
−3.7 -- −2.0
−2.0 -- −0.3
−0.3 -- 0.7
0.7 -- 1.7
1.7 -- 3.0
−4.0 -- −2.0
−2.0 -- −0.3
−0.3 -- 0.7
0.7 -- 2.0
2.0 -- 3.7
E(B-V)
0.59
1.05
0.69
0.00
0.03
0.84
0.59
0.38
0.25
0.73
aExtinction coefficient.
--
5
7
--
Table A5. ESO381-G009 - P.A. = 90◦: Flux Ratios
Line
λ
A1
A2
N
B1
−14.4 -- −3.6
−32.9 -- −14.4
−3.6 -- 5.8
5.8 -- 21.8
[OIII]
[OIII]
[NI]
HeI
[OI]
[OI]
[NII]
Hα
[NII]
[SII]
[SII]
[SII]
4959
5007
5199
5876
6300
6364
6548
6563
6583
6717
6724
6731
Hbeta 4861 = 1.00
F(Hα) (10−16 erg/s/cm2)
F(Hβ) (10−16 erg/s/cm2)
ca
E(B−V)
aExtinction coefficient.
· · ·
· · ·
· · ·
· · ·
0.31 ± 0.06
0.30
· · ·
· · ·
1.07 ± 0.66
0.94 ± 0.64
1.06 ± 0.08
1.72 ± 0.71
0.96
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
0.81 ± 1.44
0.26
0.89
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
1.00
· · ·
· · ·
0.20 ± 0.04
0.14
· · ·
· · ·
· · ·
· · ·
0.44 ± 0.04
0.25
1.59
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
8.78 ± 3.86
4.75 ± 2.00
5.01 ± 0.20
6.59 ± 1.58
2.85
2.85
2.85
2.85
2.23 ± 1.36
1.11 ± 0.79
1.13 ± 0.07
1.96 ± 0.76
0.72
0.66
0.64
0.84
2.23 ± 1.56
1.08 ± 0.79
0.61 ± 0.05
1.09 ± 0.51
0.67
0.63
0.33
0.45
4.57 ± 3.15
2.36 ± 1.63
1.10 ± 0.10
2.27 ± 1.04
1.37
1.37
0.60
0.93
2.35 ± 1.60
1.28 ± 0.84
0.49 ± 0.05
1.17 ± 0.53
0.70
0.74
0.27
0.48
61.8 ± 7.4
606.0
7.0 ± 2.3
212.7
1.48
1.03
34.1 ± 3.1
745.5 ± 0.0
28.8 ± 1.7
96.2
2343.0
158.1
7.2 ± 2.4
148.8 ± 6.0
4.4 ± 0.8
33.8
0.67
0.47
821.9
0.74
0.52
55.5
1.10
0.77
58
Table A6. ESO381-G009 - P.A.=125◦: Flux Ratios
Line
λ
A1'
A2'
A3'
N'
−12.8 -- −3.7
−21.2 -- −12.8
−33.3 -- −21.2
−3.7 -- 6.4
[OII]
[NeIII]
Hγ
[OIII]
HeII
[OIII]
[OIII]
[NI]
HeI
[OI]
[OI]
[NII]
Hα
[NII]
[SII]
[SII]
[SII]
3727
3869
4340
4363
4686
4959
5007
5199
5876
6300
6364
6548
6563
6583
6717
6724
6731
Hbeta 4861 = 1.00
F(Hα) (10−15 erg/s/cm2)
F(Hβ) (10−15 erg/s/cm2)
ca
E(B-V)
aExtinction coefficient.
2.80 ± 0.53
5.58 ± 2.96
6.08 ± 3.47
2.27 ± 0.30
3.73
0.89 ± 0.59
1.14
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
8.82
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
8.42
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
3.20
0.81 ± 0.21
1.09
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
0.26 ± 0.06
0.75 ± 0.40
0.47 ± 0.30
0.27 ± 0.03
0.25
0.73
0.46
0.26
0.63 ± 0.07
1.70 ± 0.48
1.41 ± 0.49
0.87 ± 0.05
0.61
· · ·
· · ·
0.14 ± 0.03
0.11
0.14 ± 0.04
0.10
· · ·
· · ·
1.61
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
1.36
· · ·
· · ·
0.84
· · ·
· · ·
0.23 ± 0.15
0.20 ± 0.02
0.18
· · ·
· · ·
· · ·
· · ·
0.15
0.06 ± 0.03
0.04
· · ·
· · ·
0.32 ± 0.05
0.24 ± 0.19
0.23 ± 0.41
0.38 ± 0.03
0.23
0.14
0.16
0.25
4.01 ± 0.20
4.93 ± 1.04
4.21 ± 1.26
4.30 ± 0.17
2.85
2.85
2.85
2.85
1.23 ± 0.09
1.16 ± 0.36
0.78 ± 0.53
1.08 ± 0.05
0.87
0.67
0.53
0.71
0.90 ± 0.08
1.35 ± 0.54
0.91 ± 0.46
0.57 ± 0.05
0.63
0.75
0.60
0.37
1.59 ± 0.16
2.25 ± 0.99
1.58 ± 0.85
1.04 ± 0.08
1.10
1.25
1.04
0.67
0.68 ± 0.07
0.90 ± 0.45
0.68 ± 0.40
0.47 ± 0.04
0.47
0.50
0.45
0.30
27.2 ± 0.3
54.4
6.8 ± 0.3
19.1
0.45
0.31
5.6 ± 0.2
17.0
1.1 ± 0.2
5.9
0.72
0.50
8.5 ± 0.8
65.1 ± 0.7
18.8
150.1
2.0 ± 0.4
15.1 ± 0.5
6.6
0.51
0.36
52.6
0.54
0.38
59
|
astro-ph/0301148 | 1 | 0301 | 2003-01-09T11:37:10 | Constraining the Parameters of AGN Jets - Comparisons with Herbig-Haro Jets | [
"astro-ph"
] | Comparing the properties AGN and Herbig-Haro jets can be a useful exercise for understanding the physical mechanisms at work in collimated outflows that propagate in such different environments. In the case of Herbig-Haro jets, the presence of emission lines in the spectra and the continuous evolution of the observation techniques greatly favor our knowledge of the physical parameters of the jets instead, for AGN jets, the process of constraining the jet parameters is hampered by the nature of the emission from these objects that is non-thermal. I will discuss how one cannot directly constrain the basic parameters of extragalactic jets by observations but must treat and interpret the data either by statistical means or by comparing observed and simulated morphologies in order to gain some indications on the values of these parameters. | astro-ph | astro-ph |
Constraining the Parameters of AGN Jets
Comparisons with Herbig-Haro Jets
Silvano Massaglia
Dipartimento di Fisica Generale dell'Universit`a, Via Pietro Giuria 1, 10125
Torino, Italy
November 19, 2002
Abstract. Comparing the properties AGN and Herbig-Haro jets can be a useful
exercise for understanding the physical mechanisms at work in collimated outflows
that propagate in such different environments. In the case of Herbig-Haro jets,
the presence of emission lines in the spectra and the continuous evolution of the
observation techniques greatly favor our knowledge of the physical parameters of
the jets instead, for AGN jets, the process of constraining the jet parameters is
hampered by the nature of the emission from these objects that is non-thermal.
I will discuss how one cannot directly constrain the basic parameters of ex-
tragalactic jets by observations but must treat and interpret the data either by
statistical means or by comparing observed and simulated morphologies in order to
gain some indications on the values of these parameters.
Keywords: Active Galactic Nuclei, Herbig-Haro objects, Jets, Radio Sources
1. Introduction
The main difficulty one faces when trying to comprehend the nature
of jets from Active Galactic Nuclei is due to the absence of lines in
the radiation spectrum of these objects. In fact, their emission is typ-
ically non-thermal emission (synchrotron or inverse Compton). This
very simple fact constitutes a sort of "original sin" of AGN jets and is
basically the reason why after almost five decades since the discovery
of the radio emission from Cygnus A (Jennison & Das Gupta 1953),
observers and theorists are still debating about very basic questions
such as extragalactic jets composition, i.e. whether they are made of
ordinary matter or of electron-positron pairs.
Models of AGN jets, that consider the jet overall dynamics, depend
on a minimum of three parameters: the Lorentz factor, the Mach num-
ber, the jet-to-ambient density contrast. Another crucial parameter for
jet modeling would be the magnetic field intensity, however one may
assume, at a zeroth order approximation, that the bulk kinetic energy
density in the jet dominates on the magnetic one and that the general
behavior of the jet propagation can be captured by a hydrodynamic
model. In any event, none of the above jet parameters can be directly
constrained by observations.
c(cid:13) 2018 Kluwer Academic Publishers. Printed in the Netherlands.
porto.tex; 4/06/2018; 9:40; p.1
2
Silvano Massaglia
In contrast, Herbig-Haro jets show copious emission lines in their
spectra, in the optical and infrared bands mainly. Spectral lines give
information on local temperature and density, on the bulk velocity of
the jet emitting matter and on the presence of shocks along the jet.
Moreover, the augmented angular resolution and sensitivity achievable
with the recent telescopes (see Bacciotti, this issue) allow to investigate
the velocity structure across the jet and reveal the jet characteristics
close to the central source.
In this review I will present efforts to constrain the basic parameters
in extragalactic jets, both by statistical and numerical means, discuss
the role of these efforts in interpreting the FR I/FR II radio source di-
chotomy, and made comparisons with observations and interpretations
of Herbig-Haro jets. For an extended review on jet modelling, I remind
the reader to Ferrari (1998).
In the next Section I briefly introduce the bases of the Fanaroff-Riley
classification of extragalactic radio sources and recall the interpre-
tations proposed; in Section 3 I examine a method for deriving jet
velocities; in Sections 4 and 5 I discuss the problem of constraining
the jet Mach number , density and composition and in Section 6 make
comparisons between Herbig-Haro and AGN jets.
2. The Fanaroff-Riley classification: Taxonomy
Historically, the extragalactic radio sources have been classified into
two categories (Fanaroff & Riley 1974) based upon their radio mor-
phology: a first class of objects, preferentially found in rich clusters
and hosted by weak-lined galaxies, shows jet-dominated emission and
two-sided jets and was named FR I (Fig. 1); a second one, found iso-
lated or in poor groups and hosted by strong emission-line galaxies,
presents lobe-dominated emission and one-sided jets and was called
FR II (or "classical doubles") (Fig. 2). Besides morphology, FR I and
FR II radio sources were discriminated in power as well: objects below
100 W Hz−1 str−1 were typically referred as FR I sources.
∼ 2 × 1025h2
A perhaps more illuminating criterion has been found by Owen &
Ledlow (1994) who plotted the radio luminosity against the optical
absolute magnitude of the host galaxy: they found the bordering line
of FR I to FR II regions correlating as LR ∝ L1.7
opt, i.e. in a luminous
galaxy more radio power is required to form a FR II radio sources.
This correlation is important since can be interpreted as an indication
that the environment may play a crucial role in determining the source
structure. The above argument yields the basic question of the origin
of FR I/FR II dichotomy, whether intrinsic or ambient driven.
porto.tex; 4/06/2018; 9:40; p.2
Parameters of AGN Jets
3
Figure 1. VLA image of the FRI source 3C449 (Feretti et al. 1999)
Figure 2. VLA image of the FRII source Cygnus A
Two kind of explanations for the FR I/FR II dichotomy can be
found in the literature: intrinsic and extrinsic interpretations (see the
review by Wiita 2002). Among the intrinsic explanations, the jet com-
position was invoked to interpret this dichotomy: Celotti & Fabian
(1993) argued that FR II radio jets were made of ordinary matter
(e+ − p) and Reynolds et al. (1996a) instead FR I jets (in particular
M87) were made of e+ − e− pairs. Another possibility was considered
by Wilson & Colbert (1995) and Meier (1999) who suggested that FR
II jets originated from rapidly rotating black-holes; the structure of the
porto.tex; 4/06/2018; 9:40; p.3
4
Silvano Massaglia
accretion was also considered to explain the dichotomy: according to
Reynolds et al. (1996b) when accretion is advection dominated (ADAF)
FR I jets results, while standard accretion disks generate FR II jets. The
extrinsic explanations assume, apart from the total power, FR I and FR
II jets are basically similar close to the nucleus and that differences in
the environment are able to destabilize, possibly via onset of turbulence
in the flow, and decelerate FR I jets effectively, while FR II jets succeed
to propagate, nearly unchanged, up to the working surface to produce
the hot-spots (Bicknell 1995, Komissarov 1990, Bowman et al. 1996).
A possible clue for discriminating among these two kind of inter-
pretations (Gopal-Krishna & Wiita 2000) was the observations of six
HYbrid MOrphology Radio Sources (HYMORS) that show FR I mor-
phology on one side of the core and FR II morphology on the other
one: this is a clear indication that the environment play a basic role in
determining the radio source appearance.
3. The Jet Velocity
As discussed before, none of the basic parameters (Mach number, jet-
ambient density ratio and jet Lorentz factor) required for analytical and
numerical modeling of jets can be directly constrained by observations
of radio sources. Observers must therefore interpret their data relying
upon statistical analyzes and look at these data through an assumed
basic model. This procedure is typically employed for the interpretation
of the jet one-sideness and superluminal velocities that are observed in
several radio sources at milliarcsecond scales using VLBI techniques.
The basic model adopted can be synthesized as follows: responsible for
the emission is a distribution of relativistic electrons that are advected
at relativistic speed by the jet, assuming homogeneous and isotropic
jets the flux ratio of the approaching jet to the receding one (Doppler
boosting) is:
Fa
Fr
= 1 + βj cos θ
1 − βj cos θ!2+α
,
(1)
Where Fa, Fr are the fluxes of the approaching and receding jet respec-
tively, βj is the jet bulk velocity in units of c, θ is the angle between
the jet axis and the line of sight and F ∝ ν −α, with α ∼ 0.5 typically.
If the emission region moves towards the observer at the jet speed with
the angle θ, photons emitted at a later time must travel a distance
to the observer that is shorter than for photons emitted earlier, this
makes the apparent time interval of the emission smaller and thus the
porto.tex; 4/06/2018; 9:40; p.4
Parameters of AGN Jets
apparent speed βapp possibly larger than one:
βapp =
βj sin θ
1 − βj cos θ
.
5
(2)
The observation of the jet-counterjet flux ratio Fa/Fr and of βapp in a
source would yield the actual jet speed and inclination. Unfortunately
this is not possible in most cases: typically the counterjet is too faint to
be visible and superluminal proper motions are detected in a relatively
few objects only.
Jets in AGNs are clearly highly non isotropic objects, accordingly
their appearance crucially depends on the viewing angle. This question
is discussed in the unified model for AGNs (Antonucci 1993, Urry &
Padovani 1995). According to this model the appearance of an AGN
is dictated by the jet axis orientation with respect to the line of sight:
radio-loud quasars and FR II radio galaxies are the same kind of objects
seen with increasing angle (high-luminosity unified scheme), while BL
Lac objects and FR I radio sources also belong to the same class viewed
at increasing angles (low-luminosity unified scheme). Therefore one may
ask the following question: when observing with the highest possible
resolution, i.e. with VLBI techniques, a sample of different kind on
AGNs jets showing mixed FR I and FR II morphologies at kpc-scales,
is their jet Lorentz factor distribution mirroring the various large scale
morphologies or not?
For attempting to answer this question a possible way, as mentioned
before, is to employ a statistical analysis. Giovannini et al. (1988)
collected a sample of 187 radio galaxies, observed with the VLA, and
plotted their core (5 GHz) against total (408 MHz) power. Arguing that
the objects of this large sample had jets randomly oriented with the
line of sight, they expected that data were scattered about a best-fit
line that correlated core against total radio power and corresponded to
the mean orientation angle of 60◦ to the line of sight. The correlation
found was:
log Pc(60◦) = 0.62 log Pt + 7.6 .
(3)
For investigating the jet properties close to the AGN, Giovannini et al.
(1994, 2001) examined a complete sample of 27 radio galaxies limited
in flux observed with the VLBI. The sample included FR I and FR II
sources in nearly the same quantity. They again plotted the observed
core at 5 GHz, and affected by Doppler boosting, against the total radio
power at 408 MHz, therefore not boosted, of the objects (Fig. 3), with
the correlation of Eq. (3) given as a comparison. Since both proper
motions and jet-counterjet flux ratios were typically not available for
the sources of the sample, therefore Eqs. (1) and (2) could not be
porto.tex; 4/06/2018; 9:40; p.5
6
Silvano Massaglia
Figure 3. Observed arcsecond core radio power at 5 GHz against the total radio
power at 408 MHz. Solid line represents the correlation found by Giovannini et al.
(1988)
solved for θ and βj, Giovannini et al. (2001) fixed γ and obtained θ
from data, once assuming α = 0, not a bad approximation for the
core fluxes. In this way they could constrain the Doppler factor δ (=
[γ(1 − βj cos θ)]−1) and deboost the observed core power to obtain the
intrinsic one, according to:
Pc obs = Pc intr × δ2 .
(4)
The correlation in Eq. (3) was scaled as well in the same way to obtain
the intrinsic core power and in Fig. 4 one can see the result of this
procedure after having set γ = 5. Repeating the same trick for different
values of γ, Giovannini et al. (2001) found that the correlation line
went nicely through the data for γ = 3 − 10, outside this interval the
correlation failed, as in Fig. 3, independently of the FR I/FR II type of
the sources. Giovannini et al. (2001) concluded that FR I and FR II,
despite showing different kpc-scale morphologies, have all the Lorentz
factors in the range 3-10 on the parsec scale.
4. Jet Mach Number
According to the above reasoning, both class of AGN jets are rela-
tivistic, and thus supersonic, on parsec scale. On the kpc-scale FR II
jets may still be relativistic in most cases due to the observed one-
sideness, if this is interpreted as Doppler boosting. A clear indication
porto.tex; 4/06/2018; 9:40; p.6
Parameters of AGN Jets
7
Figure 4. The solid line represents the correlation from Eq. (3), after the Pc(60◦)
has been scaled. Bullets show the intrinsic core power vs the total radio power. γ = 5
has been assumed for deboosting
of supersonic speed is the presence of shocks in the FR I jets, that can
be resolved in the transverse direction, and the signature of shocks is
the behavior of the polarization vector. The M 87 jet is among the
best studied objects since originate from the closest AGN and detailed
polarization maps have been obtained in the radio and optical bands
(Perlman et al. 1999). The magnetic field is parallel to the jet axis and
becomes perpendicular in the regions where we see emission knots, such
as knot A of the M 87 jet that is shown in detail in Fig. (5). This is
a clear indication of a shock compression of the field component along
the shock extent.
Jets of FR II radio sources are too faint to produce resolved po-
larization maps. Polarization can be easily detected in the prominent
radio lobes and hot spots, where the jet terminate into the intergalactic
medium. Again one expects a field direction oriented along the lobe's
border, i.e. the working surface of a supersonic jet at its terminal point.
In Fig. (6) the polarization map of the western hot-spot of Pictor A is
shown (Perley et al. 1997), and the projected magnetic field is perpen-
dicular to the electric field vector shown in the figure. The magnetic
field results aligned with the shocked working surface.
porto.tex; 4/06/2018; 9:40; p.7
8
a
r
X
i
v
:
a
s
t
r
o
-
p
h
/
9
9
0
1
1
7
6
1
3
J
a
n
1
9
9
9
Silvano Massaglia
C
E
S
C
R
A
6.5
6.0
5.5
5.0
4.5
4.0
3.5
3.0
C
E
S
C
R
A
6.5
6.0
5.5
5.0
4.5
4.0
3.5
3.0
LOP-1
B1
HOP-1
B2
HOP-2
HOP-3
I
A
-10
-11
-12
-13
ARC SEC
M87 - OPTICAL POLARIZATION
-14
-15
-10
-11
Figure 5
-12
-13
ARC SEC
M87 - RADIO POLARIZATION
-14
-15
Figure 5. Polarization maps of the M 87 jet (the projected magnetic field is given);
optical (top panel) and radio (bottom panel)
Figure 6. Polarization map of the western lobe of Pictor A (the electric field vector
is given)
porto.tex; 4/06/2018; 9:40; p.8
Parameters of AGN Jets
9
5. Jet Density and Composition
We have seen that jets are likely relativistic on parsec scale, and that
remain supersonic on kpc scale, provided the direction and intensity of
the magnetic field is correctly interpretated in terms of shock compres-
sion. A third parameter to constrain is the jet density, or more properly
the jet-to-ambient density contrast. For constraining this parameter one
has to rely upon numerical simulations and compare qualitatively the
spatial density distribution, resulting from the calculations, with the
observed brightness distributions.
In Fig. 7 I show the radio brightness distribution of the FR II
source Cygnus A and point out the different regions of interest for
the model interpretation: the radio hot-spots are the regions where the
jets terminate ("splash points"), the lobes are the manifestation of the
back-flow, and the diffuse region between the lobes is the "cocoon",
typically invisible in radio but evident in X-ray (see Smith et al. 2002).
When simulations of supersonic, underdense jets are carried out (see
e.g. Massaglia et al. 1996) the overall picture that emerges from these
studies can be summarized as follows (see Fig. 8): the deceleration
of the jet flow at its head is accomplished through the formation of
a strong shock (Mach disk), which partially thermalizes the jet bulk
kinetic energy; the overpressured shocked jet material forms a back-
flow along the sides of the jet and inflates a cocoon whose size becomes
more prominent as the density ratio between jet and ambient material
decreases; finally, a second shock (bow-shock) is driven into the external
medium. Simulations of supersonic, overdense jets yield qualitatively
different morphologies that are close to Herbig-Haro jets. Therefore
AGN jets, at least those found in FR II sources, are underdense with
respect to external medium and the density ratio can be very small
(∼ 10−5) in case pair plasma jets.
The question of the jet density reminds the problem of establishing
whether AGN jets are made of proton and relativistic electrons or
instead of relativistic electron-positron pairs. This of the jet compo-
sition is a long outstanding and yet unsolved problem for the simple
reason that electrons and positron emit synchrotron radiation of iden-
tical spectrum and linear polarization. A possible observational clue for
discriminating between e− − p and e+ − e− jets is the detection of radio
circular polarization (CP). In fact (see Wardle et al. 1998), the large
linear polarization degree observed in AGN jets at parsec scales has
survived the internal Faraday rotation process, that is mainly due to
electrons with γmin < 100. Therefore e− − p jets should have relativistic
electrons with γmin ≥ 100, while e+ − e− jets do not suffer Faraday
rotation since electron and positron gyrate in opposite directions and
porto.tex; 4/06/2018; 9:40; p.9
10
Silvano Massaglia
Figure 7. Observed brightness distribution in the FR II source Cygnus A
thus γmin ≥ 1. Values of γmin ≪ 100 should be a clear signature of
e+ − e− jets. Low-energy electrons can yield to Faraday conversion of
linear into circular polarization, besides producing Faraday rotation.
For an e+ − e− jet internal Faraday rotation should not be present,
as mentioned before, and a small but appreciable fraction (below 1%)
of the linear polarization should be converted into CP by low energy
electrons and positrons since this mechanism is sign-independent, being
proportional to the electron gyrofrequency to the square. Observation
of CP should be thus be a signature of the presence of low-energy
emitting particles that do not cause Faraday rotation and maintain a
high degree of LP, therefore should be e+ − e− plasma.
CP has been indeed observed in about 20 AGN jets (Wardle &
Homan 2001) and we should have a clear indication on the e+ − e−
nature of the jet composition. Unfortunately, there are other possible
origins of CP, besides Faraday conversion: intrinsic CP, scintillation
and coherent radiation mechanisms. Therefore the question of the jet
composition remains open.
Authors have attempted different ways for solving this important
and basic problem. Scheck et al. (2002) have simulated relativistic e+ −
e− and e− − p jets of given kinetic luminosities and jet-ambient density
ratios: they have found that both the morphology and the dynamical
behavior is almost independent of the assumed jet composition.
porto.tex; 4/06/2018; 9:40; p.10
Parameters of AGN Jets
11
Figure 8. Simulated density distribution for a relativistic, supersonic and under-
dense jet
6. AGN and HH Jets
AGN and Herbig-Haro jets (see a recent review by Reipurth & Bally
2001) share some similarities but many basic differences, besides the
different scale powers and sizes. Both are collimated outflows that
originated from accretion processes and are accelerated, possibly by
MHD mechanisms. Since jet velocity must be of the order of the escape
velocity from the central gravitational well, AGN jets are likely to be
relativistic, at least in their initial part, while HH jets have velocities
comparable with stellar winds. Both kind of jets appear knotty and ter-
minate into the ambient medium forming structures whose morphology
that can be interpreted as bow-shocks, that indicates supersonic bulk
velocity in the jets.
HH jets are observed in the optical and infrared bands where they
emit optically thin thermal radiation. Their spectra are rich of emission
lines, mainly of hydrogen, sulfur and oxygen with intensities, ratios and
profiles are clearly indicative of shocks, since these observational values
can be related to the predictions from plane-parallel and bow-shock
models. Combining the data of shock velocities, Doppler shifts of the
emission lines and proper motions it is possible to constrain the jet
velocity, density and temperature (i.e. the Mach number). Typically,
the jet velocity is ∼ 200 − 400 km−1, density ∼ 103 − 104 cm−3 and
Mach number ≈ 20 − 40.
AGN jets instead emit continuous, non-thermal radiation in a wide
frequency range that goes from the radio up to the X-ray bands; these
jets are likely relativistic on parsec scale but may decelerate down to
non-relativistic speeds at kpc scales, at least in the case of FR I objects,
but are likely to remain supersonic both FR I and FR II jets. The basic
reason why shocks lead to thermal emission in HH objects and non-
thermal radiation in AGN jets is due to about 6-8 orders of magnitude
ratio in the ambient density: in the case of HH jets propagate in the
high density ambient of molecular clouds, thus shocks can heat the
matter and produce radiative losses, while extragalactic jets propagate
porto.tex; 4/06/2018; 9:40; p.11
12
Silvano Massaglia
in the tenuous intergalactic or intracluster medium therefore shocks,
that can be considered with good approximation adiabatic, have the
main effect to accelerate relativistic particles via Fermi-like processes of
the first kind that, in turn, yield synchrotron radiation in the shocked
ambient magnetic field and the only possible signature we observe is
the reorientation of the polarization vector of the radiation.
References
Antonucci, R. ARA&A, 31:473, 1993.
Bicknell, G. V. ApJS, 264:228, 1995.
Bowman, M., J. P. Leahy and S. S. Komissarov. MNRAS, 279:899, 1996.
Celotti, A. and A. C. Fabian. MNRAS, 101:29, 1993.
Fanaroff, B. L. and J. M. Riley. MNRAS, 167:31, 1974.
Feretti, L., R. A. Perley, G. Giovannini and H. Andernach. A&A, 341:29, 1999.
Ferrari, A. ARA&A, 36:539, 1998.
Giovannini, G., L. Feretti, T. Venturi and P. Parma. A&A, 199:73, 1988.
Giovannini, G., L. Feretti, L. Lara, J. Marcaide, M. Rioja, S. R. Spangler and A. E.
Wehrle ApJ, 435:116, 1994.
Giovannini, G., W. D. Cotton, L. Feretti, L. Lara, T. Venturi. ApJ, 552:508, 2001.
Gopal-Krishna, S. and P. J. Wiita. A&A, 363:507, 2000.
Jennison, R. C. and M. K. Das Gupta. Nature, 172:996, 1953.
Komissarov, S. S. Ap&SS, 171:105, 1990.
Massaglia, S., G. Bodo and A. Ferrari. A&A, 307:997, 1996.
Meier, D. L. ApJ, 522:753, 1999.
Owen, F. N. and M. J. Ledlow.
In G. V. Bicknell, M. A. Dopita and P. J. Quinn,
editors, The First Stromlo Symposium: The Physics of Active Galaxies. ASP
Conference Series, 54:319, 1994.
Perley, R. A., H-J. Roser and K. Meisenheimer. A&A, 283:873, 1996a.
Perlman, E. S., J. A. Biretta, F. Zhou, W. B. Sparks and D. Macchetto. AJ, 117:2185,
1999.
Reipurth, B. and J. Bally. ARA&A, 39:403, 2001.
Reynolds, C. S., A. C. Fabian, A. Celotti and M. J. Rees. MNRAS, 328:12, 1997.
Reynolds, C. S., T. Di Matteo, A. C. Fabian, U. Kwang and C. R. Canizares.
MNRAS, 283:111, 1996b.
Scheck, L., M. A. Aloy, J. M. Mart´ı, J. L. G´omez and E. Muller. MNRAS, 331:615,
2002.
Smith, D. A., A. S. Wilson, S. Andrew, K. A. Arnaud, Y. Terashima and A. J.
Young. ApJ, 565:195, 2002.
Urry, C. M. and P. Padovani. PASP, 107:803, 1995.
Wardle, J. F. C., D. C. Homan, R. Ojha and D. H. Roberts. Nature, 395:457, 1998.
Wardle, J. F. C. and D. C. Homan. In R. A. Laing and K. M. Blundell, editors, Par-
ticles and Fields in Radio Galaxies. ASP Conference Series, astro-ph/0011515,
in press, 2001.
Wiita, P. J. Cosmic Radio Jets, astro-ph/0103020, 2002.
Wilson, A. S. and E. J. M. Colbert. ApJ, 438:62, 1995.
porto.tex; 4/06/2018; 9:40; p.12
|
astro-ph/0509424 | 1 | 0509 | 2005-09-15T08:36:41 | Multi-Wavelength Studies of the Optically Dark Gamma-Ray Burst 001025A | [
"astro-ph"
] | We identify the fading X-ray afterglow of GRB 001025A from XMM-Newton observations obtained 1.9-2.3 days, 2 years, and 2.5 years after the burst. The non-detection of an optical counterpart to an upper limit of R=25.5, 1.20 days after the burst, makes GRB 001025A a ``dark'' burst. Based on the X-ray afterglow spectral properties of GRB 001025A, we argue that some bursts appear optically dark because their afterglow is faint and their cooling frequency is close to the X-ray band. This interpretation is applicable to several of the few other dark bursts where the X-ray spectral index has been measured. The X-ray afterglow flux of GRB 001025A is an order of magnitude lower than for typical long-duration gamma-ray bursts. The spectrum of the X-ray afterglow can be fitted with an absorbed synchrotron emission model, an absorbed thermal plasma model, or a combination thereof. For the latter, an extrapolation to optical wavelengths can be reconciled with the R-band upper limit on the afterglow, without invoking any optical circumburst absorption, provided the cooling frequency is close to the X-ray band. Alternatively, if the X-ray afterglow is due to synchrotron emission only, seven magnitudes of extinction in the observed R-band is required to meet the R-band upper limit, making GRB 001025A much more obscured than bursts with detected optical afterglows. Based on the column density of X-ray absorbing circumburst matter, an SMC gas-to-dust ratio is insufficient to produce this amount of extinction. The X-ray tail of the prompt emission enters a steep temporal decay excluding that the tail of the prompt emission is the onset of the afterglow (abridged). | astro-ph | astro-ph |
Multi-Wavelength Studies of the Optically Dark Gamma-Ray
Burst 001025A
K. Pedersen,1 K. Hurley,2 J. Hjorth,1 D. A. Smith,3 M. I. Andersen,4 L. Christensen,4
T. Cline,5 J. P. U. Fynbo,1 J. Goldsten,6 S. Golenetskii,7 J. Gorosabel,8 P. Jakobsson,1
B. L. Jensen,1 B. Milvang-Jensen,9 T. McClanahan,5 P. Møller,10 V. Palshin,7
N. Schartel,11 J. Trombka,5 M. Ulanov,7 D. Watson1
ABSTRACT
We identify the fading X-ray afterglow of GRB 001025A from XMM-Newton
observations obtained 1.9 -- 2.3 days, 2 years, and 2.5 years after the burst. The
non-detection of an optical counterpart to an upper limit of R = 25.5, 1.20 days
after the burst, makes GRB 001025A a "dark" burst. Based on the X-ray af-
terglow spectral properties of GRB 001025A, we argue that some bursts appear
1Dark Cosmology Centre, Niels Bohr Institute, University of Copenhagen, Juliane Maries Vej 30, DK-2100
Copenhagen Ø, Denmark; kp, jens, jfynbo, pallja, brian j, darach @astro.ku.dk
2Space Sciences Laboratory, 7 Gauss Way, University of California, Berkeley, CA 94720-7450, USA;
[email protected]
3NSF Postdoctoral Fellow, University of Michigan, Department of Physics Ann Arbor, MI 48109, USA;
[email protected]
4Astrophysikalisches Institut Potsdam, An der Sternwarte 16, 14482 Potsdam, Germany; mandersen,
lchristensen @aip.de
5NASA Goddard Space Flight Center Code 661 Greenbelt, MD 20771, USA; Thomas.L.Cline, Timo-
thy.P.McClanahan, Jacob.I.Trombka.1 @nasa.gov
6The
Johns Hopkins University Applied Physics Laboratory Laurel, MD 20723, USA;
[email protected]
7Ioffe Physico-Technical Institute, St. Petersburg, 194021, Russia; golen, val, mulanov @mail.ioffe.ru
8IAA-CSIC, P.O. Box 03004, 18080 Granada, Spain and Space Telescope Science Institute, 3700 San
Martin Drive, Baltimore, MD21218, USA; [email protected]
9Max-Planck-Institut
fur
extraterrestrische Physik, Giessenbachstr., D-85748 Garching; mil-
[email protected]
10European Southern Observatory, Karl Schwarzschild-Strasse 2, D-85748 Garching bei Munchen, Ger-
many; [email protected]
11XMM-Newton Science Operations Centre, ESA, Villafranca del Castillo, P.O. Box 50727, 28080 Madrid,
Spain; [email protected]
-- 2 --
optically dark because their afterglow is faint and their cooling frequency is close
to the X-ray band. This interpretation is applicable to several of the few other
dark bursts where the X-ray spectral index has been measured. The X-ray af-
terglow flux of GRB 001025A is an order of magnitude lower than for typical
long-duration gamma-ray bursts. The spectrum of the X-ray afterglow can be
fitted with an absorbed synchrotron emission model, an absorbed thermal plasma
model, or a combination thereof. For the latter, an extrapolation to optical wave-
lengths can be reconciled with the R-band upper limit on the afterglow, without
invoking any optical circumburst absorption, provided the cooling frequency is
close to the X-ray band. Alternatively, if the X-ray afterglow is due to syn-
chrotron emission only, seven magnitudes of extinction in the observed R-band is
required to meet the R-band upper limit, making GRB 001025A much more ob-
scured than bursts with detected optical afterglows. Based on the column density
of X-ray absorbing circumburst matter, an SMC gas-to-dust ratio is insufficient
to produce this amount of extinction. The X-ray tail of the prompt emission en-
ters a steep temporal decay excluding that the tail of the prompt emission is the
onset of the afterglow. To within the astrometric uncertainty, this afterglow was
coincident with an extended object, seen in a deep VLT R-band image, which
we identify as the likely host galaxy of GRB 001025A.
Subject headings: gamma rays: bursts -- X-rays: general
1.
INTRODUCTION
One of the enduring mysteries of cosmic gamma-ray bursts (GRBs) is that of the so-
In the
called "dark" bursts - those with no detected optical or near-infrared afterglows.
fireball model of GRBs (for a review, see Meszaros 2002), an explosive event produces a
relativistically expanding blast wave. Internal shocks within the blast wave are the source of
the high energy prompt emission while afterglows arise from external shocks when the blast
wave run into an interstellar medium.
GRBs may be divided into "long-duration" and "short-duration" categories; no unam-
bigous detection of an optical afterglow has yet been made for any short-duration event
(e.g. Bloom et al. 2005; Hjorth et al. 2005; Gorosabel et al. 2002; Hurley et al. 2002) and
only in one case has an X-ray afterglow been detected (Gehrels et al. 2005). On the other
hand, virtually all of the long-duration bursts display fading soft X-ray afterglows (e.g. Costa
1999; De Pasquale et al. 2003), and they have been shown to be associated with supernova
explosions (Hjorth et al. 2003b; Stanek et al. 2003; Malesani et al. 2004).
-- 3 --
For more than half of the long-duration GRBs no optical and/or radio afterglow has
been detected (e.g. Fynbo et al. 2001; Lazzati et al. 2002; Berger et al. 2002); however, rapid
and systematic follow-up of bursts has decreased the relative fraction of dark bursts in recent
years (Rol et al. 2005; Berger et al. 2005). Recently, Jakobsson et al. (2004) suggested using
the value of the optical-to-X-ray spectral index as a diagnostic for early identification of
dark bursts, and they found five dark bursts (including GRB 001025A) in their sample of
52 bursts.
There are several possible explanations for dark bursts (Fynbo et al. 2001). They could
be at very high redshifts (Lamb & Reichart 2000), or they could be heavily obscured by dust
in their host galaxies (e.g. Groot et al. 1998; Taylor et al. 1998; Reichart & Price 2002). They
could also occur in regions where the surrounding medium is tenuous, far from the centers
of their host galaxies, where the strong shocks required for particle acceleration could not
form (Kumar & Panaitescu 2000).
Observations of bursts detected by the BeppoSAX, the High Energy Transient Explorer,
and the Rossi X-ray Timing Explorer spacecrafts have indeed shown that bursts can be
dark for a variety of reasons. Some bursts seem in fact to be dark due to extinction, e.g.
GRB 000210 (Piro et al. 2002), GRB 970828 (Djorgovski et al. 2001), and GRB 001109
(Castro Cer´on et al. 2004). Other bursts have faint optical afterglows either because of
moderately high redshift, e.g., GRB 020124 (Hjorth et al. 2003a; Berger et al. 2002), or
because they are intrinsically faint without extinction, e.g. GRB 980613 (Hjorth et al. 2002)
and GRB 021211 (Crew et al. 2003). Analysis of bursts detected by Swift will undoubtly shed
more light on the nature of dark bursts. Already now it has become evident that afterglows
of bursts detected by Swift on average are fainter than afterglows from bursts detected by
previous instruments (1.7 magnitudes in the optical/near-infrared (Berger et al., 2005) and a
factor three in X-rays Jakobsson, 2005). The rapid fading of some Swift bursts demonstrates
that at least these become intrinsically faint after the first few hundred seconds (Tagliaferri
et al. 2003).
Here we examine the case of the dark burst GRB 001025A. We report on our temporal
and spectral analysis of the prompt emission, on our very deep optical upper limits on the
afterglow, and on the results of our temporal and spectral analysis of the X-ray afterglow.
Finally, we discuss the nature of GRB 001025A within the framework of the fireball model,
and the implications for dark bursts in general.
-- 4 --
2. OBSERVATIONS
2.1. GRB 001025A Prompt Emission
GRB 001025A was first reported as an event detected by the Rossi X-Ray Transient
Explorer (RXTE) All Sky Monitor (ASM) at 11405 s UT October 25, 2000 (Smith et al.
2000). Its duration in the ASM 1.5 -- 12 keV energy band was ∼15 s, and its peak flux was ∼
4 Crab (10−7 erg cm−2 s−1) in the 5 -- 12 keV band. The event was detected by a single camera
and localized to a 4′ × 1.◦6 error box. It was also observed by Ulysses, the X-Ray/Gamma-
Ray Spectrometer (XGRS) aboard the Near Earth Asteroid Rendezvous mission (NEAR),
and Wind (Konus experiment) in the Interplanetary Network (IPN), and triangulated to a
preliminary 5.6′ wide annulus which crossed the ASM error box to form a 25 arcminute2
error box (Hurley, Cline & Smith 2000; Hurley 2000).
We have used the final Ulysses, NEAR, and Wind-Konus data to obtain a refined IPN
error box for this burst. The Ulysses - Konus annulus is centered at R.A.(2000)= 169.◦8656,
decl.(2000) = −56.◦6466, and has a radius of 53.◦5052 ± 0.◦0073 (3σ). The Ulysses - NEAR
annulus is centered at R.A.(2000) = 145.◦4502, decl.(2000)= −36.◦9464, and has a radius of
27.◦9969 ± 0.◦0275 (3σ). The intersection of these annuli defines a 96 arcminute2 error box.
However, the intersection of the Ulysses-Konus annulus with the ASM error box defines a
3.3 arcminute2 error box, whose coordinates are given in Table 1. The combined ASM-IPN
error box is shown in Figure 5.
The ASM, Konus, and XGRS lightcurves are shown in Figure 1. With a T90 duration
of 2.9 s (i.e., the time to accumulate between 5% and 95% of the photons) in the 50 -- 200 keV
energy range, this burst falls into the "long-duration" category. Figure 1 shows the lightcurve
in several energy ranges, and indicates the intervals used for time-resolved spectral analysis.
This figure also shows the hardness ratio as a function of time. A hard-to-soft evolution,
which is often observed in GRBs (e.g. Preece et al. 1998), is evident. The energy spectra are
presented in Figure 2. The Band (Band et al. 1993) model, which consists of two smoothly
joined power laws, was used to perform spectral fits. The Band function is:
f (E) = (cid:26) A(E/100)α exp (−E/E0)
A{(α − β)E0/(100)}(α−β) exp (β − α)(E/100)β
if E < (α − β)E0
if E ≥ (α − β)E0
(1)
This yields a peak energy in the range E0 ∼ 85 −140 keV during the burst (see Table 2).
The 15 -- 2000 keV peak flux of this burst was 3.3 × 10−5 erg cm−2 s−1 over 0.048 s, and the
15 -- 2000 keV fluence was 1.3×10−5 erg cm−2. Neither the time history, nor the spectrum, nor
the intensity of this event was exceptional in any way. Atteia (2003) has proposed a simple
-- 5 --
redshift estimator based on the photon flux, energy spectrum, and duration of a GRB. Using
this method, we obtain z ≈ 0.8, with a statistical uncertainty of about 50%.
2.2. Afterglow Searches
In the weeks immediately following GRB 001025A, optical, X-ray, and radio observations
were carried out in attempts to identify an afterglow.
An optical afterglow was not identified, either in observations starting 1.16 days after
the burst (Fynbo et al. 2000, R > 22.5), or in observations starting 1.66 days after the burst
(Uemura et al. 2000, R > 18.0).
An XMM-Newton observation ∼ 2 days after the burst revealed one apparently fading
source (S1, see Figure 5) in the initial ASM/IPN error box (Altieri et al. 2000a). In late-time
follow-up XMM-Newton observations this source had disappeared, strongly indicating that
(S1) is the X-ray afterglow of GRB 001025A.
Radio observations with the Very Large Array (VLA) were carried out at two epochs,
but they did not reveal any radio afterglow. The first VLA observation took place Novem-
ber 1, 2000, approximately 6 days after the GRB (D. Frail 2003, private communication).
Unfortunately, the XMM-Newton attitude was subsequently revised (Altieri et al. 2000b),
and the refined position of S1 was ∼ 2′ from the initial one. Due to the off-set between
the VLA pointing and the revised position of S1, the VLA sensitivity at the position of
S1 was relatively low. VLA observations took place again on November 21 and 24, 2000
(27 -- 30 days after the burst), but given the timescales and sensitivies (∼ 200µJy) most GRB
radio afterglows would not have been detected. Thus it is not clear whether this burst was
radio-quiet.
In the following three sections, we present our analysis and results of optical observations
and X-ray observations of the GRB 001025A afterglow.
2.3. Optical Observatons with the VLT
The field of GRB 001025A was observed at the Very Large Telescope (VLT) equipped
with the FORS1 camera at three epochs, under photometric and good seeing conditions (see
Table 3 for details). The first epoch was a few hours after the release of the IPN error box
(Hurley et al. 2000), the second epoch was 24 hours later, and the final epoch was 5 months
later.
-- 6 --
By comparing the combined R-band images of the three epochs we find no transient
sources in the IPN error box and in particular near the XMM-Newton error circle of S1 to a
3σ detection limit of R = 25.5. Figure 3 shows a 12×12 arcsec2 section of the last epoch VLT
image with the XMM-Newton 2σ error circle overplotted. The 2σ error circle is consistent
with the eastern part of what seems to be either a galaxy with two main components or a
random projection of two unrelated objects. Following Piro et al. (2002), we estimate that
the probability for a chance alignment is less than 1%, so we consider this object the likely
host galaxy of GRB 001025A. The total magnitude of this complex is R = 24.01±0.04 and
B = 25.13±0.12 (corrected for a modest foreground extinction of E(B − V ) = 0.07 ?).
GRB afterglows display a wide variety of fading behaviors (see Figure 4). Of the pre-
Swift detected optical afterglows, all but GRB 030115 would have been detected in the VLT
observations. The failure to detect the optical afterglow of GRB 001025A is thus not because
our observations are less sensitive than studies of other detected GRB afterglows.
2.4. X-ray Observations with XMM-Newton
XMM-Newton carried out observations of the region around the error box at three
different epochs, spanning 1.9 -- 2.3 days, 761.0 -- 761.2 days, and 910.2 -- 910.4 days after the
GRB (see Table 4). The XMM-Newton target of opportunity observation ∼ 2 days after the
burst revealed at least four X-ray sources in the initial ASM/IPN error box (Altieri et al.
2000a), one of which (S1) appeared to be fading. As shown in Figure 5, only S1 is in the final
ASM/IPN error box. This, along with marginal evidence for a fading behavior (see Section 4
below, and Watson et al. 2002), suggested that it was the X-ray afterglow of GRB 001025A.
In order to determine with certainty whether S1 was indeed the fading afterglow (as opposed
to a random, variable X-ray source such as an active galactic nucleus), we carried out a follow-
up XMM-Newton observation. This observation started on November 25, 2002, but had to
be terminated because of high background radiation; it was completed on April 23, 2003.
The first observation has already been reported on in Watson et al. (2002), but for the
analysis presented here, we have made two improvements: we have omitted data taken during
periods with high background, and we have analyzed data from the second and third epochs
as well. The effective EPIC pn exposure times of the three XMM-Newton observations, after
screening out high background intervals, are 15 ks, 9 ks, and 18 ks respectively.
The data were processed and analysed with SAS version 5.4.1 and XSPEC version
11.3. Periods with flares were identified based on the 10 -- 12 keV lightcurve of the full EPIC
detector and events registered during these periods (where the pn count rate was above 1.5
-- 7 --
counts/s, and where the MOS count rate was above 0.25 counts/s) were filtered out. For the
pn detector, single and double events were included in the analysis and the two exposures
taken in the first epoch were co-added. For the MOS detectors, single, double, triple, and
quadruple events were included in the analysis and the two MOS data sets for each epoch
were co-added. pn events in the energy interval 0.3 -- 9 keV and MOS events in the energy
interval 0.4 -- 7 keV were included in the analysis. Spectra and lightcurves of sources were
extracted using a circular aperture of radius 18′′. Background events were extracted from an
annulus centered on the source in question with an inner radius of 18′′ and an outer radius
of 50′′. The background regions are entirely on the same chip as the relevant sources, and
they do not include other contaminating sources.
Below we present results from our analysis of the XMM-Newton images, lightcurves,
and spectra of the GRB 001025A X-ray afterglow.
2.4.1.
Imaging
In the first epoch observation two sources were detected, S1 R.A.(2000)= 8h36m35.s93,
decl.(2000)= −13◦04′10.′′82, ±0.′′3), and S2, in and in the vicinity of the final ASM/IPN error
box, respectively (see Figure 5). Over the first XMM-Newton observation 264 net counts
(0.3 -- 12 keV) were detected for source S1 by the pn detector, corresponding to a flux of
(6.4 ± 0.2) × 10−14 erg cm−2 s−1. For source S2, 78 net counts (0.3 -- 12 keV) were detected,
corresponding to a flux of (2.0 ± 0.4) × 10−14 erg cm−2 s−1. By the time of the second epoch
observation, S1 had faded beyond detection (fX(0.3 -- 12 keV)< 6 × 10−15 erg cm−2 s−1) while
the flux of S2 was consistent with that of the first epoch. In the third epoch observation
S1 was still undetectable (fX (0.3 -- 12 keV)< 5 × 10−15 erg cm−2 s−1) and the flux of S2 was
consistent with the flux at the first two epochs. Hence, source S2 is persistent over several
years, displaying flux variations less than about 25% from epoch to epoch. S2 is coincident
with an R ∼ 20 optical source which based on a spectrum obtained with the Nordic Optical
Telescope seems to be a QSO at redshift z ≈ 2.5 (M. I. Andersen, private communication).
2.4.2. Lightcurve
The pn lightcurve for S1 is well fitted by a power law with decay index δX = −1.6 ± 0.35
(χ2 = 4.80 for 8 d.o.f., 0.3-12 keV). This is somewhat lower than, but consistent with, the
steep decay index δX = −3.0 ± 1.9 derived by Watson et al. (2002). The data quality of the
corresponding co-added MOS1+2 lightcurve does not allow us to further constrain the fading
-- 8 --
of S1. The decay of S1 is typical of X-ray afterglows having a decay index of δX ≈ −1.4
(Piro 2001). By extrapolating the best-fit pn lightcurve we predict the total number of
counts in the second and third epoch XMM-Newton observations from S1 to be 0.04 counts
and 0.02 counts, respectively (i.e., well below detectability). Hence, the second and third
epoch observations establish that S1 is a non-persistent source; hence we identify S1 as the
X-ray afterglow of GRB 001025A.
In order to compare the X-ray afterglow to the prompt X-ray emission as observed by
the RXTE ASM we extracted the pn 1.5 -- 12 keV lightcurve. A power law fit to the 1.5 --
12 keV light curve is not well constrained (decay index ∼ −1). Hence, we extrapolated
the best fit power law to the 0.3 -- 12 keV light curve (renormalized to the 1.5 -- 12 keV flux
level) to the prompt emission phase. This extrapolation reproduces the RXTE ASM flux
within the uncertainties (see Figure 6). However, fitting a power law to the fading of the
prompt emission in the 1.5 -- 12 keV band yields a steep power law (decay index −3.1 ± 0.4)
underestimating the pn data points. Such a steep decay has been reported in several early
X-ray afterglows observed with the Swift X-ray Telescope (Tagliaferri et al. 2003).
It is
not possible to fit a power law decay jointly to the fading prompt emission and the X-
ray afterglow. The X-ray afterglow observed by XMM-Newton is therefore not a simple
continuation of the prompt emission.
2.4.3. Spectral Analysis
Spectra of S1 were extracted from the first epoch observations and binned with a mini-
mum of 20 counts per bin. Several spectral models were fitted to the pn and MOS1+2 spectra
simultaneously: (i) power law models, representing synchrotron emission from a population
of relativistic electrons in the fireball (Sari, Piran & Narayan 1998), (ii) thermal plasma
emission giving rise to X-ray emission lines as seen in several other GRB afterglows (Reeves
et al. 2002; Watson et al. 2003) (the MEKAL model as is typically used to represent thermal
emission Mewe et al. 1985; Liedahl, Osterheld & Goldstein 1995), and (iii) combinations of
(i) and (ii). In all models Galactic absorption was included, and additional absorption at the
burst redshift was included in some of the models. A summary of the spectral fits is given
in Tables 5,6,7 with errors quoted as 90% confidence intervals.
An absorbed synchrotron model is an acceptable fit to the spectrum (see Table 5), but
the best-fit photon index is rather steep (Γ ∼ 2.8, consistent with the findings of Watson et al.
2002). The Galactic H I column density in the direction of GRB 001025A is 6.1 × 1020 cm−2
and can in worst case be uncertain by up to 50% (Dickey & Lockman 1990). Power law
models indicate absorption in excess of Galactic absorption, but the column density of extra-
-- 9 --
galactic absorbing material is not strongly constrained. In this model the burst redshift is
unconstrained.
An absorbed thermal plasma model (MEKAL) is as good a fit as the absorbed syn-
chrotron emission model (see Table 6) and implies a plasma temperature around 3 keV and
a burst redshift around z ∼ 0.3. Watson et al. (2002) favor a redshift in the range 0.5 -- 1.2
from their MEKAL model fits, and we note that the thermal plasma could be outflowing
resulting in a host galaxy redshift lower by ∆z ∼ 0.1 (e.g. Reeves et al. 2002). Hence, ther-
mal emission models do not provide strong constraints on the burst redshift. The absorbing
column density at the burst redshift is at most twice the Galactic column density. The
plasma abundances are not well constrained, but they are consistent with the solar values
(ZF e . 2Z⊙, ZS,Si . 13Z⊙).
Naturally, a model with a combination of a synchrotron emission component and a
thermal emission component (see Figure 7) is also a good fit (see Table 7). The relative
normalization of the synchrotron emission component and the thermal emission component
is not well constrained, but the synchrotron emission component contributes ∼ 2/3 of the
total flux and above ∼ 2 keV the synchrotron emission component dominates. The best-fit
photon index is lower than in the pure synchrotron emission model fits (Γ ∼ 1.8), but close
to typical photon indices for X-ray afterglows (Piro 2001). The favored burst redshift is
z ≈ 2, and there is an indication of circumburst absorbtion (NH ∼ 10 − 20 × 1021 cm−2).
The abundances of the thermal plasma are not well constrained, but they are consistent with
the solar values (ZF e,S,Si . 6Z⊙).
In conclusion, an absorbed synchrotron model, an absorbed thermal plasma model, and
a combination of the two all represent a good fit to the data. Consequently, we consider all
these models in the following.
2.5. Spectral Energy Distribution
We derived the broadband spectral energy distribution by extrapolating the best-fit X-
ray spectral models to the optical band. We used the X-ray decay index δX = −1.6 to recast
the X-ray spectrum to the epoch of the first R-band observations (1.20 days after the burst).
For the synchrotron emission model (spectral index βX = 1.8) an observed R-band extinction
of at least 7 magnitudes is required to bring the model in agreement with the R-band upper
limit (see Figure 8). On the other hand, extrapolating the synchrotron emission component
from the best-fit combined synchrotron emission/thermal emission model (spectral index
βX = 0.8) to the R-band shows that the R-band upper limit is consistent with the model, if
-- 10 --
the cooling break is close to the X-ray band.
3. DISCUSSION
The spectral index of the (X-ray) afterglow is a key parameter for constraining the
properties of the burst and its environment. The pure synchrotron emission fit to the
GRB 001025A X-ray afterglow has a very steep spectral index for a GRB X-ray afterglow.
On the other hand, the synchrotron component of the combined thermal plasma/synchrotron
model has a spectral index in line with what is found for other GRBs (Piro 2001). Detec-
tion of thermal afterglow emission has been reported in several other GRBs (see Watson et
al. 2003, for an account of GRBs with thermal emission) increasing the soft X-ray flux in
the ∼ 0.5 − 2 keV range. One of the most prominent claims of thermal emission is in the
GRB 011211 X-ray afterglow where thermal emission is detected during the first 5 ks of the
XMM-Newton observation (Reeves et al. 2002). We investigated the effect on the spectral in-
dex of fitting an absorbed synchrotron model to a spectrum with thermal emission by fitting
an absorbed synchrotron emission model to the first 5 ks of the GRB 011211 XMM-Newton
spectrum. We obtain a spectral index of 1.6 ± 0.3; this is very similar to the spectral index
(βX = 1.8 ± 0.3) we find fitting the same model to GRB 001025A. The steep spectral index
of a pure synchrotron emission fit to GRB 001025A compared to the shallower index from a
synchrotron emission plus thermal emission fit thus suggests that thermal emission may be
steepening the X-ray afterglow spectrum of GRB 001025A.
Due to the different best fit spectral indices of the synchrotron emission in the pure
synchroton model and in the combined thermal plasma/synchrotron model, these two models
have quite different implications for the deduced properties of the GRB 001025A fireball and
its environment.
3.1. Physics of the Fireball
The prompt soft X-ray emission has a steep decay as seen in the tails of other GRBs
(Smith et al. 2002; Giblin et al. 2002). Observations with the Swift XRT has revealed a
population of bursts with an early steep decay and a shallower late time decay (Tagliaferri
et al. 2003) similar to the behavior of GRB 001025A.
Several models predicts an early steep X-ray decay, including a hot cocoon surrounding
a relativistic jet (Ramirez-Ruiz et al. 2002), emission from outside the relativistic beaming
cone, θ > Γ−1, (Kumar & Panaitescu 2000), and a single spherical shell emitting instanta-
-- 11 --
neously (Fenimore et al. 1996). The steepness of the prompt emission decay excludes that
the tail of the prompt emission is the onset of the afterglow (Sari, Piran & Narayan 1998).
However, the afterglow could appear at any time > 16 s after the burst at which epoch the
GRB 001025A afterglow is expected to be below the sensitivity of RXTE ASM.
The X-ray afterglow has a steep spectral index, if it is due to synchrotron emission
only. The standard fireball model predicts in this case a temporal decay index of −2.2 ±
0.45 to −3.2 ± 0.45 for a spherical blast wave and an index of −3.6 ± 0.6 to −4.6 ± 0.6 for
a jet (Zhang & Meszaros 2004). This is steeper than observed (δX = −1.6 ± 0.35), but only
the jet model can be ruled out. If the X-ray afterglow is due to a combination of thermal
plasma and synchrotron emission, the fireball model predicts a shallower afterglow decay
index of the synchrotron component (index −0.4 ± 0.3 to −1.7 ± 0.9 for a spherical blast
wave, index −1.9 ± 0.3 to −2.6 ± 1.2 for a jet, Zhang & Meszaros 2004) more in line with the
X-ray lightcurve. In this case models where the X-ray afterglow originates from fast cooling
are ruled out.
3.2. Burst Redshift and Luminosity
In the combined synchroton/thermal plasma spectral model of the X-ray afterglow,
the suggested burst redshift is z ≈ 2. This yields an isotropic prompt energy release of
1.5 × 1053 erg (for a H0 = 72 km/s/Mpc, Ωm = 0.3, ΩΛ = 0.7 cosmology) which is quite
typical (e.g. Bloom, Frail & Kulkarni 2003). The isotropic 0.3 -- 12 keV afterglow luminosity
is LX = (3.2 ± 2.6) × 1045 erg/s time-averaged over the first XMM-Newton observation. We
now compare the X-ray luminosity of GRB 001025A to the luminosities of the GRB sample
of Berger et al. (2003). We use our derived X-ray afterglow decay index, flux and spectral
index to calculate the isotropic luminosity (LX,iso, eq. (1) in Berger et al. 2003, using their
cosmology) of GRB 001025A 10 hours after the burst. In the combined synchrotron/thermal
plasma model with a burst redshift z = 2, we find LX,iso = (2.4 ± 1.9) × 1046 erg/s. This
is close to the typical X-ray afterglow luminosity at this epoch for the Berger et al. (2003)
burst sample. However, in that sample, redshifts are not known for most of the bursts, in
which case the luminosity has been derived assuming a redshift of z = 1.1. If GRB 001025A
is at a redshift of z = 1.1 its isotropic luminosity is LX,iso = (3.5 ± 2.5) × 1045 erg/s which is
in the low tail of the Berger et al. (2003) X-ray luminosity distribution. If the burst redshift
is z = 0.33, as predicted in the thermal plasma only model, the luminosity of GRB 001025A
LX,iso = (2.3 ± 1.5) × 1044 erg/s is lower than all GRBs in the Berger et al. (2003) sample.
The B-band detection of the likely host galaxy implies a redshift z < 2.7. The magnitude
and color of the galaxy are consistent with a redshift around two, but they do not constitute
-- 12 --
strong constraints on the redshift.
The pseudo-redshift estimate (Atteia 2003) for this burst z = 0.8 ± 0.4, and although
not strongly constraining, this is in line with the lower redshift derived from the thermal
emission models of the X-ray afterglow.
3.3. The Burst Environment
The appearance of the candidate host galaxy is similar to other GRB hosts, i.e., an
irregular morphology and a color that for a redshift of z > 0.3 is indicative of active star
formation. From the absorbing column density in the soft X-ray band, the extinction towards
the burst can be estimated by assuming a gas-to-dust ratio. The best-fit column densities
at the burst redshift are NH . 20 × 1021 cm−2. Taking a Galactic gas-to-dust ratio (Predehl
& Schmitt 1995), this translates into a restframe extinction of AV . 11 for z < 2. Using
an SMC gas-to-dust ratio results in AV . 1, i.e. not enough to account for the extinction
required in the pure synchrotron model for the X-ray afterglow. A Galactic gas-to-dust ratio
is at variance with the findings for GRBs with detected optical afterglows generally having
SMC-like gas-to-dust ratios (Galama & Wijers 2001). Either the pure synchrotron model
is not appropriate for the GRB 001025A X-ray afterglow or GRB 001025A is much more
obscured than bursts with detected optical afterglows (typically having AV . 0.2 Hjorth et
al. 2003a).
3.4. What Makes GRB 001025A a Dark Burst?
The dark bursts in the GRB sample of Jakobsson et al. (2004) (updated as of July 8,
2005, Jakobsson 2005) has a mean 1.6 -- 10 keV flux of 3.0 × 10−12 erg cm−2 s−1 11 hours
after the burst. Extrapolating the best-fit X-ray lightcurve of GRB 001025A to this epoch,
the estimated 1.6 -- 10 keV flux is 5.0 ± 2.7 × 10−13 erg cm−2 s−1. The X-ray afterglow of
GRB 001025A is fainter than any of the dark bursts in the Jakobsson et al. (2004) sample
and much fainter than the mean of GRBs with detected optical afterglows in this sample
(4.2 × 10−12 erg cm−2 s−1).
The low R-band flux of GRB 001025A is not caused solely by a general faintness of the
afterglow at all wavelengths. The R-band-to-3 keV spectral index is . 0.43 so GRB 001025A
is clearly dark according to the definition of Jakobsson et al. (2004). For GRBs with detected
optical afterglows, the R-band-to-3 keV spectral index is 0.73 and for Swift detected bursts
it is 0.65 (Jakobsson 2005) so the afterglow of GRB 001025A is optically faint relative to
-- 13 --
the X-ray band. The low optical-to-X-ray spectral index could be due to heavy obscuration
in the burst environment. Alternatively, the cooling frequency could be close to the X-ray
band, giving rise to a shallow spectral index requiring no optical obscuration.
The X-ray spectral index has been measured only for a few dark GRBs. By re-
normalizing the X-ray spectrum and the optical observations to a common epoch (e.g. 11 h
after the burst) and extrapolating the X-ray spectral index to the optical band, assuming
a position of the cooling break, the R-band flux can be predicted. For three of the five
GRBs with measured spectral index in the dark GRB sample of De Pasquale et al. (2003)
(GRB 990704, GRB 990806, GRB 000214) the predicted R-band flux is lower than the ob-
servational upper limit, when invoking a cooling break at 1 keV. These three GRBs have
steep X-ray spectral indices, possibly reflecting the presence of a thermal component on top
of the synchrotron emission. The true synchrotron spectral index may thus be lower than
the index obtained from a pure synchrotron emission fit, giving rise to even lower predictions
for the optical flux. For the remaining two GRBs in the dark GRB sample of De Pasquale
et al. (2003) (GRB 000210 and GRB 001109) obscuration is required in order to bring the
observational upper limit in line with the model prediction (or the derived X-ray spectral
index is too step due to the presence of thermal emission). Two of the bursts detected by
Swift and with published X-ray spectral index (GRB 050401 and GRB 050408, Chincarini
et al. 2005) are dark according to the definition of Jakobsson et al. (2004). For both bursts
obscuration is required in order to be consistent with the upper limit on the R-band flux,
even when invoking a cooling break at 1 keV.
4. CONCLUSIONS
GRB 001025A was a long-duration GRB with a prompt emission spectrum well fit-
ted by the Band model. The tail of the prompt soft X-ray emission decays too steeply to
be the beginning of the afterglow. Three epoch XMM-Newton observations show that the
GRB 001025A afterglow is X-ray faint. Furthermore, the optical flux is relatively low com-
pared to the X-ray band. Either (i) the optical afterglow suffers extinction of at least seven
magnitudes, or (ii) a significant fraction of the X-ray afterglow flux originates from a thermal
plasma, and the cooling frequency is close to the X-ray band at 1.2 days after the burst. In
the latter case, we predict that the burst redshift is around two, and we find that the X-ray
luminosity and the spectral index are fairly typical of GRB afterglows. Alternatively, the
GRB 001025A X-ray afterglow spectrum is characterized by a very steep synchrotron index,
and GRB 001025A is situated in an environment with an unusually large gas-to-dust ratio,
consistent with the Galactic ratio.
-- 14 --
The present study of GRB 001025A, and the properties of the few other dark GRBs
with measured spectral index, raises the possibility that some bursts appear optically dark
because their afterglow is faint, and their cooling frequency is close to the X-ray band. The
viability of this scenario can be tested from early observations of X-ray afterglows from dark
GRBs yielding the temporal and spectral behaviour of the X-ray afterglows, which in turn
constrain the geometry of the blast wave and the underlying physical processes. This would
provide a crucial test for physical differences between dark bursts and bursts with an optical
afterglow.
We are grateful to Dale Frail for making the results of his VLA observations known to
us. We thank the anonymous referee for useful and very detailed comments that improved
the presentation of our results. KH is grateful for Ulysses support under JPL Contract
958056, for support from the Long Term Space Astrophysics program under NAG5-3500, and
for support as a Participating Scientist in the NEAR mission under NAG5-9503. KP and
JPUF acknowledge support from the Carlsberg foundation. KP acknowledges support from
Instrument Center for Danish Astrophysics. PJ acknowledges support from a special grant
from the Icelandic Research Council. This work was supported by the Danish Natural Science
Research Council (SNF). From the Russian side this work was suported by the Russian Space
Agency contract and RBRF grant 03-02-17517. DAS is supported by an NSF Astronomy and
Astrophysics Postdoctoral Fellowship under award AST-0105221. The authors acknowledge
benefits from collaboration within the EU FP5 Research Training Network "Gamma-Ray
Bursts: An Enigma and a Tool". Based on observations obtained with XMM-Newton, an
ESA science mission with instruments and contributions directly funded by ESA Member
States and NASA,and based on observations made with ESO Telescopes at the La Silla or
Paranal Observatories under programme 66.A.-0386(A).
Facilities: XMM(EPIC), VLT(ANTU), RXTE(ASM), NEAR(XGRS), Wind(Konus).
REFERENCES
Altieri, B., Schartel, N., Santos, M., Tomas, L., Guainazzi, M., Piro, L., & Parmar, A. 2000a,
GCN Circ. 869 (http://gcn.gsfc.nasa.gov/gcn/gcn3/869.gcn3)
Altieri, B., Schartel, N., Lumb, D., Piro, L., & Parmar, A. 2000b, GCN Circ. 884
(http://gcn.gsfc.nasa.gov/gcn/gcn3/884.gcn3)
Atteia, J.-L. 2003, A&A 407, L1
Band, D., et al. 1993, ApJ 413, 281
-- 15 --
Berger, E., et al. 2002, ApJ 581, 981
Berger, E., Kulkarni, S.R., Frail, D.A., 2003, ApJ 590, 379
Berger, E., et al. 2005, ApJ submitted, astro-ph/0505107
Bloom, J.S., Frail, D.A. & Kulkarni, S.R. 2003, ApJ 594, 674
Bloom, J.S., et al. 2005, ApJ submitted (astro-ph/0505480)
Castro Cer´on, J.M., et al. 2004, A&A 424, 833
Costa, E., 1999, A&AS 138, 425
Chincarini, G. et al. 2005, astro-ph/0506453
Crew, G. et al. 2003, ApJ 599, 387
De Pasquale, M. et al., 2003, ApJ 592, 1018
Dickey, J.M. & Lockman, F.J., 1990, ARA&A, 28, 215
Djorgovski, S.G., et al. 2001, ApJ 562, 654
Fenimore, E.E., et al. 1996, ApJ 473, 998
Fynbo, J. P. U., et al. 2000, GCN Circ. 867 (http://gcn.gsfc.nasa.gov/gcn/gcn3/867.gcn3)
Fynbo, J. P. U., et al. 2001, A&A 369, 373
Galama, T.J. & Wijers, R.A.M.J. 2001, ApJ 549, L209
Gehrels, N., et al. 2005, Nature submitted (astro-ph/0505630)
Giblin, T.W., et al. 2002, ApJ 570, 573 A&A 383, 112
Gorosabel, J., et al. 2002, A&A 383, 112
Groot, P., et al. 1998, ApJ 493, L27
Hjorth, J., et al. 2002, ApJ 576, 113
Hjorth, J., et al. 2003a, ApJ 597, 699
Hjorth, J., et al. 2003b, Nature 423, 847
Hjorth, J., et al. 2005, ApJ submitted (astro-ph/0506123)
-- 16 --
Hurley,
K.,
Cline,
T.,
&
Smith,
D.
2000,
GCN
Circ.
863
(http://gcn.gsfc.nasa.gov/gcn/gcn3/863.gcn3)
Hurley, K. 2000, GCN Circ. 864, (http://gcn.gsfc.nasa.gov/gcn/gcn3/864.gcn3)
Hurley, K. et al. 2002, ApJ 567, 447
Jakobsson, P. et al. 2004, ApJ 617, L21
Jakobsson, P. 2005, http://www.astro.ku.dk/pallja/dark.html
Kumar, P., & Panaitescu, A. 2000, ApJ 541, L51
Lamb, D.Q., & Reichart, D. 2000, ApJ 536, 1
Lazatti, D., Covino, S., & Ghisellini, G., 2002, MNRAS 330, 583
Liedahl, D.A., Osterheld, A.L., & Goldstein, W.H., 1995, ApJ 438, L115
Malesani, D., et al. 2004, ApJ 609, L5
Meszaros, P. 2002, ARAA 40, 137
Mewe, R., Gronenschield, E.H.B.M., & van den Oord, G.H.J., 1985, A& AS 62, 197
Piro, L., 2001, in Gamma-ray bursts in the afterglow era, eds. E Costa, F. Frontera, & J.
Hjorth (Springer - Berlin), p. 97
Piro, L., et al., 2002, ApJ 577, 680
Predehl, P. & Schmitt, J.H.M.M., 1995, A&A 293, 889
Preece, R., Pendleton, G., Briggs, M., Mallozzi, R., Paciesas, W., Band, D., Matteson, J.,
& Meegan, C. 1998, ApJ 496, 849
Ramirez-Ruiz, E., et al., 2002, MNRAS 337, 1349
Reichart, D., & Price, P. 2002, ApJ 565, 174
Rol, E. et al. 2005, accepted for publication in ApJ
Reeves, J. et al. 2002, Nature 416, 512
Sari R., Piran, T. & Narayan, R., 1998, ApJ 497, L17
Schlegel, D.J. et al. 1998, ApJ 500, 525
-- 17 --
Smith, D.A., Levine, A., Remillard, R., Hurley, K., & Cline, T. 2000, GCN Circ. 861
(http://gcn.gsfc.nasa.gov/gcn/gcn3/861.gcn3)
Smith D.A., et al., 2002, ApJS 141, 415
Stanek, K.Z., et al., 2003, ApJ 591, L17
Tagliaferri, G., et al., Nature accepted, astro-ph/0506355
Taylor, G.B., et al., 1998, ApJ 502, L115
Uemura, M., Kato, T., Ishioka, R., Iwamatsu, H., & Yamaoka, H. 2000 GCN Circ. 866
(http://gcn.gsfc.nasa.gov/gcn/gcn3/866.gcn3)
Watson, D., Reeves, J., Osborne, J., O'Brien, P., Pounds, K., Tedds, J., Santos-Lleo, M., &
Ehle, M. 2002 A&A 393, L1
Watson, D., Reeves, J., Hjorth, J., & Pedersen, K. 2003 ApJ 595, L29
Zhang, B. & Meszaros, P. 2004 Int. J. Mod. Phys. A 19, 2385
This preprint was prepared with the AAS LATEX macros v5.2.
-- 18 --
Fig. 1. -- Lightcurves of GRB 001025A. Plotted from top to bottom are the background
subtracted lightcurves of RXTE ASM (1.5 -- 12 keV, 1 s resolution), NEAR XGRS (100 keV --
2 MeV, 1 s resolution), Wind-Konus in three different energy bands (18 -- 70 keV, 70 -- 305 keV,
305 -- 1150 keV, 0.128 s resolution), and the hardness ratio between the Wind-Konus 70 --
305 keV and 18 -- 70 keV energy bands as function of time. The vertical dotted lines indicate
the intervals where time-resolved spectroscopy was done, see Fig 2.
-- 19 --
0 - 0.256s
0.256 - 0.768s
0.768 - 8.960s
0 - 0.768s
104
103
102
101
104
103
102
1
-
s
2
-
m
c
V
e
k
,
Fn
1
-
s
2
-
m
c
V
e
k
,
Fn
101
101
102
103
101
102
103
E, keV
E, keV
Fig. 2. -- Time-resolved energy spectra of GRB 001025A in νFν units from Wind-Konus.
Spectral fitting was done using the Band model. The fitting parameters are given in Table 2.
The time intervals used for the spectral fits are indicated in Figure 1.
n
n
-- 20 --
Fig. 3. -- 12 × 12 arcsec2 VLT R-band image with the 2σ error circle of source S1. North is
up and East is to the left.
-- 21 --
18
4
7
)
g
a
m
(
R
20
22
24
26
6
5
2
3
1
001025A
0.1
tobs - tburst (days)
1.0
Fig. 4. -- R-band magnitudes and upper limits as a function of time for various gamma-ray
bursts in the pre-Swift era, compared to the observations of GRB 001025A (filled circle).
The dotted lines sketch the decay of optically dim bursts, starting at the time of the first
observation. The labels are as follows: (1) GRB 980329, (2) GRB 980613, (3) GRB 000630,
(4) GRB 020124, (5) GRB 020322, (6) GRB 030115, (7) GRB 021211.
-- 22 --
S2
S1
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
n
o
i
t
a
n
i
l
c
e
D
-13:09:00
45
8:36:40
35
30
25
20
15
Right ascension
S2
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
n
o
i
t
a
n
i
l
c
e
D
-13:09:00
45
8:36:40
35
30
25
20
15
Right ascension
S2
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
n
o
i
t
a
n
i
l
c
e
D
-13:09:00
45
8:36:40
35
30
25
20
15
Right ascension
Fig. 5. -- XMM-Newton co-added pn+MOS1+MOS2 images showing the afterglow, S1, and
the source S2, with the final RXTE-ASM/IPN error box overlaid, at three epochs: Top:
1.88 days after the burst. Middle: 761 days after the burst. Bottom: 910 days after the
burst. See Table 4 for the exposure time of invididual images.
-- 23 --
Fig. 6. -- RXTE ASM and XMM-Newton pn 1.5 -- 12 keV lightcurves with power law fits
overlaid. Dotted line: Fit to the tail of the RXTE ASM lightcurve. Full line: Fit to XMM-
Newton pn 0.3 -- 12 keV lightcurve renormalized to the 1.5 -- 12 keV flux level. Dashed lines:
1σ envelope of fit to XMM-Newton pn 0.3 -- 12 keV lightcurve renormalized to the 1.5 -- 12 keV
flux level.
-- 24 --
Fig. 7. -- Top panel: XMM-Newton pn (squares) and MOS1+2 (stars) spectra of the after-
glow of GRB 001025A with a spectral binning of min. 10 counts per bin. Overplotted is the
best-fit combined synchrotron/thermal emission model with fixed Galactic absorption and
absorption at the burst redshift. Lower panel: residuals of model fit.
-- 25 --
Fig. 8. -- The optical to X-ray spectral energy distribution of GRB 001025A 1.20 days after
the burst. The XMM-Newton pn spectrum recast to this epoch is shown as diamonds (only
data points above 1.3 keV, where absorption is neglibible), the R-band upper limit is shown
as the filled circle. Overlaid is the synchrotron emission component of the best-fit from the
combined synchrotron emission/thermal emission model (spectral index 0.8, full line) and
the best-fit synchrotron emission model (spectral index 1.8, dotted line). The extrapolation
to the optical band assumes that the cooling break is at 1 keV (2140 × 1014 Hz) and that
the spectral index changes by −0.5 (Sari, Piran & Narayan 1998) when going to the short
frequency side of the break.
-- 26 --
Table 1. Corners of the final ASM-IPN Error Box for GRB 001025A
R.A.(2000)
decl.(2000)
129.◦1456
129.◦1619
129.◦0891
129.◦1054
−13.◦0579
−13.◦0663
−13.◦0854
−13.◦0938
-- 27 --
Table 2. Spectral Fits to the Prompt Emission of GRB 001025Aa
Interval
(s)
Normalization
(photons cm−2 s−1)
α
β
E0
(keV)
0 -- 0.256
0.256 -- 0.768
0.768 -- 8.960
0 -- 0.768
0 -- 8.960
0.26 ±0.1
0.37 ±0.04
0.03 ±0.1
0.29 ±0.04
0.036 ±0.004
+0.81 ± 0.57
-0.39 ± 0.11
-0.99 ± 0.3
-0.30 ± 0.12
-0.91 ± 0.1
-1.84 ± 0.12
-2.75 ± 0.2
-2.40 ± 0.2
-2.55 ± 0.16
-2.40 ± 0.16
85 ± 33
140 ± 19
103 ± 46
148 ± 20
191 ± 37
aNote: See Eq. 1 for the definition of α, β, and E0.
-- 28 --
Table 3. Journal of VLT Observations
UT
∆t
Filter
Exp. time
(days)
(s)
Seeing
(arcsec)
2000 Oct 26.34
2000 Oct 27.32
2001 Mar 23.02
2001 Mar 23.04
1.21
2.19
146.89
146.91
R
R
B
R
12×200
8×200
3×200
10×200
0.50 -- 0.75
0.50 -- 0.75
0.45 -- 0.55
0.45 -- 0.55
-- 29 --
Table 4. XMM-Newton Observations of GRB 001025A
Datea
Time since GRB
Instrument
Exposure timeb
(days)
1.88
761
910
2000 Oct 27.003c
2002 Nov 25.16d
2003 Apr 23.29e
aStart of observation
(ks)
15
32
8.8
8.6
18
19
pn
MOS
pn
MOS
pn
MOS
bExposure time for each detector used in the analysis, i.e. with high
background periods left out
cTarget of opportunity observation (Altieri et al. 2000a,b)
dAO-2 observation
eContinuation of AO-2 observation
-- 30 --
Table 5. Synchrotron Model Fits to the X-ray Afterglow Spectrum of GRB 001025A
Model
Γa
Absorptionb
(1021 cm−2)
Redshiftc
χ2/d.o.f.
Fixed index, fixed Galactic abs.
Free Galactic abs.
Fixed Galactic abs, free abs. at z
2.0
2.8 ± 0.3
2.6 ± 0.3
0.61
2.6+0.4
15+26
−0.6
−5
0
0
< 7.5
66.8/33
32.6/32
31.3/31
aPhoton index.
bColumn density of absorber at the quoted redshift.
cRedshift of absorber.
-- 31 --
Table 6. Thermal Emission Model Fits to the X-ray Afterglow Spectrum of GRB 001025A
Model
T
(keV)
Absorptiona
(1021 cm−2)
Redshiftb
χ2/d.o.f.
Fixed Gal. abs.
Free Gal. abs.
Fixed Gal. abs, free abs. at z
3.0+0.5
−0.4
2.8 ± 0.4
2.8 ± 0.5
aAbsorber column density.
0.61
0.9+0.4
0.5+0.7
−0.3
−0.5
0.32 ± 0.05
0.33 ± 0.05
0.33 ± 0.05
36.7/32
34.1/31
33.7/31
bCommon redshift of thermal emission and extra-galactic absorber.
-- 32 --
Table 7. Combined Synchrotron Model and Thermal Emission Model Fits to the X-ray
Afterglow Spectrum of GRB 001025A
Model
Γa
T
(keV)
Absorptionb
(1021 cm−2)
Redshiftc
χ2/d.o.f.
Fixed index, free Gal. abs.
Fixed Gal. abs.
Free Gal. abs.
Fixed Gal. abs, free abs. at z
Fixed index, free abs. at z
2.0
−1.5
1.2+1.6
1.9+0.6
−0.4
1.8 ± 0.6
2.0
−0.7
1.5 ± 0.6
2.8+0.5
1.6+0.5
−0.4
1.8 ± 0.6
1.7 ± 0.5
aPhoton index.
bAbsorber column density.
−1.0
3.0+1.4
0.61
2.8+1.0
17+9
−6
17 ± 6
−0.7
1.98+0.13
−0.15
0.32 ± 0.05
2.00+0.12
−0.15
2.00 ± 0.13
2.00+0.12
−0.14
17.9/27
34.1/29
18.4/28
19.6/28
19.8/29
cCommon redshift of thermal emission and extra-galactic absorber.
|
0805.1597 | 2 | 0805 | 2008-06-04T14:02:08 | X-ray absorption in Compton-thin AGN: the predictions of a model revisited | [
"astro-ph"
] | The evidence of a decrease with increasing luminosity of the fraction f_{abs} of absorbed and Compton-thin among X-ray (2-10 keV) selected AGN is observationally rather well supported, while that of an increase of f_{abs} with redshift is rather controversial. In Lamastra, Perola & Matt (2006) the gravitational effect of the SMBH on the molecular interstellar gas, in the central region of the host galaxy, was shown to predict an anti-correlation between f_{abs} and M_{BH}. The most recent findings on the distribution of the Eddington ratio \lambda=L_b/L_E as a function of M_{BH} and z are used to convert that relationship into one between f_{abs} and both bolometric (L_b) and X-ray (L_X) luminosities at various values of z. The findings on \lambda(M_{BH},z) are properly treated in order to ensure completeness in the prediction of f_{abs} above a certain luminosity, at values of z=0.1, 0.35, 0.7 and >1. To verify the consequence of these findings alone, we adopted in a first istance a distribution of gas surface density \Sigma, observed in a sample of local spiral galaxies, irrespective of the galaxy morphological type and z. Assuming in the \lambda(M_{BH},z) distribution the Eddington limit, \lambda=1, as a ``natural'' cut-off, the predictions are consistent with the existence of an anti-correlation between f_{abs} and L_X, but fail to reproduce an increase of f_{abs} with z. Because the early type galaxies on average are much poorer in molecular gas than late type ones, a quantitative agreement with the local value of f_{abs} requires the existence of a correlation between \Sigma and the central activity. An increase of typical values of \Sigma with z, correlated with the activity, might explain an increase of f_{abs} with z. However, at the highest luminosities f_{abs} could hardly exceed about 0.3. | astro-ph | astro-ph | Astronomy&Astrophysicsmanuscript no. paper3030608
November 5, 2018
c(cid:13) ESO 2018
8
0
0
2
n
u
J
4
]
h
p
-
o
r
t
s
a
[
2
v
7
9
5
1
.
5
0
8
0
:
v
i
X
r
a
X-ray absorption in Compton -- thin AGN: the predictions of a
model revisited
Alessandra Lamastra, G. Cesare Perola and Giorgio Matt
Dipartimento di Fisica "E. Amaldi", Universit`a degli Studi Roma Tre, via della Vasca Navale 84, I-00146 Roma, Italy
Received ; Accepted
ABSTRACT
Context. The evidence of a decrease with increasing luminosity in the fraction fabs of absorbed and Compton-thin among X-ray-
selected (2-10 keV) AGN is observationally well supported, while that of an increase in fabs with redshift is fairly controversial. In
Lamastra, Perola & Matt (2006), the gravitational effect of the SMBH on the molecular interstellar gas, in the central region of the
host galaxy, was shown to predict an anti-correlation between fabs and the black-hole mass MBH.
Aims. The most recent findings on the distribution of the Eddington ratio λ=Lb/LE as a function of MBH and z are used to convert that
relationship into one between fabs and both bolometric (Lb) and X-ray (LX) luminosities at various values of z.
Methods. The findings for λ(MBH, z) are properly treated to ensure completeness in the prediction of fabs above a certain luminosity,
at values of z=0.1, 0.35, 0.7, and ≥1. To verify the consequence of these findings alone, we first adopted a distribution of gas surface
density Σ, observed in a sample of local spiral galaxies, irrespective of the galaxy morphological type and z.
Results. Assuming the Eddington limit, λ=1, in the λ(MBH, z) distribution as a "natural" cut-off, the predictions are consistent with
the existence of an anti-correlation between fabs and LX, but they fail to reproduce an increase in fabs with z. Because the early type
galaxies are on average much poorer in molecular gas than late type ones, a quantitative agreement with the local value of fabs requires
the existence of a correlation between Σ and the central activity. An increase in typical values of Σ with z, correlated with the activity,
might explain an increase in fabs with z. However, fabs could hardly exceed about 0.3 at the highest luminosities.
Key words. Galaxies: active -- X-rays: galaxies -- ISM: clouds
1. Introduction
The selection of active galactic nuclei in the hard X-ray band (up
to about 10 keV with the XMM-Newton and Chandra satellites),
together with their redshift z measurement after optical identifi-
cation, immediately yields the spectroscopic estimate of the pho-
toelectrically absorbing column density, NH, intrinsic to each of
them. A schematic split between objects with NH either smaller
or larger than 1022 Hcm−2 corresponds fairly well (although not
exactly one-to-one) with the optical classification of AGN as ei-
ther type 1 (broad-line) or type 2 (narrow-line) (see, e. g., Perola
et al. 2004). Among the objects with NH larger than 1022 Hcm−2,
a further subdivision is necessary, those with NH smaller than
σ−1
= 1.5 × 1024cm−2, called Compton-thin, and those with NH
T
> σ−1
T , called Compton-thick. The latter category is hardly repre-
sented in samples extracted from images with maximum X-ray
energy about 10 keV; therefore, the investigations on the frac-
tion of absorbed AGN do in fact apply only to the Compton-thin
sources: fabs = n(1022 ≤ NH ≤ 1024)/n(total).
The empirical results for this fraction, reported in the liter-
ature, have not yet reached full agreement. To first mention the
two papers (Ueda et al. 2003, La Franca et al. 2005) they ad-
dressed the issue of fabs in detail simultaneously with that of the
AGN luminosity function in LX(2 −10 keV), LX hereafter, and its
cosmological evolution, they agree on the existence of an anti-
correlation between fabs and LX, but not on its behaviour with
z. While its normalization and shape stay constant,,according to
Ueda et al. (2003), according to La Franca et al. (2005, LF2005),
fabs(LX) increases with z and becomes shallower, so that the in-
Send offprint requests to: [email protected]
crease is more marked at high than at low luminosities. Akylas
et al. (2006) confirm the existence of the anti-correlation with
LX, but argue that the apparent increase in fabs with z in their
data sample might be due to a systematic overestimate of NH at
high redshifts. Evidence of the anti-correlation with LX has also
been reported by Hasinger (2004) and by Treister & Urry (2005,
see also Barger et al. 2005). To avoid the uncertainties associated
with the estimate of NH, Treister & Urry (2006) use the optical
spectroscopic classification of the counterparts in a combination
of seven X-ray-selected samples and claim evidence (after cor-
recting for selection bias) of an increase with z of the fraction of
obscured AGN. They do not, however,explicitly address the is-
sue of whether the shape of the anti-correlation might change as
well. A correlation between the fraction of broad-line AGN and
their luminosity is reported by Simpson (2005), after analysing
an optically selected sample (z <0.3) and using the [OIII] narrow
emission line to indicate the bolometric luminosity.
Although Dwelly & Page (2006), after a detailed study of
one X-ray deep and narrow beam survey, find no evidence
of both the anti-correlation and of a redshift dependence in
fabs(LX), we consider it reasonable at present to adopt the view
that the existence of the anti-correlation between fabs and LX
is fairly well-supported. For X-ray-selected samples in particu-
lar, we regard the results as more trustworthy that are based on
the combination of deep-narrow and shallow-wide surveys (see
LF2005 for a discussion on this point). On the other hand, we re-
gard the evidence of an increase in fabs(LX) as still controversial
with increasing z, even more so that of a change in its "slope".
In a previous paper (Lamastra, Perola & Matt 2006, PapI), we
proposed a simple model to explain the anti-correlation, which
2
A. Lamastra, G.C. Perola and G. Matt: Compton -- thin type 2 AGN
calls into play interstellar gas for the Compton-thin sources,
while the molecular torus, as invoked in the standard "unifica-
tion" model (Antonucci 1993), remains the best explanation for
the Compton-thick sources. Our proposal is no more than an ex-
tension of the standard model, whose best empirical support is
the finding by Maiolino & Rieke (1995) of a correlation between
absorption and disc inclination in (optically classified as) inter-
mediate type Seyfert galaxies, which is absent in extreme type 2
Seyferts. A finding whose simplest explanation is that the torus
and the disc are, in general, not necessarily coplanar. The model
assumes that the obscuring matter is distributed in a rotation-
ally supported disk, extending out to a few hundred of parsec, as
observed in nearby spiral galaxies mainly in the form of molec-
ular hydrogen. The gravitational pull of the central supermas-
sive black hole (SMBH), combined with that of the bulge (see
Ferrarese & Ford 2005 for a review of the connections between
these two galactic components) on the gaseous disc, as a func-
tion of the distance R from the centre, reduces the gaseous disc
covering factor C as the SMBH mass, MBH, increases. This ef-
fects leads to prediction of an anti-correlation between fabs and
MBH, which is likely to be reflected into the one observed be-
tween fabs and LX. In PapI it was shown that the latter could be
reproduced fairly well in the "local" universe by adopting a "typ-
ical" value of 0.1, irrespective of the mass, of the ratio between
the bolometric and the Eddington luminosity, λ = Lb/LE. This
value of λ was consistent with the estimates available then (see
references in PapI).
Recently, by means of a wealth of spectroscopic data from
the Sloan Digital Sky Survey (SDSS), the empirical knowledge
of the behaviour of λ as a function of MBH has had a substantial
turn, thanks in particular to the work by Netzer & Trakhtenbrot
(2007, NT2007) on broad-line AGN. They explored it for masses
from 107 to 109 M⊙ in the z interval 0.3-0.7, and found a strong
dependence on both MBH and z. It should be noted that several
earlier papers seemed to agree on a practically constant typical
value of λ with increasing z, irrespective of luminosity and mass,
after selection effects had been taken into account (a good ex-
ample, over the range z=0.3-4 in Kollmeier et al. 2006, see also
Lamastra, Matt & Perola 2006). In the framework of a model
where, by definition, the intrinsic properties of the central en-
gine are the same in broad and narrow-line objects, it is nat-
ural to assume that the behaviour found by NT2007 holds for
the narrow-line AGN as well. As a matter of fact, in the "local"
(around z=0.1) universe, a qualitatively similar result for λ(MBH)
was found earlier on by Heckman et al. (2004, H2004) in a large
sample of SDSS selected narrow-line AGN. Despite some criti-
cism of the reliability of the methods used by H2004 expressed
by Netzer et al. (2006), we use their findings as well.
The new light on the MBH -- λ -- z relationships in the range of z
up to 0.7 prompted us to revisit the model of PapI, with the fur-
ther aim of verifying whether the z-dependence of fabs claimed
by LF2005 might be reproduced or otherwise.
2. New light on the MBH -- λ -- z relationships
NT2007 uses a sample of almost ten thousand broad-line AGN,
with mi ≤ 19.1, from the SDSS to investigate the distribution of
λ in different range values of MBH, from 107 to 109M⊙, as a func-
tion of the redshift z. They estimate MBH using the L5100 − RBLR
(the radius of the broad-line region) relationship (as given by
Kaspi et al. 2005) in combination with the FWHM(Hβ). The
value of λ is obtained after applying a (constant) bolometric cor-
rection to L5100. The results, as summarised in their Table 1, ap-
ply to intervals of z up to zmax=0.75, except for the mass range
around 107M⊙ for which zmax=0.3, and down to zmin=0.05 for the
mass ranges around 107 and 107.5M⊙, with zmin increasing up to
0.25 for the mass range around 109M⊙. Thus, in the z interval
0.25 -- 0.75, the results apply to all masses from 107.5 to 109M⊙,
while the overlap is confined to the narrow z interval 0.25 -- 0.3 for
the mass range around 107M⊙. These results concern a feature,
λpeak, defined as the maximum in the observed λ-distributions,
and its dependence on MBH and on z. NT2007 apply reliable sta-
tistical tests to demonstrate that the dependence found is most
likely real, rather than a consequence of the flux limit of the
sample.
To avoid misunderstandings on the extent to which we rely
on NT2007 in the following , it is appropriate to comment on
the "incompleteness" of their sample. As extensively discussed
in particular by Richards et al. (2006), whose aim was to esti-
mate the luminosity function of QSO as a function of the red-
shift, samples of such objects from the SDSS are affected by
incompleteness due to the target selection algorithm, based on
the observed photometric colors, and to the limiting magnitude.
From the statements in NT2007, we infer that the value of λ
(see the marks in their Fig. 3, call it λmark), below which they
regard their sample incomplete, relates to the limiting magni-
tude mi ≤ 19.1. Figure 6 in Richards et al. (2006) shows clearly
that the incompleteness degree increases very steeply for mag-
nitudes above this limit. In the same figure it is apparent that
the incompleteness due to the color selection is of the order of
20% (averaged over 15< mi <19.1) and stays constant up to z
about 2. Going through the way Richards et al. (2006) then used
these results to obtain the proper normalization of the luminosity
function, we are led to conclude that the color-induced selection
effects do impact on the "absolute" values, not on the "relative"
ones. In other words, the "shape" of the distributions of λ above
λmark should not be affected by these effects. Since for our pur-
pose, as will be apparent later, it is the "shape" of these distribu-
tions that is essential, we now describe the way we have adopted
to quantify this shape.
We introduce the distribution function Φ(MBH, λ, z). Figure
3 in NT2007 displays the fraction of AGN as a function of λ
for two ranges of black hole mass (107.5 -- 107.8; 108.5 -- 108.8 M⊙),
each one for three values of z. Except for the range MBH=107.5 --
107.8 M⊙ at z=0.5, λmark is placed somewhat to the left of the
peak in the "observed" distribution in the other five cases. We
thence looked for an analytical function that could represent the
five "observed" distributions above λmark reasonably well. We
eventually adopted a Lorentzian profile:
This paper is organised as follows. Section 2 is devoted to the
new results mentioned above and to their quantitative represen-
tation as is used later, Sect. 3 to the mass function of the SMBH,
Sect. 4 to the molecular gas surface density in the inner region
of galaxies, both late and early-types. Sections 5 and 6 present
the model "predictions" of fabs as a function of Lbol and of LX,
and their comparison with the LF2005 results. A discussion and
a conclusion follow in Sects. 7 and 8.
Φ(MBH, λ, z) = A(MBH ,z)
2π
w
(λ−λpeak)2 +w2
λ > λmark,
(1)
with a "natural" cut -- off at λ=1.
After assigning statistical error bars corresponding to a total
number of 150 objects (a relatively arbitrary choice between the
total numbers from 50 to 230 declared in NT2007 without fur-
ther specifications)to each one of the five histograms in NT2007,
A. Lamastra, G.C. Perola and G. Matt: Compton -- thin type 2 AGN
3
Fig. 3, we operated a χ2 fit to obtain the five values of λpeak and
of w. After noting that all five values of w turned out to be equal
to λpeak multiplied by approximately the same constant, we pro-
ceeded with a simultaneous fit after requiring that this constant
should be exactly the same. The result obtained is w = 1.5λpeak,
with a reduced χ2=0.68. We then adopted this result to w to hold
for any mass value from 107 to 109 M⊙, independent of z. It is
to be noted, at last, that our values of λpeak do not differ sig-
nificantly from those, apparently obtained by NT2007 directly
from the position of the peak in their fractional distributions,
displayed in their Fig. 4. (We later return on the normalization
A(MBH, z).)
For the z -- dependence of λpeak(MBH), NT2007 find (see also
their Fig. 4) that it can be represented (of course only in the
intervals of z effectively covered) with a power law, namely:
λpeak ∝ zγ(MBH )
(2)
where the exponent γ increases with MBH. We therefore adopt
the following z-dependence:
log λpeak(MBH, z) = log a(MBH) + γ(MBH) log z.
(3)
For our application, it is convenient to assign an analytical
form to the function γ(MBH) and a(MBH). For the former, it turns
out that the five values of γ in NT2007, Table 1, are well-fitted
with
γ(MBH) = −3.01 + 0.58 log(MBH).
(4)
For a(MBH), which is not given in NT2007, we have to rely upon
the only two values obtained with the above-mentioned fits. We
choose to connect them with a straight line in the log-log plot of
a and MBH, and obtained
a(MBH) = 105.5(MBH/M⊙)0.74.
(5)
The power-law relationship (3) is illustrated in Fig. 1 for five
representative mass values, from 107 to 109 M⊙. Except for
MBH=107 M⊙, the normalization comes from (5). The extrap-
olation of Eq. (5) to this mass places the power law well above
the three λpeak(z) points in NT2007, Fig. 4, which are reported
in our Fig. 1. We therefore decided to adopt a power law with
a slope as given by Eq. (4) and a normalization in agreement
with these points. The figure illustrates very well that the power
law (3) pertaining to a given mass always stays above the one
pertaining to the next higher mass.
To use the function Φ, there are three other elements to be
taken into consideration. The first would be the extrapolation
of Φ below λpeak(MBH, z). This is extremely uncertain, so we
avoided even trying. We exploit the fortunate circumstance that
the information available on the shape of Φ above λpeak will al-
low us (see Sects. 5 and 6) to predict the quantity fabs for Lbol ≥
1045ergs−1 (corresponding to Lx ≃ 1043.5ergs−1, Marconi et al.
2004), namely the range of Lx where the increase in fabs with z,is
most pronounced according to LF2005. Incidentally, this is also
the luminosity interval where the SDSS quasar luminosity func-
tion was determined by Richards et al. (2006); see also Hopkins
et al. (2007).
The second is the normalization A(MBH, z). Since we use Φ
only for λ ≥ λpeak, it is convenient to introduce the integral quan-
tity
Z 1
λpeak
Φ(MBH, λ, z)dλ = q(MBH, z)
(6)
Fig. 1. The redshift dependence of λpeak for five values of MBH in
M⊙, derived from NT2007 as described in the text. The crosses
are the values obtained by fitting the Lorentzian profile (1), the
hexagons are from Fig. 4 in NT2007
This quantity, which depends on A(MBH, z), represents the frac-
tion of S MBH, as a function of their mass and cosmological
epoch, which are active above λpeak.
There are no elements in NT2007 to infer the value of q. For
our purpose, however, what matters is whether this fraction is
either a constant or a function of MBH; in other words, whether
it preserves the same shape as the active S MBH mass function
(q = constant) or not (q depends on MBH). In Sect. 5 we illustrate
the consequences of both types of assumption.
The third is the mass function (MF) of the SMBH in galaxies,
F(MBH), in particular that of the active ones. The MF is needed
to weight the fractional function Φ when the contributions to
fabs by different masses are combined within the same bin of
luminosity. The MF will be dealt with in Sect. 3.
We lack a similar analysis for λ(MBH, z) carried out explic-
itly on broad-line AGN in the low-redshift universe. Thence we
resorted to the work by H2004 on a large (23,000) sample of
narrow-line AGN drawn from the SDSS. To the extent that this
other type of AGN can be assumed to share the same intrinsic
properties of the broad-line type, we can confidently use their
results to include the local (out to z=0.1) universe in our work.
Unlike NT2007, H2004 estimate the mass using the MBH −
σ∗ (stellar velocity dispersion in the galaxy bulge) relationship
(as given by Tremaine et al. 2002). The need to measure σ∗ prac-
tically excluded from the sample the brightest Seyfert galaxies
and the QSO, a cautionary remark emphasised by H2004. They
derive Lbol from LOIII multiplied by a constant factor (3,500, es-
timated variance of 0.38 dex). This procedure is not completely
reliable, because there seems to be a correlation between the cor-
rection factor and the luminosity (Netzer et al. 2006), but we re-
gard it as acceptable in the first approximation for our purpose.
The quantitative results that we use are those summarised
in H2004, Fig. 3 (left panel), where the integral distribution in
λ of the fraction of SMBH, which are active, is given for six
values of MBH, from 3x106 to 109 M⊙. It can be immediately
seen that, at least qualitatively, the trend of λ with MBH is the
4
A. Lamastra, G.C. Perola and G. Matt: Compton -- thin type 2 AGN
same as found by NT2007. This form of presentation, namely
the fraction of active objects with respect to the "local" F(MBH),
is particularly convenient because it can be directly used, as we
do in Sect. 6, without resorting to assumptions on the quantity
q, Eq. (6). It is also interesting to check whether these distribu-
tions can be represented by the function (1). Since the curves in
H2004, Fig. 3, are devoid of error bars, we could only adjust,
as best as we could, the integral of this function to them. The
outcomes are in general quite satisfactory, and one example is
shown in Fig. 2 (where the curve from H2004 is represented by
a selection of points). Typically the high λ tails stay somewhat
above the H2004 curves. One might require that the excess in the
Lorentzian tail does to some extent take care of the lack of the
brightest objects in the original sample, but we need not insist
on this issue, because the difference is of little consequence for
our final results.
Marconi et al. 2004, Shankar et al. 2004 and references therein).
The one adopted here is from Marconi et al. (2004) and is shown
in Fig. 3, where the MF which includes all morphological types
(their Fig. 2b) is given along with the MF in late, FL, and early,
FE, type galaxies separately (A. Marconi, priv. comm.). This FL
is preferred to the one estimated by Shankar et al. (2004) for the
attention paid to the change in the bulge to the total luminosity
ratio along the Hubble sequence of late type galaxies. The reason
we keep the late separate from the early type galaxies will be
explained in Sect. 7.
For our purpose, what does matter is the shape of the MF.
Insofar as the growth of the SMBH is a consequence of accretion
and is therefore accompanied by the AGN type activity, it is most
likely that the shape of the MF for the "active" SMBH changes
with redshift and is different from that of the whole population
as observed in the local universe. In principle, the redshift depen-
dence of the "active" SMBH mass function, F ∗(MBH, z), could
be inferred from the AGN luminosity function as a function of z,
provided that a simple relationship holds between MBH and the
luminosity (for instance, λ = constant). In practice, as discussed
in Marconi et al. (2004) and as has become particularly evident
after NT2007, the conversion is neither straightforward nor uni-
vocal. Hence F ∗(MBH) can only be reliably estimated through
a "direct" evaluation of MBH in a properly selected sample of
AGN, with special care devoted to correcting for incomplete-
ness selection effects. On the basis of SDSS samples, H2004
(z ≤ 0.1) for narrow emission line AGN and Greene & Ho (2007,
z ≤ 0.3, with masses derived as in NT2007) for broad emis-
sion line AGN, both find that in the local universe F ∗(MBH) is
steeper, above 107 M⊙, than F(MBH): a difference attributed to
the so-called downsizing
In Sects. 5 and 6 we, in a first instance, adopt the "shape"
of the local F(MBH); however, the impact on the results of other
options for the downsizing is also discussed.
Fig. 2. Integral distribution in λ of the active fraction of SMBH,
MBH=108 M⊙, out to z=0.1. The crosses are derived from H2004,
Fig. 3, the line is our Lorentzian representation.
The parameters extracted from H2004 that we use in Sect. 6
are summarised in Table 1. The Lbol,0 in the second column (with
the corresponding λ0 in the third) represents a completeness
limit valid for all masses between 107 and 109 M⊙. The fourth
and fifth columns contain, repectively, λmax, the maximum value
that can be read in H2004, Fig. 3, and qH, the fraction of active
SMBH above λ0 from the same figure. From the sixth column
onward the quantities given were obtained using the Lorentzian
profile, as explained above. The quantity λpeak is very low and
badly determined, but the quantity that really matters is w; the
fraction qHl in the last column differs only marginally from qH.
It should be noted that the fraction of active SMBH in H2004
is given with respect to the local F(MBH), such as the one given
by Marconi et al. (2004) and described in the next section.
3. The mass function of SMBH
The number density of SMBH as a function of mass, F(MBH),
has been estimated in the local universe by several authors (see
Fig. 3. The local SMBH mass function from Marconi et al.
(2004). Dotted line: late type galaxies; broken line: early type
galaxies.
A. Lamastra, G.C. Perola and G. Matt: Compton -- thin type 2 AGN
5
Table 1. The parameters used out to z=0.1, inferred from H2004
log MBH (M⊙)
log Lbol,0 (erg/s)
7
7.5
8
8.5
9
44.8
44.8
44.8
44.8
44.8
λ0
0.5
0.16
0.05
0.016
0.005
qH(λ > λ0)
0.32x10−3
1.44x10−3
2.3x10−3
1.26x10−3
0.17x10−3
λmax
1
0.8
0.4
0.13
0.04
λpeak
0.035
0.006
0.005
0.0005
0
w
0.053
0.02
0.012
0.004
0.0003
qHl(λ > λ0)
0.41x10−3
1.53x10−3
2.3x10−3
1.26x10−3
0.17x10−3
nearby and bright spiral galaxies mapped at the 3 mm CO J=1 --
0 line, the BIMA SONG survey (Helfer et al. 2003). At their
average distance of 12 Mpc, the angular resolution of 6 arcsec
corresponds to a radial distance from the centre of about 210 pc.
4. The molecular gas surface density Σ
The basic feature of the model in PapI is the gravitational force
exerted by the BH (and by the associated stellar bulge) on a ro-
tationally supported disc made of molecular clouds. This force
shapes the gas distribution profile, as a function of the radial dis-
tance R from the BH, in such a way as to determine an anticorre-
lation between MBH and the "covering factor" determined by the
disc. Quantitatively this factor was calculated in PapI assuming
that the distance out to which the disc extends is at least equal
to 2Rin f l (Rin f l is the distance from the BH at which the molec-
ular gas distribution profile has a point of inflection, see Eq. (9)
in PapI), that is, about 25 pc when MBH=106 and extending to
about 450 pc when MBH=109 M⊙.
Figure 4 shows this "covering factor", C(Σ, MBH), as a func-
tion of the surface density Σ, from 0 to 2000 M⊙ pc−2, for four
values of MBH, from 106 to 109 M⊙. After a steep growth, the
curves enter a regime where the increase is very slow. The rather
sharp change in slope takes place around a Σ value from about
200 to about 400 M⊙ pc−2 when moving down from MBH=109
to MBH=106 M⊙. Thus, in predicting fabs, attention must be paid
to the distribution of Σ adopted, particularly on the fraction of
objects below those values.
Fig. 5. The molecular gas central, face-on surface density Σ of
the spiral galaxies in the BIMA SONG survey, versus the mass
estimate of the SMBH hosted by these galaxies. Hexagons: MBH
from the MBH − Lbulge relationship; open: from Dong & De
Robertis (2006); filled: our work. Triangles: from the MBH − σ∗
relationship, Batcheldor et al. (2005), Merloni et al. (2003), Woo
& Urry (2002), Pellegrini (2005). Crosses and pentagon: from
maser and from stellar kinematics, Ho (2002). Square: from re-
verberation mapping, Kaspi et al. (2000).
In Fig. 5 the central face-on Σ of these objects is plotted as a
function of the mass of the SMBH (estimated as explained in the
figure caption). For the majority of these galaxies MBH is com-
prised between 106.5 and 108.5 M⊙. In practice this is the range
of masses for which the information described in Sect. 2 will be
used to predict fabs in Sects. 5 and 6. Given the small number of
objects in this sample, we use the observed distribution of Σ in
the 44 objects to evaluate the uncertainty on fabs.
In PapI we stressed that the effect under study does not re-
quire any type of correlation between the central activity and
the presence of the absorbing gas, the latter being "in place" as
a property of the galactic interstellar medium. This statement
can be regarded as basically correct for the late type galaxies
(not only in the local universe, as shown by the survey just men-
tioned, but presumably also at cosmological distances), and now
Fig. 4. The covering factor C as a function of Σ, according to
the model, for different values of the SMBH mass. From top to
bottom: 106, 107, 108, and 109 M⊙.
In PapI (to which we refer the reader for more details), we
used the only results obtained with sufficient angular resolution
in a molecular line. These are represented by a sample of 44
6
A. Lamastra, G.C. Perola and G. Matt: Compton -- thin type 2 AGN
we need to consider to what extent it might be applied to the
early type galaxies.
Elliptical (E) and lenticular (S0), unlike the spiral (S) galax-
ies, are rare in the field and frequent (the bright E's in particular)
in galaxy groups and clusters. Therefore, useful interferometric
observations are only available for a dozen of them, optically lu-
minous or very luminous (thus hosting SMBH well above 107
M⊙), out to a maximum distance of about 85 Mpc and with a
strong CO line signal. It is convenient to concentrate first on
these results, and we limit ourselves to those obtained by Young
(2002, 2005) with the BIMA interferometer on seven "true" E
(that is galaxies with an r1/4 photometric profile). Molecular gas
is found distributed in rotating, mostly symmetric discs, with
radii ranging from 1 to 6 kpc. For each of them we report the
peak surface brightness, translated into a surface density of H2
gas, along with the galaxy surface area corresponding to the
beam. UGC1503 (field): 60 M⊙ pc−2 (2.4x2.1 kpc2); NGC807
(field): 40 M⊙ pc−2 (2.4x2.0 kpc2); NGC3656 (merger rem-
nant): 720 M⊙ pc−2 (1.7x1.3 kpc2); NGC4476 (Virgo cluster):
112 M⊙ pc−2 (0.73x0.49 kpc2); NGC5666 (field): 115 M⊙ pc−2
(1.7x1.3 kpc2) (from Young 2002). NGC83 (group): 135 M⊙
pc−2 (2.7x2.5 kpc2); NGC2320 (A569): 353 M⊙ pc−2 (2.4x2.2
kpc2) (from Young 2005). On account of the values of the
area/beam, it is evident that these selected cases correspond to
the values of Σ, which imply a substantial covering factor C and
which occur in a large fraction of the S galaxies in Fig. 5. But
how frequent the occurrence is of such a situation in E and S0
galaxies? The answer to this question does not appear to be well
settled yet. Furthermore, most data are from single-dish obser-
vations, which can be properly converted only into a total H2
mass.
On this issue it seems to us appropriate to first quote the com-
pilation by Bettoni, Galletta & Garcia-Burillo (2003) of a wealth
of data obtained prior to the year 2003. Most useful for our pur-
pose are this mean sample values, subdivided per morphological
class according to the mean numerical Hubble stage index t as
tabulated in the Third Reference Catalogue of Bright Galaxies
(RC3 de Vaucouleurs et al. 1991), of the quantity M(H2)/D2
25,
where D25 is the linear size corresponding to the 25 mag/arcsec2
isophote. After assembling their classes into three groups, G1
(t from -5 to -3, E), G2 (t from -2 to 0, S0), and G3 (t from
1 to 6, Sa to Sc), from Table 5 in Bettoni, Galletta & Garcia-
Burillo (2003), we obtain the following proportions for the quan-
tity M(H2)/D2
25, G1:G2:G3 = 0.06:0.29:1. Thus, with the cau-
tionary remark that the size of the H2 gas structure in the three
groups may not scale similarly with D25, the immediate conclu-
sion one can draw is that in the ellipticals the surface density of
molecular gas is, on average, more than one order of magnitude
less than in spirals, while it is a factor of about three less in the
lenticulars. This conclusion we believe is stronger than the one
that can be drawn from the detection rate, which is very sen-
sitive to the sample selection, as discussed in Combes, Young
& Bureau (2007); see also Welch & Sage (2003), Sage, Welch
& Young. (2007) and references in these papers. An additional
important result emerging from these papers is that, among the
early type galaxies, the ratio M(H2)/L decreases with increasing
luminosity L, despite the increase in depth of the potential well.
Taken together with the results quoted from Bettoni, Galletta &
Garcia-Burillo (2003), this finding leads us to conclude that the
more luminous (and therefore hosting the more massive SMBH)
an early type galaxy, the less the chance of finding "in place" H2
surface densities similar to the ones reported above from Young
(2002, 2005).
For high z objects, we would like to mention the most obvi-
ous argument used to account for the MBH − σ∗ and MBH − L
observational relationships (see Ferrarese & Ford 2005 for a re-
view), where σ∗ and L are the stellar velocity dispersion and
the luminosity, respectively, of galactic "bulges" or "spheroids".
The argument goes that the formation of the spheroids is closely
connected with the growth of the SMBH sitting at their cen-
ters. If the two phenomena had progressed hand-in-hand from
the earliest to the present times, it would be legitimate to pos-
tulate that a phase of intense activity (AGN phase) should in
general be accompanied by star formation on the spatial scale
of the bulge. The latter requires the accumulation on this scale
of gas (whatever its origin) in a dense, most likely molecular
state, wherein the star formation takes place. From an obser-
vational point of view the validity of such a postulate rests on
results obtained with studies of vast samples of galaxies, based
on SDSS spectrography and photometry. Here we refer in par-
ticular to Kauffmann et al. (2003a), Kauffmann et al. (2003b),
H2004. Quoting from the review by Heckman & Kauffmann
(2006): "Strong emission -- line AGN inhabit those unusual galax-
ies that are both relatively massive and dense, yet have a sig-
nificant young stellar population". The denomination "unusual"
emphasises the contrast with the usual make-up of galaxies with
similar mass and structure, which, within the limits in sharpness
of the observational distinctions employed, contain only a popu-
lation of old stars.
A cautious theoretical elaboration on these results must in-
evitably contemplate the possibility that the SMBH activity is
a consequence of the transformation into stars of the dense gas
and the dynamical implications of supernovae explosions in par-
ticular. In this instance, it could well be that, during most of the
active phase of the SMBH, the dense gas is no longer present
in sufficient quantity on the spatial scale required by our model.
Furthermore, it is doubtful that the structure of the molecular gas
were dominated, in such a phase, even if present, by a rotation-
ally supported configuration out to several 100 pc. Only direct
observations in molecular lines, with the adequate angular reso-
lution, will answer these questions.
To reach simple and sharp conclusions, we therefore con-
sider two instances. In Sects. 5 and 6, in order to emphasise only
the consequences of what was described in Sect. 2, we shall posit
that, for all AGN no matter the morphological type of the host
galaxies, the model applies equally well and a single represen-
tative distribution of Σ will be used. In Sect. 7, we posit that the
above applies only to the late type galaxies, while in the active
early type galaxies the "covering factor" is basically unknown.
The latter instance requires that FL(MBH) be kept separate from
FE(MBH).
5. Predicting fabs as a function of luminosity for
MBH > 107 M⊙, at z=0.35
The fraction of absorbed AGN, fabs, in a given luminosity bin
(Lb1 < Lb < Lb2) can be computed using the equation
fabs = P jPiC(Σ j,MBH,i)f (Σ j)∆Σ jF(MBH,i)∆MBH,iR λ2
λ1
Φ(MBH,i,λ,z)dλ
(7)
Pi F(MBH,i)∆MBH,iR λ2
λ1
Φ(MBH,i, λ, z)dλ
Lb1
LEdd (MBH,i) (λ1 ≥ λpeak) and λ2 =
where λ1 =
LEdd (MBH,i) (λ2 ≤ 1),
f (Σ j) is the fractional distribution in the quantity Σ, introduced
to take properly into account the strong dependence of C(Σ) on
Lb2
A. Lamastra, G.C. Perola and G. Matt: Compton -- thin type 2 AGN
7
low values of the argument. In our application we used mass bins
0.3 dex wide, and the results are presented in bins of Lb that are
0.5 dex wide.A normalization of the function Φ, as explained in
Sect. 2, will depend on the assumption we make on the quantity
q(MBH) given by Eq.(6).
To predict fabs(Lb) with Eq. (7), we only consider the ac-
tive SMBH with λ > λpeak(MBH). When the relationshps (3) and
(5) are combined, it turns out that, at one particular value of the
redshift, z=0.35, λpeak(MBH) ∝ M−1
BH very closely. Therefore, all
masses in the range 107.5 -- 109 M⊙, to which at z=0.35 both (3)
and (5) apply, contribute above the same value of Lb, equal to
1045ergs−1. In other words, the active BH in this range of mass
with λ < λpeak contributes only below 1045ergs−1. Notably, this
is the Ledd of MBH = 107 M⊙. Thus, to obtain a "complete" es-
timate of fabs in the interval of logLb = 45 -- 45.5, only the con-
tribution of the masses in the range 107 -- 107.5 M⊙ needs to be
added, which we did by resorting to a small extrapolation of the
curve labelled 107 M⊙ in Fig. 1 beyond its formal limit at z=0.3.
This convenient situation will be exploited to make some clear
point, and to predict fabs at a given z to be used in Sect. 6 for
comparison with the predictions at higher and lower values of
the redshift.
As anticipated in Sects. 3 and 4, we first adopt the "shape"
of the local F(MBH), Fig. 3, as representative of the ensamble
of active SMBH in both late and early type galaxies, and assume
that the "model" applies equally well to the two types of hosts
with the same distribution of Σ. Concerning the quantity q (Eq.
(6)), we start with the assumption that it is equal to a constant,
namely independent of MBH. It is perhaps worth emphasising
that the value of this constant is irrelevant for obtaining the frac-
tion fabs, while its precise knowledge would be necessary if we
wanted to predict the number, for instance per unit volume, of
the two types of AGN. The result is given as a histogram in Fig.
6. The first aspect deserving a comment is the sensitivity of the
result to the choice concerning Σ. The values of fabs(Lb) in the
individual bins of the histogram have been computed using the
"observed", local, distribution of the 44 values of Σ referred to in
Sect. 4, and the bars represent the 68% confidence intervals cal-
culated on the basis of the limited number of objects in that dis-
tribution. For the sake of neatness, in the following similar his-
tograms, the bars will no longer be reproduced. For comparison,
in the same figure fabs is also given for two choices of a "single"
representative value of Σ, namely the median (150 M⊙ pc−2) and
the mean (270 M⊙ pc−2) of the same distribution. In both cases,
especially in the second, the prediction is significantly higher in
all bins. This follows from the fact that (see Fig. 4), when a sub-
stantial fraction of objects span the interval from 0 to about 300
M⊙ pc−2, the covering factor is extremely sensitive to the way
the values of Σ are distributed. In this respect, the prediction of
fabs obtained adopting Σ=270 M⊙ pc−2 (as it can be judged from
Fig. 4) is rather close to the ceiling attainable by pushing the
distribution toward much higher values of Σ than adopted here:
a point on which we return in Sect. 7.
It is important to note that, without the "physical" limit
of unity for λ, the result would be different. For instance, if
no cut-off at λ=1 in Eq. (1) is imposed, the computation of
fabs results is approximately a constant value, irrespective of
luminosity. This occurs because a single mass range, 107-107.3
M⊙, turns out to dominate at all luminosities, and the value
of fabs to be close to the one pertinent to these mass values.
Moreover, in this instance, we have no information to adopt
from NT2007 in order to estimate the contribution of the
masses below 107 M⊙, a very serious cause of uncertainties,
Fig. 6. The predicted fabs at z=0.35, obtained using the local
SMBH MF, q=const, with three different choices for Σ. Full line
histogram: the observed distribution of Σ described in the text,
the bars represent the systematic uncertainty due to the limited
number of objects in this distribution. Dotted and broken line
histograms: Σ=150 and 270 M⊙ pc−2 respectively.
due to the relatively large number of SMBH in the mass range
106-107 M⊙. Because the relationship between λpeak and MBH
remains qualitatively the same at
the redshift values dealt
with in the following section, the same type of result, with the
same type of uncertainties, is also obtained at the other redshifts.
To illustrate the consequence of taking into account the
downsizing effect mentioned in Sect. 3, we only need to change
the leading assumption on q, namely adopt a proper dependence
of q on MBH. The shape of the broad-line AGN mass function
in the local universe, as obtained by Greene & Ho (2007) above
107 M⊙, suggests a dependence of the type q ∝ (MBH/107)−0.5.
To show that the knowledge of the exact difference between F ∗
and F is not very important for this range of mass, we assumed
an even steeper correction as an exercise, namely q ∝ M−1
BH. The
result is shown in Fig. 7. The anticorrelation becomes somewhat
steeper, but the difference overall is not particularly marked.
This is so because, even though the absolute number of objects as
a function of Lb has changed radically, the fraction of absorbed
sources depends solely on the mix of values of MBH contributing
to each bin of luminosity. In particular, for instance, in the high-
est lumonosity bin the change is almost imperceptible because,
despite the greatest change (decrease) in the absolute number of
objects, the interval of MBH values contributing to it have re-
mained almost unchanged.
To proceed to a comparison with the results obtained in the
hard X-rays, we need to turn from Lb to Lx. The bolometric cor-
rection used comes from Marconi et al. (2004) and is not linear.
We therefore recalculated Eq. (7) after converting Lb into Lx.
In the right panel of Fig. 7 the expectation for the two options
on q is compared with the results obtained by LF2005 around z
= 0.35. The two data points, with Poissonian error bars, repre-
sent the objects in their samples within the interval z = 0.2-0.5,
while the solid line represents the outcome at z=0.35 of their
8
A. Lamastra, G.C. Perola and G. Matt: Compton -- thin type 2 AGN
Fig. 7. Left panel: full line histogram, same as in Fig. 6; dotted
line histogram, q ∝ M−1
BH. Right panel: histograms as in left panel
as a function of LX; solid line from the global fit by LF2005; the
data points, with Poissonian uncertainties, include the objects
with z=0.2-0.5 in the sample used by LF2005.
Fig. 8. Same as Fig. 7, but only with q=const, at z=0.7. The data
points include the objects with z=0.55-0.85 in the sample used
by LF2005.
"global" fit of the intrinsic distribution after correcting for se-
lection effects. It must be noted that the "global" fit given by
LF2005 (dotted lines in their figure 11) also includes Compton-
thick AGN with 1024 < NH < 1025, so we recalculated their
best-fit relation to take only the contribution of the Compton-
thin AGN into account.
The difference between the data points and the line illustrates
the relevance of the various selection effects that LF2005 take
into account As it can be seen, our model predicts a decrease
in the absorbed AGN fraction with increasing X-ray luminosity,
but the slope of the predicted anticorrelation is not as steep as in
LF2005.
6. Redshift dependence in fabs
We next move up to z=0.7, the highest redshift where the
NT2007 findings do apply. The results on fabs (Eq. 7), keeping
q = constant, are shown in the left panel of Fig. 8, as a function
of Lb, for Lb ≥ 1045.4ergs−1, where the completeness is ensured,
and in the right panel of Fig. 8, as a function of Lx. Notably the
values of fabs are almost identical to those calculated at z=0.35
for the same assumption on q. Despite the significant differences
in λpeak(MBH) at the two redshifts, the parameters into play con-
jure to give essentially the same quantitative outcome: not sur-
prisingly in the highest bin of Lx, which remains dominated by
the highest masses. Even if, at this particular redshift, the global
best fit at z=0.7 from LF2005 (solid line) appears to be in rea-
sonably good agreement with our prediction, the fundamental
point is that the z-dependence of the λ(MBH) distribution does
not in itself introduce any increasing trend in fabs, at least be-
tween z=0.35 and z=0.7. This conclusion is strengthened by the
fact that a possible dependence of q on MBH, reflecting changes
in the shape of F ∗(MBH) above 107 M⊙, is likely to be less steep
at z=0.7 than assumed at z=0.35, for demonstrative purpose, in
the previous section, and would therefore make no significant
difference for the conclusion.
We now move to the local Universe. We used the informa-
tion extracted from H2004, as described in Sect. 2 and sum-
marised in Table 1. Figure 9, left panel, shows the results
on fabs(Lb) obtained in two ways, one directly from the curves
in Fig. 3 of H2004, the other from their Lorentzian represen-
tation. The latter allowed us to go beyond logLb=46, although
one should be aware that the absolute number of very lumi-
nous objects is in any case very small in the local universe. Thus
for our purpose, we consider uniportant that the H2004 sample
lacks, by construction, the most luminous objects. In the interval
logLb=44.8-46, the fabs expected from the H2004 curves or from
their Lorentzian representation are practically indistinguishable.
Fig. 9. Same as Fig. 7 at z=0.1, obtained using the H2004 results,
see text. The dotted line histogram refer to the H2004 plots, the
full line one to their representation with the function given by
(1).The data points include the objects with z≤0.2 in the sample
used by LF2005.
After correcting Lb to LX, the expected fabs is diplayed in
Fig. 9, right panel. Here again the data points (z≤0.2) and the
best fit result from LF2005 are also given. The comparison
shows that the predicted slope of the anticorrelation is definitely
flatter than obtained by LF2005 at z=0.1.
To summarise these results, the fabs predicted as a function
of Lb and LX, for the three values of z covered (with q=const for
A. Lamastra, G.C. Perola and G. Matt: Compton -- thin type 2 AGN
9
z=0.35 and 0.7) are collected in Fig. 10. The simple conclusion
that can be drawn immediately is that, despite the strong depen-
dence of λ on MBH and on z, the normalization and slope of the
anticorrelation between fabs and LX, in the range of luminosities
where they appear to change most quickly with z according to
LF2005, is expected to change very little, if at all.
Fig. 11. The curve represent fabs at z greater than 1, predicted af-
ter adopting a fixed value of λ=0.25 irrespective of the SMBH
mass. The histogram of the prediction at z=0.35, q=const, is
shown for comparison.
Fig. 10. The predicted fabs at z=0.1 and, with q=const, at z=0.35,
0.7. In the right panel the dotted, solid, and dashed lines are from
the global fit by LF2005 at z=0.1, 0.35, and z=0.7 respectively.
We conclude this section by tentatively pushing our predic-
tions beyond z=1. In this respect we note that, up to z=0.7 (see
Sect. 2) the λ − MBH relationships for the different values of MBH
do converge. Thus we see no contradiction with results, in partic-
ular those obtained by Kollmeier et al. (2006) and mentioned in
Sect. 1, going well beyond z=1, namely that out to z about 4 the
mean value of λ seems to stabilize at a value, <λ>=0.25 indepen-
dent of both MBH and z, and that its distribution is approximately
lognormal. A recent investigation by Netzer et al. (2007) does es-
sentially confirm those results for what concerns the mean value
of λ in the redshift bin 2.3 -- 3.4 and the mass range 108.8-1010.7
M⊙, while the distribution is somewhat broader than found by
Kollmeier et al. (2006). A simplified approach, which ignores
the distibution and associates the same value of λ, equal to 0.25,
to any value of MBH ≥ 107 M⊙, leads, when adopting the same
Σ distribution used previously, to the prediction of fabs given in
Fig. 11, which, by construction, is independent of the SMBH
mass function, and hence of z. The only, albeit small, difference
relative to the prediction at z=0.35, shown for comparison in the
same figure, is that the expected fabs becomes somewhat steeper,
contrary to the LF2005 results.
7. Discussion
Despite the uncertainty on the actual form of F ∗(MBH, z) com-
pared to that of the local F(MBH), we have demonstrated that the
predicted behaviour of fabs at high luminosities does not repro-
duce the increase with z found by LF2005, as long as the dis-
tribution of Σ is kept fixed and indifferent to the morphological
type of the host galaxy. This result can be understood easily, be-
cause the trend in λ(MBH, z), in particular the way this quantity
for different masses converge upwards with increasing z, implies
that the higher the redshift, the higher the relative contribution
of the more massive SMBH (with the lower C) at high luminosi-
ties. Another way to appreciate this point is to note that, being
LE(108M⊙)=1046 ergs−1 (LX = 1044.5 ergs−1), beyond this lumi-
nosity only SMBH with higher masses contribute whose C stays
in the range about 0.3; therefore, an increase in fabs above this
range cannot be predicted.
We are therefore left to operate on the other parameter of
the model, namely Σ. On this quantity the empirical information
we would need is practically missing, and only ad hoc assump-
tions can be made. It should be noted immediately that, as shown
in Fig. 6, the increase in a "typical" value of Σ shifts fabs up-
wards, without markedly altering its slope, in fact just rendering
it somewhat steeper. Furthermore, in Sect. 5 we emphasised that
the uppermost values in Fig. 6 practically represent a ceiling, as
a consequence of the almost flat progression of the C with in-
cresing Σ beyond about 300 M⊙ pc−2, irrespective of MBH (see
Fig. 4). Thus, in the context of our model, there are limits on the
values that fabs can achieve, no matter the assumptions we are
willing to make.
As an exercise, let us assume that, in the local universe, the
Σ distribution adopted in the previous two sections applies only
to the late type galaxies, the early types being comparatively, on
average, much poorer in molecular gas (Sect. 4). For simplicity,
we assume that Σ is practically zero in the early type galaxies.
Because the SMBH above 107.5−8 M⊙ are mainly hosted by this
type of galaxies, above LX=1043.5 this assumption implies a ma-
jor drop in fabs, as illustrated in Fig. 12 at z=0.1, and the dis-
crepancy with respect to the best fit by LF2005 is much stronger
than in Fig. 9. Thence, to obtain a prediction closer in the local
universe, as well as with increasing z, to the results in LF2005,
one would need to assume that the presence of obscuring gas in
at least the more massive early type galaxies (with the appropri-
ate values of Σ and organised as the model requires) is correlated
with the nuclear activity and Σ itself increases with z. One might
adopt as a hint, in favour of the latter assumption, the increase
with z in the "volume-averaged" star formation rate (SFR) (see
also H2004 for further considerations). However, this is not so
straightforward, because the evolution of the X-ray luminosity
function of AGN is not only in their luminosity but also in their
number density: thus, what occurs in the unit volume does not
necessarily apply to the individual objects. Perhaps the simplest
speculation, that one can naively propose, is that the quantity Σ
effectively present, or at least surviving during the activity phase,
10
A. Lamastra, G.C. Perola and G. Matt: Compton -- thin type 2 AGN
increases with λ. However, as noted above, it would remain hard
for our model to explain values of fabs greater than about 0.3 at
the highest luminosities.
Unfortunately, no theoretical picture of the gas spatial distri-
bution and kinematics, the interplay with the SMBH growth, on
the one hand, and the SFR on the other, has been developed so
far to a degree of (physically consistent) detail necessary to sen-
sitively adjust the prediction of fabs. On the observational side,
we lack sufficient data to proceed empirically, as we have done
with λ.
We finally need to point out the non-negligible difference
between the optical and the X-ray selection of AGN. Following
Rigby et al. (2006) (see also Fiore et al. 2003, Cocchia et al.
2007 and references in all three papers), we stress that a sub-
stantial fraction of AGN selected through their X-ray emission
and classified as such on the basis of their LX would be missed
through an optical spectroscopic classification (this fraction in-
creases from about 15% locally to about 50% at z around 1).
These objects could perhaps reduce the difference between our
predictions and the fabs(z) found by LF2005. Such a possibility
could be tested only after a properly complete investigation of
the λ of these objects is carried out.
Fig. 12. Same as Fig. 9, with the addition of the fabs (full line
histogram) predicted assuming Σ=0 in early type galaxies.
we did concentrate, might possibly be reproduced by assuming
that the typical values of Σ do increase with the redshift and,
furthermore, that in the early type galaxies, at least in the more
massive ones, the presence of this quantity is correlated in time
with the activity of their nucleus. At the highest luminosities,
however, according to our model, fabs can hardly exceed 0.3.
For the sake of completeness, it should be noted that the results
just summarised are sensitive to the presence of a cut-off in the
λ(MBH, z) distribution, and they correspond to the case where
the cut-off is placed at the physical limit λ=1.
Steps forward, that are relevant not only to the model dis-
cussed in this paper can be made at least in two ways. The first
and most obvious one requires systematic investigation of the
amount of molecular gas in an appropriately selected sample of
early type galaxies, both active and inactive, out to redhifts of
0.5 at least. The line strength and profile may not tell us enough
about the spatial distribution, but much about the amount and the
kinematics of the gas. The former property needs to wait for ob-
servations with a large interferometric array of telescope, such
as ALMA. The second step is somewhat indirect and therefore
requires interpretation. The method is described and applied to a
handful of broad-line objects in Maiolino et al. (2007). It allows
an estimate of C to be inferred for each object on the basis of
measurements of infrared spectral features due to dust reradia-
tion of the ultraviolet from the active nucleus. It is worth noting
that Maiolino et al. (2007) find an anti-correlation between C and
either the luminosity or the mass, thus leaving completely open
at present which of the two parameters is the one "physically"
driving the anti-correlation. This sort of degeneracy could be re-
solved by comparing the covering factor of objects with similar
MBH and widely different values of Lb, hence of λ. According to
our model a similar value of C should be found in such objects.
Acknowledgements
The authors are particularly grateful to Fabio La Franca for his
help in assessing the figures with the X-ray data, and thank sev-
eral colleagues for enlightening discussions. They acknowledge
financial support from the ASI (grant 1/023/05/0) and MiUR
(2006025203).
References
Akylas, A., Georgantopoulos,
I., Georgakakis, A., Kitsionas, S., &
Hatziminaoglou, E., 2006, A&A, 459, 693
8. Conclusions
In the light of recent new results on the behaviour of the
Eddington ratio λ as a function of the SMBH mass and of the
redshift, we have revisited the predictions of the model pre-
sented in PapI. This model naturally predicts an anti-correlation
between the fraction fabs of X-ray obscured, Compton-thin
AGN, and MBH, provided there is interstellar molecular gas with
appropriate values of surface density Σ approximately within
25 to 450 pc (depending on MBH) from the active nucleus.
We used the newly found λ(MBH, z) dependence (z from the
local universe to 0.7) and a distribution of Σ, irrespective of
the morphological type of the host galaxies and invariant with
z, to convert that prediction into one versus LX. The results
appear to reproduce, at least qualitatively, the anti-correlation
between fabs and LX claimed by several authors, but not the
still controversial claim of an increase in fabs with z. The latter
increase, particularly in the form claimed by LF2005 on which
Antonucci, R.R.J., 1993, ARA&A, 31, 473
Barger, A.J., Cowie, L.L., Mushotzky, R.F., et al., 2005, AJ, 129, 578
Batcheldor, D., Axon, D., Merritt, D., et al., 2005, ApJS, 160, 76
Bettoni, D., Galletta, G., & Garcia-Burillo, S., 2003, A&A, 405, 5
Combes, F., Young, L.M. & Bureau, M., 2007, MNRAS, 377, 1795
Cocchia, F., Fiore, F., Vignali, C., et al., 2007, A&A, 466, 31
de Vaucouleurs, G., de Vaucouleurs, A., Corwin H. G., et al., 1991, Third
Reference Catalogue of Bright Galaxies (RC3) (New York Springer-Verlag)
Dong, X.Y. & De Robertis, M.M., 2006, AJ, 131, 1236
Dwelly, T. & Page, M.J., 2006, MNRAS, 372, 1755
Ferrarese, L. & Ford, H., 2005, SSRv, 116, 523
Fiore, F., Brusa, F., Cocchia, F., et al., 2003, A&A, 409, 79
Greene, J.E. & Ho, L.C., 2007, ApJ, 667, 131
Hasinger, G., 2004, Nucl. Phys. B, 132, 86
Heckman, T., Kauffmann, G., Brinchmann, J., et al., 2004, ApJ, 613, 109
(H2004)
Heckman, T. & Kauffmann, G., 2006, New Astr Rev, 50, 677
Helfer, T.T., Thornley, M.D., Regan, M.W., et al., 2003, ApJS, 145, 259
Ho, L.C., 2002, ApJ, 564, 120
Hopkins, P. F., Richards, G., T., Hernquist, L., 2007, ApJ, 654, 731
Kaspi, S., Smith, P.S., Netzer, H., et al., 2000, ApJ, 533, 631
Kaspi, S., Maoz, D., Netzer, H., et al., 2005, ApJ, 629, 61
A. Lamastra, G.C. Perola and G. Matt: Compton -- thin type 2 AGN
11
Kauffmann, G., Heckman, T.M., White, D.M., et al., 2003a, MNRAS, 341, 54
Kauffmann, G., Heckman, T.M., Tremonti, C., et al., 2003b, MNRAS, 346, 1055
Kollmeier, J. A., Onken, C. A., Kochanek, C. S., et al., 2006, ApJ, 648, 128
La Franca, F., Fiore, F., Comastri, A., et al., 2005, ApJ, 635, 864 (LF2005)
Lamastra, A., Perola, G.C., & Matt, G., 2006, A&A, 449, 551 (PapI)
Lamastra, A., Matt, G., & Perola, G.C., 2006, A&A, 460, 487
Maiolino, R., Rieke, G. H., 1995, ApJ, 454, 95
Maiolino, R., Shemmer, O., Imanishi, M., et al., 2007, A&A, 468, 979
Marconi, A., Risaliti, G., Gilli, R., et al., 2004, MNRAS, 351, 169
Merloni, A., Heinz, S. & di Matteo, T., 2003, MNRAS, 345, 1057
Netzer, H., Mainieri, V., Rosati,P., & Trakhtenbrot, 2006, A&A, 453, 525
Netzer, H. & Trakhtenbrot, B., 2007, ApJ, 654, 754 (NT2007)
Netzer, H., Lira, P., Trakhtenbrot, B., Shemmer, O., & Cury, I., 2007, ApJ, in
press (astro-ph/0708.3787)
Pellegrini, S., 2005, ApJ, 624, 155
Perola, G.C., Puccetti, S., Fiore, F., et al., 2004, 421, 491
Richards, G. T., Strauss, M. A., Fan, X., et al., 2006 AJ, 131, 2766
Rigby, J.R., Rieke, G.H., Donley, J.L., Alonso-Herrero, A. & Perez-Gonzales,
P.G., 2006, ApJ, 645, 115
Sage, L.J., Welch, G.A. & Young, L.M., 2007, ApJ, 657, 232
Shankar, F., Salucci, P., Granato, G.L., De Zotti, G. & Danese, L., 2004,
MNRAS, 354, 1020
Simpson, C., 2005, MNRAS, 360, 565
Treister, E. & Urry, M., 2005, ApJ, 630, 115
Treister, E. & Urry, M., 2006, ApJ, 652, L79
Tremaine, S., Gebhardt, K., Bender, R., et al., 2002, ApJ, 574, 740
Ueda, Y, Akiyama, M., Ohta, K., Miyaji, T., 2003, ApJ, 598, 886 (eds. J.H.
Knapen, J.E. Beckman, I. Shlosman and T.J. Mahoney), ASP Conference
Series, Vol. 249, 389
Welch, G.A. & Sage, L.J., 2003, ApJ, 584, 260
Woo, J. & Urry, C.M., 2002, ApJ, 579, 530
Young, L.M., 2002, ApJ, 124, 788
Young, L.M., 2005, ApJ, 634, 258
|
astro-ph/0409650 | 1 | 0409 | 2004-09-27T20:00:27 | Mismatch between X-ray and emission-weighted temperatures in galaxy clusters: cosmological implications | [
"astro-ph"
] | The thermal properties of hydrodynamical simulations of galaxy clusters are usually compared to observations by relying on the emission-weighted temperature T_ew, instead of on the spectroscopic X-ray temperature T_spec, which is obtained by actual observational data. In a recent paper Mazzotta et al. show that, if the cluster is thermally complex, T_ew fails at reproducing T_spec, and propose a new formula, the spectroscopic-like temperature, T_sl, which approximates T_spec better than a few per cent. By analyzing a set of hydrodynamical simulations of galaxy clusters, we find that T_sl is lower than T_ew by 20-30 per cent. As a consequence, the normalization of the M-T_sl relation from the simulations is larger than the observed one by about 50 per cent. If masses in simulated clusters are estimated by following the same assumptions of hydrostatic equilibrium and \beta--model gas density profile, as often done for observed clusters, then the M-T relation decreases by about 40 per cent, and significantly reduces its scatter. Based on this result, we conclude that using the observed M-T relation to infer the amplitude of the power spectrum from the X-ray temperature function could bias low \sigma_8 by 10-20 per cent. This may alleviate the tension between the value of \sigma_8 inferred from the cluster number density and those from cosmic microwave background and large scale structure. | astro-ph | astro-ph |
DRAFT VERSION OCTOBER 19, 2018
Preprint typeset using LATEX style emulateapj v. 11/12/01
MISMATCH BETWEEN X-RAY AND EMISSION-WEIGHTED TEMPERATURES IN GALAXY
CLUSTERS: COSMOLOGICAL IMPLICATIONS
E. RASIA1, P. MAZZOTTA 2,3 , S. BORGANI4,5, L. MOSCARDINI6, K. DOLAG1,6, G. TORMEN1, A.
DIAFERIO7, G. MURANTE8
Draft version October 19, 2018
ABSTRACT
The thermal properties of hydrodynamical simulations of galaxy clusters are usually compared to observations
by relying on the emission-weighted temperature Tew, instead of on the spectroscopic X-ray temperature Tspec,
which is obtained by actual observational data. In a recent paper Mazzotta et al. show that, if the cluster is
thermally complex, Tew fails at reproducing Tspec, and propose a new formula, the spectroscopic-like temperature,
Tsl, which approximates Tspec better than a few per cent. By analyzing a set of hydrodynamical simulations
of galaxy clusters, we find that Tsl is lower than Tew by 20 -- 30 per cent. As a consequence, the normalization
of the M -- Tsl relation from the simulations is larger than the observed one by about 50 per cent. If masses in
simulated clusters are estimated by following the same assumptions of hydrostatic equilibrium and β -- model gas
density profile, as often done for observed clusters, then the M -- T relation decreases by about 40 per cent, and
significantly reduces its scatter. Based on this result, we conclude that using the observed M -- T relation to infer
the amplitude of the power spectrum from the X -- ray temperature function could bias low σ8 by 10-20 per cent.
This may alleviate the tension between the value of σ8 inferred from the cluster number density and those from
cosmic microwave background and large scale structure.
Subject headings: galaxies: clusters: general -- X-rays: galaxies masses -- cosmology: observations
1. INTRODUCTION
Thanks to their position at the top of the hierarchy of cosmic
structure formation, galaxy clusters represent powerful tools to
constrain cosmological parameters, like the present value of
the matter density parameter Ω0m, and the normalization of the
primordial power spectrum, usually expressed in terms of σ8,
the r.m.s. matter fluctuation in a sphere of radius of 8h- 1 Mpc
(Rosati et al. 2002; Voit 2004 for recent reviews). The funda-
mental quantity entering theoretical models is the cluster mass,
but from the observational point of view its high-precision mea-
surement is still precluded by the presence of different system-
atic effects affecting the lensing data and the dynamical models.
A possible short cut to overcome this difficulty is represented
by the determination of the cluster temperature from X-ray ob-
servations. In fact simple theoretical arguments, supported by
numerical hydro-N-body simulations, suggest the existence for
virialized gravitational systems of a tight relation between mass
M and temperature T .
The enormous progresses in the angular and spectral reso-
lutions of the latest generation X-ray satellites, Chandra and
XMM-Newton, allowed us to obtain temperature measurements
for extended sets of clusters, showing in many cases the pres-
ence of a rather complex thermal structure. The theoretical
interpretation of these high-quality data requires state-of-the
1 Dipartimento di Astronomia, Università di Padova, vicolo dell'Osservatorio
2, I-35122 Padova, Italy (rasia,kdolag,[email protected])
2 Dipartimento di Fisica, Università di Roma Tor Vergata, via della Ricerca
Scientifica 1, I-00133 Roma, Italy ([email protected])
3 Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge,
MA02138, USA
4 Dipartimento di Astronomia dell'Università di Trieste, via Tiepolo 11, I-
34131 Trieste, Italy ([email protected])
5 INFN -- National Institute for Nuclear Physics, Trieste, Italy
6 Dipartimento di Astronomia, Università di Bologna, via Ranzani 1, I-40127
Bologna, Italy ([email protected])
7 Dipartimento di Fisica Generale "Amedeo Avogadro", Università di Torino,
via Giuria 1, I-10125, Torino, Italy ([email protected])
8 INAF, Osservatorio Astronomico di Pino Torinese, Strada Osservatorio 20,
I-10025 Pino Torinese, Italy ([email protected])
1
art numerical simulations, in which the physical processes act-
ing on the gas component (radiative cooling, non-gravitational
heating, etc.) are faithfully reproduced. The comparison be-
tween observations and simulations is, however, not always
simple. As for X-ray data, the cluster gas temperature is ob-
tained by fitting the source spectrum to a suitable thermal model.
This procedure cannot be reproduced in the analysis of simula-
tions, without using devoted software packages (like X-MAS;
Gardini et al. 2004), which simulate the spectral properties of
the simulated clusters taking into account the instrument re-
sponse, as well as sky and instrumental backgrounds. To over-
come this problem, it has been of common use to resort to the
emission-weighted temperature,
Tew ≡ R Λ(T )n2T dV
R Λ(T )n2dV
.
(1)
Here n is the gas density and Λ(T ) is the cooling function, for
√T , as-
which most of the works in the literature use Λ(T ) ∝
suming bremsstrahlung emission.
Recently, Mazzotta et al. (2004, M04 hereafter) demonstrated
that, if clusters have a complex thermal structure, Tew always
overestimates the spectroscopic temperature Tspec obtained from
X-ray observations. The discrepancy between Tew and Tspec
strongly depends on the complexity of the cluster thermal struc-
ture (see also Mathiesen & Evrard 2001; Gardini et al. 2004).
M04 proposed a new definition of temperature to be imple-
mented in the analysis of simulations, the spectroscopic-like
temperature,
Tsl ≡ R n2T a/T 1/2dV
R n2T a/T 3/2dV
.
(2)
When applied to simulated spectra of clusters hotter than 2-3
keV observed by Chandra or XMM-Newton , this equation,
with a = 0.75, yields a temperature within a few per cent from
Tspec. M04 also showed that this result is almost independent of
the details of the detector response function, as long as its spec-
tral properties are similar to or worse than the Chandra ones.
In this way, Eq. 2 can be reliably applied to the detectors on
board of Beppo-SAX and ASCA.
2
In this Letter we address the implications of using Tsl on the
mass -- temperature relation and on the X -- ray temperature func-
tion (XTF) for an extended set of simulated clusters. These
results are then used to discuss the effect on the estimate of σ8
from the XTF.
2. THE SIMULATED CLUSTERS
The sample of simulated galaxy clusters used in this paper
has been extracted by the large-scale cosmological hydro-N-
body simulation of a "concordance" ΛCDM model (Ω0m = 0.3,
Ω0Λ = 0.7, Ω0b = 0.019 h- 2, h = 0.7, σ8 = 0.8), presented in Bor-
gani et al. (2004, B04 hereafter). Here we give only a short
summary of its characteristics, and we refer to that paper for
more details. The run, made by using the massively parallel
Tree+SPH code P-GADGET2 (Springel et al. 2001; Springel
2004, in preparation), follows the evolution of 4803 dark mat-
ter particles and an equal number of gas particles in a peri-
odic cube of size 192h- 1 Mpc. The mass of the gas particles
is mgas = 6.89 × 108h- 1M⊙, while the the Plummer-equivalent
force softening is 7.5h- 1 kpc at z = 0. Besides gravity and hy-
drodynamics, the simulation includes the treatment of radia-
tive cooling, the effect of a uniform time -- dependent UV back-
ground, a sub -- resolution model for star formation from a mul-
tiphase interstellar medium, as well as galactic winds powered
by SN explosions (Springel & Hernquist 2003). At z = 0 we ex-
tract a set of 95 temperature-selected clusters (hereafter Set#1)
with Tew > 2 keV, value below which the single spectroscopic
temperature for thermally complex systems cannot be defined
(see discussion in M04).
Due to the finite box -- size, the largest cluster found in the
cosmological simulation has Tew ≈ 7 keV. In order to extend our
analysis to more massive and hotter systems, as required for the
study of the M -- T relation, we further include 4 more galaxy
clusters (hereafter Set#2) having Mvir > 1015h- 1M⊙ and be-
longing to a different set of hydro-N-body simulations (Dolag
et al. 2004 in preparation). Since these objects have been
obtained by re-simulating at high resolution a patch of a pre-
existing cosmological simulation, they have a better mass res-
olution, with mgas = 1.69× 108h- 1M⊙. These simulations have
been performed by using the same code with the same choice
of the parameters defining star -- formation and feedback.
Therefore, our total sample (Set#1 and #2) comprises 99 ob-
jects spanning the temperature range 2 ∼< Tew ∼< 13 keV.
3. RESULTS
For each cluster in our sample we calculate both the emission --
weighted and the spectroscopic -- like temperature. As for Tew, it
is computed as in B04 by applying Eq. 1, with the cooling func-
tion estimated in the [0.5 -- 10] keV energy band so as to account
for the typical energy coverage of X -- ray detectors. For simi-
lar reasons, we apply Eq. 2 to compute Tsl by removing the gas
particles with T < 0.5 keV. We note that, since the mean tem-
perature of each cluster here considered is higher than 2 keV,
the excluded particles usually provide a small fraction of the
total emission measure. Therefore, this procedure does not sig-
nificantly affect the values obtained for both Tew and Tsl.
In Fig. 1 we plot the value of Tsl versus Tew, which confirms
the result of M04, that for all clusters Tsl is smaller than Tew.
The distribution of the points in Fig. 1 suggests that the two es-
timators can be roughly related by a linear relation. By applying
a χ2 fit, we find the relation,
Tsl = (0.70± 0.01)Tew+ (0.29± 0.05) ,
(3)
FIG. 1. -- The comparison between emission -- weighted temperature, Tew,
and spectroscopic-like temperature, Tsl. Circles and triangles refer to objects
from Set#1 and Set#2, respectively. The dotted line is the best -- fit linear re-
gression (see Eq. 3), while the dashed line corresponds to Tsl = Tew.
shown as dotted line in Fig. 1. Therefore, for objects with tem-
peratures of about 6 keV, the disagreement can be as large as 2
keV. As also discussed by M04, the relative mismatch between
Tew and Tsl depends only on the thermal structure of the ICM
and, in particular, increases as a function of its complexity. For
this reason, we expect Eq. 3 to hold for the physics considered
in our simulation sets. Any physical process (e.g. stronger feed-
back or thermal conduction), which produces a more isothermal
ICM, is expected to provide a smaller difference between the
two definitions of temperature.
The existence of a systematic trend between the two differ-
ent estimates of the cluster temperature has interesting conse-
quences on the scaling relation between mass and temperature.
Adiabatic hydrodynamical simulations (e.g. Evrard et al. 1996;
Bryan & Norman 1998) find a normalization which is about
40 per cent higher than observed (e.g. Horner et al. 1999;
Nevalainen et al. 2001; Finoguenov et al. 2001; Allen et al.
2001). This result has suggested the necessity of including in
numerical simulations the treatment of extra physical processes
for the gas component, like non -- gravitational heating and ra-
diative cooling. For instance, B04 showed that considering both
effects can be efficient in reducing the normalization of the sim-
ulated M -- T relation, which however turned out to be still 20
per cent higher than observed. In that paper, as in most nu-
merical works on the X-ray properties of galaxy clusters, the
emission -- weighted temperature is used to construct the M -- T
relation. However, since Tew tends to overestimate Tsl by a fair
amount, we expect the actual M-Tsl relation from simulations to
be even more discrepant with respect to observations. This dis-
crepancy is indeed apparent in the left panel of Figure 2, where
the simulated M -- T relation at overdensity ¯ρ/ρcr = 5001 is com-
pared to the observational results by Finoguenov et al. (2001),
who measured temperatures from ASCA data. By modelling
the scaling relation with
M500 = M0(cid:18) T500
keV(cid:19)α
,
(4)
1 Here ρcr is the cosmological critical density.
3
0
mass for a fixed temperature and, therefore, a higher normal-
ization of the power spectrum (e.g., Borgani et al. 2001; Seljak
2002; Pierpaoli et al. 2003; Henry 2004). Huterer & White
(2002) suggested an approximation for the scaling of σ8 with
M0, involving the matter density parameter: Ω0.6
0mσ8 ∝ M0.53
.
Based on this relation, we expect that a bias in the mass esti-
mate as large as that found for the simulated clusters turns into
an underestimate of σ8 by about 20 per cent.
Besides decreasing M0, estimating cluster masses from Eq.5
also significantly reduces the scatter in the M -- T relation from
about 30 per cent to about 16 per cent, as evident from Figure
2. This witnesses that the complex structure of the ICM en-
ters in determining the intrinsic scatter of the M -- T relation in
a more subtle way that just accounted for by cluster-to-cluster
variations of the β -- model fitting parameters and of the effective
polytropic index. The intrinsic scatter of the M -- T relation is a
further piece of information entering into the determination of
σ8: since the theoretical mass function needs to be convolved
with this scatter, the high -- temperature tail of the model XTF
grows with the scatter, at a fixed σ8 (e.g. Pierpaoli et al. 2003).
As a consequence, an overestimate of the scatter turns into an
underestimate of σ8. We find that σ8 decreases by about 5 per
cent by assuming 30 per cent, instead of 16 per cent, intrinsic
scatter. Although this effect only partially compensates for the
underestimate of the amplitude of the M -- T relation, it high-
lights once again how precision measurements of cosmological
parameters require having under control a number of system-
atic effects entering in the conversion between cluster mass and
observable quantities.
As a word of caution, we remind that the mismatch between
Tew and Tsl is introduced by the complexity of the cluster ther-
mal structure and, therefore, so does the estimate of any biasing
in the mass estimate. Furthermore, in order to properly quantify
the amount of a possible bias in the observational determination
of cluster masses it is required to reproduce as closely as pos-
sible the observational procedure and the scale range where the
value of βfit is fitted (see the discussion in B04). We desert
to a forthcoming paper (Rasia et al., in preparation) a detailed
analysis of the observational systematics in the determination of
cluster masses from X -- ray data. In this respect, it will be also
important to compare results on σ8 from the XTF with those
from other methods, also based on galaxy clusters, but not rely-
ing on the choice of the M -- T relation. An example is provided
by the approach based on the cluster baryon mass function (Vo-
evodkin & Vikhlinin 2004), which provides σ8 ≃ 0.7.
In Figure 3 we show the effect that using Tsl, instead of Tew,
has on the estimate of the XTF. We compare here the cumula-
tive temperature function N(> T ) from the simulation (Set#1
only) to the data by Ikebe et al. (2002). Using Tew reproduces
the observed XTF quite well (see also B04): this would sup-
port σ8 = 0.8, the value assumed in the simulation. However,
when Tsl is used to better account for the observational proce-
dure of temperature measurement, the simulation XTF is sig-
nificantly affected, especially in the high-temperature tail. This
result shows that σ8 = 0.8 is excluded at 3σ level, while favor-
ing a larger power spectrum normalization, at least for the ICM
physics introduced in our simulation. For instance, at 2 keV the
simulated cumulative XTF is lower than the observed one by
about 50-60 per cent. The value of Mvir corresponding to this
temperature in our simulation is about 1.5× 1014h- 1M⊙. Com-
puting the mass function by Sheth & Tormen (1999) for our
cosmology at this mass, we find that σ8 needs to be increased
FIG. 2. -- The mass -- temperature relation at ¯ρ/ρcr = 500, in simulations
(filled circles and triangles for Set#1 and Set#2, respectively) and for the ob-
servational data by Finoguenov et al. (2001, squares with errorbars). The left
panel is for the true masses of simulated clusters; the right panel is for masses
of simulated clusters estimated by adopting the same procedure applied by
Finoguenov et al. to observational data (see text).
a log-log least -- square fitting for the simulated clusters gives
α = 1.66± 0.09 and log(M0/h- 1M⊙) = 13.54± 0.03, while
log(M0/h- 1M⊙) = 13.60± 0.07 by fixing the slope to the value
α = 1.5 expected from the condition of hydrostatic equilibrium.
While masses of simulated clusters have been computed here
by summing over all the particles, masses of X -- ray observed
clusters are usually estimated from data (e.g., Markevitch 1998;
Nevalainen et al. 2000; Finoguenov et al. 2001) by applying
the condition of hydrostatic equilibrium to a spherical gas dis-
tribution described by a β -- model (Cavaliere & Fusco -- Femiano
1976), with the equation of state having the polytropic form. In
this way, the total self -- gravitating mass within the radius r is
M(< r) = 1.11× 1014βfitγ
T (r)
keV
r
h- 1Mpc
x2
1 + x2 h- 1M⊙ .
(5)
Here T (r) is the temperature at r, βfit is the fitted slope of the
gas density profile, x = r/rc is the radial coordinate in units of
the core radius rc, and γ is the effective polytropic index.
A number of analyses of hydrodynamical simulations of clus-
ters (e.g., Muanwong et al. 2002; B04; Rasia et al. 2004; Kay
et al. 2004; cf. also Ascasibar et al. 2003) have shown that
Eq.5 underestimates the actual cluster mass by about 20 per
cent. Here we repeat the same analysis as in B04, but using
Tsl in Eq.5 instead of Tew. Although the resulting M -- T rela-
tion (shown in the right panel of Figure 2) is still slightly higher
than the observed one, we find that the amplitude of the M -- T
relation decreases quite substantially, with masses derived from
Eq.5 being on average 40 per cent smaller than the true ones.
The discrepancy between true and recovered cluster masses is
larger than in B04 as a consequence of using Tsl, instead of Tew.
A comparable bias in the mass estimates has been also found
by Bartelmann & Steinmetz (1996), who used a spectroscopic
definition of temperatures. Fitting the simulation results to Eq.4
gives α = 1.53±0.05 and log(M0/h- 1M⊙) = 13.40±0.02, while
log(M0/h- 1M⊙) = 13.41± 0.13 when fixing α = 1.5 (we note
that the change of M0 is larger than the 40 per cent average de-
crease of the temperature, due to the non -- Gaussian distribution
of the scatter in the left panel of Fig.2).
As already mentioned, the M -- T relation is one of the key in-
gredients in the recipes to extract cosmological parameters from
the cluster X -- ray luminosity (XLF) and temperature functions
(XTF): for a fixed value of Ω0m, a larger M0 implies a larger
4
FIG. 3. -- The cumulative temperature function N(> T ), from Set#1. Filled
squares and dashed line refer to the results for the emission-weighted temper-
ature Tew, while filled triangles and solid line represent the spectroscopic-like
temperature Tsl. Error bars correspond to Poissonian uncertainties. For refer-
ence, open circles show the local temperature function measured by Ikebe et
al. (2002) and adapted for the considered cosmological model.
to about 0.9 to increase the number density of clusters above
this mass limit by the required 50 -- 60 per cent.
using Tsl instead of Tew. A comparison with the observed XTF
(Ikebe et al. 2001) indicates that for the "concordance" ΛCDM
model σ8 needs to be increased from ≃ 0.8 to ≃ 0.9.
In general, our conclusions go in the direction of alleviat-
ing a possible tension between the power -- spectrum normaliza-
tion obtained from the number density of galaxy clusters and
that arising from the first -- year WMAP CMB anisotropies (e.g.
Bennett et al. 2003) and SDSS galaxy power spectrum (e.g.
Tegmark et al. 2004). Our results rely on: i) the reliability of
our simulations to correctly describe the thermal complexity of
the ICM; ii) the capability of reproducing and understanding
any possible source of bias in the estimates of cluster masses.
There is little doubt that these two issues need to be addressed
in detail in order to fully exploit the role that hydrodynamical
simulations play in the calibration of galaxy clusters as preci-
sion tools for cosmology.
This work is supported CXC grants GO3-4163X and by Eu-
ropean contract MERG-CT-2004-510143. Simulations were
performed on the CINECA IBM-SP4 supercomputing facility
(INAF -- CINECA grants). This work is also supported by the
the PD51 INFN grant. KD acknowledges support by a Marie
Curie fellowship of the European Community program "Hu-
man Potential" under contract number MCFI-2001-01221. We
thank S. Ettori, V. Springel, L. Tornatore and P. Tozzi for useful
discussions.
4. CONCLUSIONS
REFERENCES
As demonstrated by Mazzotta et al. (2004; see also Math-
iesen & Evrard 2001), the emission-weighted temperature, Tew,
usually assumed to analyze the results of hydrodynamical sim-
ulations, fails in reproducing the "observational" temperature
obtained from the spectral fit. As an alternative, M04 proposed
and tested a new formula, named spectroscopic-like tempera-
ture, Tsl, which gives a much better approximation to the spec-
troscopic temperature, with differences of a few per cent for
T > 2 -- 3 keV.
In this Letter we have studied the consequences, for the mass --
temperature (M -- T ) relation and for the estimate of σ8 from the
X -- ray temperature function (XTF), of adopting Tsl, instead of
Tew. We have analyzed a set of hydrodynamical simulations of
galaxy clusters, which include a realistic treatment of the gas
physics. Our main results can be summarized as follows.
(a) We find in our simulations that Tsl is smaller than Tew by a
factor 0.7 -- 0.8. A linear fit approximating the relation between
the two different temperature estimators is given by Eq. 3.
(b) Using Tsl, instead of Tew, increases the normalization of
the M -- T relation making it higher than the observed one by
about 50 per cent. By assuming hydrostatic equilibrium for a
gas density distribution described by a β -- model with a poly-
tropic equation of state, masses are underestimated on average
by ∼ 40 per cent, therefore substantially reducing the discrep-
ancy with observational data; the scatter in the M -- T relation is
also underestimated by about a factor two.
(c) A bias in the M -- T relation propagates into a bias in the
power spectrum normalization, σ8, from the XTF. If such a bias
is as large as that found in the simulations, the values of σ8
obtained by combining the local XTF and the observed M -- T
relation are underestimated by about 15 per cent.
(d) The XTF from the simulation is significantly lower when
731
Allen, S.W., Schmidt, R.W., & Fabian, A.C. 2001, MNRAS, 328, L37
Ascasibar, Y., Yepes, G., Mueller, V., & Gottloeber, S. 2003, MNRAS, 346,
Bartelmann, M., & Steinmetz, M. 1996, MNRAS, 283, 431
Bennett, C.L., et al. 2003, ApJS, 148, 1
Borgani, S., et al. 2001, ApJ, 561, 13
Borgani, S., et al. 2004, MNRAS, 348, 1078 (B04)
Bryan, G.L., & Norman, M.L. 1998, ApJ, 495, 80
Cavaliere, A., & Fusco -- Femiano, R. 1976, A&A, 49, 137
Evrard, A.E., Metzler, C.R., & Navarro, J.F. 1996, ApJ, 469, 494
Finoguenov, A., Reiprich, T.H., & Böhringer, H. 2001, A&A, 369, 479
Gardini, A., Rasia, E., Mazzotta, P., Tormen, G., De Grandi, S., & Moscardini,
L. 2004, MNRAS, 351, 505
Henry, J.P. 2004, ApJ, 609, 603
Horner, D.J., Mushotzky, R.F., & Scharf, C.A. 1999, ApJ, 520, 78
Huterer, D., & White, M. 2002, ApJ, 578, L95
Ikebe, Y., Reiprich, T.H., Böhringer, H., Tanaka, Y., & Kitayama, T. 2002,
Kay, S.T., Thomas, P.A., Jenkins, A., & Pearce, F.R. 2004, MNRAS, in press
A&A, 383, 773
(astro-ph/0407058)
Markevitch, M. 1998, ApJ, 504, 27
Mathiesen, B.F., & Evrard, A.E. 2001, ApJ, 546, 100
Mazzotta, P., Rasia, E., Moscardini, L., & Tormen, G. 2004, MNRAS, in press
Muanwong, O., Thomas, P.A., Kay, S.T., & Pearce, F.R. 2002, MNRAS, 336,
(astro-ph/0404425) (M04)
527
Nevalainen, J., Kaastra, J., Parmar, A.N., Markevitch, M., Oosterbroek, T.,
Colafrancesco, S., & Mazzotta, P. 2001, A&A, 369, 459
Nevalainen, J., Markevitch, M., & Forman, W. 2000, ApJ, 532, 694
Pierpaoli, E., Borgani, S., Scott, D., & White, M. 2003, MNRAS, 342, 163
Rasia, E., Tormen, G., & Moscardini, L. 2004, MNRAS, 351, 237
Rosati, P., Borgani, S., & Norman, C. 2002, ARA&A, 40, 539
Seljak, U. 2002, MNRAS, 337, 769
Sheth, R.K., & Tormen, G. 1999, MNRAS, 308, 119
Springel, V., & Hernquist, L. 2003, MNRAS, 339, 289
Springel, V., Yoshida, N., & White, S.D.M. 2001, NewA, 6, 79
Tegmark, M., et al. 2004, Phys. Rev. D, 69, 103501
Voevodkin, A. & Vikhlinin, A. 2004, ApJ, 601, 610
Voit, G.M. 2004, Rev. Mod. Phys., in press
|
astro-ph/0602215 | 1 | 0602 | 2006-02-09T21:20:56 | On a Site of X-ray Emission in AE Aquarii | [
"astro-ph"
] | An analysis of recently reported results of XMM-Newton observations of AE Aqr within a hypothesis that the detected X-ray source is located inside the Roche lobe of the white dwarf is presented. I show this hypothesis to be inconsistent with the currently adopted model of mass-transfer in the system. Possible solutions of this problem are briefly discussed. | astro-ph | astro-ph | Draft version November 1, 2018
Preprint typeset using LATEX style emulateapj v. 11/12/01
ON A SITE OF X-RAY EMISSION IN AE AQUARII
Institute of Astronomy, University of Cambridge, Madingley Road, CB3 0HA, UK
Nazar R. Ikhsanov1
Draft version November 1, 2018
ABSTRACT
An analysis of recently reported results of XMM-Newton observations of AE Aqr within a hypothesis
that the detected X-ray source is located inside the Roche lobe of the white dwarf is presented. I show
this hypothesis to be inconsistent with the currently adopted model of mass-transfer in the system.
Possible solutions of this problem are briefly discussed.
Subject headings: accretion, accretion disks -- X-rays: binaries -- (stars:) white dwarfs -- stars:
6
0
0
2
b
e
F
9
1
v
5
1
2
2
0
6
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
individual (AE Aqr)
1.
introduction
AE Aqr is a close binary system with an orbital period
◦
◦
9.88 h, mass ratio 0.58 -- 0.89, and an inclination i ≃ 55
±7
(Welsh, Horne, & Gomer 1995). The distance to the
system is about 100 pc (Friedjung 1997). The degener-
ate companion is a magnetized white dwarf rotating with
a period of Ps = 33 P33 s and braking with a rate of
P0 = 5.64×10−14 s s−1, which implies the spin-down power
of (de Jager et al. 1994; Welsh 1999),
Lsd
∼= 6 × 1033 P −3
(1)
Here I50 is the moment of inertia of the white dwarf ex-
pressed in units of 1050 g cm2.
33 I50 ( P / P0) erg s−1,
The normal companion is a K3 -- K5 red dwarf which
overflows its Roche lobe and loses material through the
L1 point towards the white dwarf. This material man-
ifests itself in a form of the optical/UV continuum and
emission lines. It is neither accreted onto the surface of
the white dwarf nor stored in a disk around its magneto-
sphere. Instead, it interacts with the magnetic field of the
white dwarf via a drag term (King 1993) and is leaving
the system without forming a disk (Wynn, King, & Horne
1997; Welsh, Horne, & Gomer 1998; Ikhsanov, Neustroev,
& Beskrovnaya 2004).
the white dwarf. According to their scenario the heat-
ing occurs at an adiabatic shock located at a distance of
∼ 1010 cm from the white dwarf and the observed emission
is powered by the gravitational energy of the streaming
material.
In this letter I show that the above mentioned hypothe-
sis is inconsistent with the currently adopted model of the
mass transfer in the system and thus, a question about a
location of the source of X-ray emission in AE Aqr remains
open.
2. can the x-ray source be associated with an
expanded stream ?
Within the hypothesis suggested by Itoh et al. (2005)
the flow of hot material responsible for the observed X-
rays is feeded by the stream flowing into the Roche lobe
of the white dwarf through the L1 point. The numerical
simulations of Hα Doppler tomogram of AE Aqr (Wynn
et al. 1997; Welsh et al. 1998; Ikhsanov et al. 2004) have
shown that a distance to which the stream could approach
the white dwarf, rmin, significantly exceeds its corotation
radius rcor ≃ 1.5 × 109 M 1/3
cm. This distance, in
the general case, is limited to rmin & max[r0, rA], where
rA is the Alfv´en radius of the white dwarf, and
0.8 P 2/3
33
a
0.64(cid:17)−0.464(cid:18)
r0 ≃ 1010(cid:16) q
1.8 × 1011 cm(cid:19) cm (2)
The system X-ray emission has recently been stud-
ied with XMM-Newton (RGS, EPIC) by Itoh, Ishida,
& Kunieda (2005). As they have reported, the num-
ber density of plasma responsible for the detected X-rays
is nx ∼ 1011 cm−3 and the linear scale of the source is
ℓx ∼ (2 − 3) × 1010 cm. The observed spectrum has
been well fitted using the 4-temperature (0.14, 0.59, 1.4
and 4.6 keV) VMEKAL model (Mewe, Kaastra & Liedahl
1995). The analysis of the centroids of N and O Lyα lines
shows no evidence for any significant orbital Doppler mod-
ulation. The widths of these lines are close to 1000 km s−1.
Analyzing these results Itoh et al. (2005) have discarded
a possibility that the detected X-rays are emitted from the
surface of the white dwarf indicating that nx is a few orders
of magnitude smaller than corresponding conventional es-
timates in the post-shock accretion column and ℓx exceeds
the radius of the white dwarf by almost two orders of mag-
nitude. Instead, they have suggested a hypothesis in which
the detected emission is associated with a heating and ex-
pansion of the material streaming through Roche lobe of
1 Central Astronomical Observatory, Russian Academy of Sciences, 65-1 Pulkovo, 196140 St. Petersburg, Russia
The radial distance r0 represents a minimum possible
distance to which the material flowing through the L1
point can approach the white dwarf. The value of this
parameter does not depend on either the magnetic field
strength of the white dwarf or the structure of the inflow-
ing material and is only based on the angular momen-
tum conservation law. For the material to come closer to
the white dwarf its angular momentum must be reduced.
However a question about the mechanism which could be
responsible for such a reduction in the case of AE Aqr re-
mains open. Indeed, due to a lack of a disk the canonical
is a distance to which the material could approach the
white dwarf if its angular momentum along the whole tra-
jectory remains constant (see e.g. Eq. 2.14 in Warner
1995). Here q and a are the mass ratio and orbital sepa-
ration of the system components.
1
2
viscosity model of the angular momentum transport is not
applicable. On the other hand, the interaction between
the stream and the magnetic field at a distance r0 tends to
increase the angular momentum of the material since the
velocity of field lines, Ωr, significantly exceeds the velocity
of the material, which is limited to the free-fall velocity,
Vff = p2GMwd/r.
A quantitative analysis of the mass-transfer process in
AE Aqr has been first presented by Wynn et al. (1997). As
they have shown, the Hα Doppler tomogram of the system
can be reproduced assuming that the stream is inhomoge-
neous (a sequence of large diamagnetic blobs) and interacts
with the magnetic field of the white dwarf via a drag term.
The efficiency of this interaction is ∝ r−n, where n ≥ 2.
That is why, the strongest interaction between the blobs
and the magnetic field occurs at their closest approach to
the white dwarf. The initial radius of the blobs at rmin is
lb ≃ 109 Q1/2
19 cm,
(3)
where Q19 is the effective cross-section of the mass transfer
throat at the L1 point expressed in units of 1019 cm2. The
initial sound speed and number density of the material
entrained in the blobs are, respectively
cs ≃ 106 T 1/2
4
cm s−1,
(4)
and
nb ≃ 1014 M17 M −1/2
(5)
Here l9 = lb/109 cm, r10 = rmin/1010 cm, and T4 is the
mean temperature of the flowing material expressed in
units of 104 K (see Beskrovnaya et al. 1996; Eracleous &
Horne 1996).
r1/2
10 l−2
9
cm−3.
0.8
Let us now consider a scenario in which the hot flow
forms at rmin as a result of strong interaction between the
blobs and the magnetic field (independently of the mech-
anism of this interaction). This implies a heating of the
blobs (or their surface layers) and their expansion up to
a size of ℓx. As the gas expands its density decreases by
a factor of ℓ3
b∆s), where ∆s is the thickness of the
heated layer. This indicates that the initial number den-
sity of material in the layer should satisfy the following
condition
x/(l2
n0 & nxℓx/(l2
b∆s).
(6)
Otherwise, the density of the hot flow would be signifi-
cantly smaller than that evaluated by Itoh et al. (2005).
Combining Eqs. (5) and (6), and setting n0 = nb one finds
∆s & 8 × 108 M−1
17 M 1/2
0.8 r−1/2
nx
ℓ3
10
10(cid:16)
1011 cm−3(cid:17) cm. (7)
Thus, for the considered scenario to realize the entire blobs
at a distance rmin should be heated to a temperature of
5 keV.
However, as soon as the temperature of the blob reaches
5 keV, its X-ray luminosity proves to be
14 l3
Lx,sl ≃ 6 × 1032 T 1/2
7.7 n2
9 erg s−1,
(8)
i.e. a factor of 60 larger that the X-ray luminosity of
the system (≃ 1031 erg s−1, see e.g. Clayton & Osborne
1995; Choi, Dotani, & Agrawal 1999). Here n14 =
nb/1014 cm−3.
Furthermore, it remains unclear how the temperature
of the blob could reach the value of 5 keV. Indeed, for the
heating of the blobs to occur the characteristic time of the
heating process should be smaller than the cooling time.
But the bremsstrahlung cooling time of the blobs,
0.8
tbr ≃ 15 T 1/2
7.7 n−1
14 s,
10 M −1/2
(9)
is significantly smaller than the free-fall time at rmin
(tff ≃ 70 r3/2
s) and correspondingly, the charac-
teristic time of the drag interaction, which according to
Wynn et al. (1997) is limited to tdrag > tff . The value of
tbr is even smaller than the turbulent diffusion time eval-
uated by Meintjes & Venter (2005). Thus, the assumption
about a significant heating of the entire blobs at the ra-
dius rmin within any currently considered mechanisms of
interaction between the blobs and the magnetic field of
the white dwarf is not valid. This means, that formation
of the hot flow with the number density ∼ 1011 cm−3, lin-
ear scale & 1010 cm, and temperature ∼ 5 keV inside the
Roche lobe of the white dwarf is impossible within the
currently adopted model of the mass-transfer.
Instead,
the source associated with the hot surface layers of the
blobs within this model is expected to be dense, ∼ nb,
and compact, ∆s . 3 × 104 n−2
cm.
14 T −1/2
7.7
3. discussion
The problem posed in the previous section suggests that
either our current view on the mass-transfer picture in
AE Aqr is incomplete or a significant part of detected X-
rays is generated in a region situated beyond the Roche
lobes of the system components, or both. Here I briefly
address these two possibilities.
The situation can be partly improved if one invokes an
assumption that the mass transfer through the Roche lobe
of the white dwarf operates via more than one component.
In particular, a formation of the X-ray source with the
required parameters inside the Roche lobe of the white
dwarf could be expected if the blobs were surrounded by a
medium of a linear size ∼ 1010 cm and mean number den-
sity ∼ 1011 cm−3. The required rate of mass transfer by
this additional flow is ∼ 5×1015 g s−1, i.e. significantly (by
more than an order of magnitude) smaller than the rate of
mass transferred by the blobs. The heating of the flow, ac-
cording to Eq. (9), can be expected even if its interaction
with the magnetic field is gentle. In particular, the cooling
time of this flow would be larger than the time of heating
governed by the drag interaction (see Eq. 1 in Wynn et al.
1997). The temperature of the flow, in the first approxi-
mation, can be limited by equating the thermal pressure
of the material, (1/2)nxmpc2
s , with the energy density of
the external magnetic field, B2/8π:
nx
K.
(10)
T . 5.8 × 107 B2
2 (cid:16)
1011 cm−3(cid:17)−1
Here B2 is the strength of the external magnetic field ex-
pressed in units of 102 G.
It should be noted that both the density and size of the
additional flow component significantly exceed the corre-
sponding parameters of the interblob material considered
by Wynn et al. (1997). The hot flow under these conditions
screens the blobs from the magnetospheric field making
the drag interaction between them ineffective. However,
the suppression of the drag-driven propeller action does
not occur at the closest approach of the flow to the white
dwarf since the blobs and the hot flow at this distance
follow different trajectories. Indeed, being more dense the
blobs would significantly deeper penetrate into the magne-
tosphere of the white dwarf (for a discussion see e.g. Welsh
et al. 1998; Ikhsanov et al. 2004). This allows an ample
room for a direct interaction between the blobs and the
magnetic field of the white dwarf at the distance r0, i.e.
just at the point where the efficiency of the drag interac-
tion reaches its maximum value.
At the same time, the origin of this additional compo-
nent within the currently adopted model of the system re-
mains uncertain. An incorporation of the additional com-
ponent into the mass transfer scheme implies that either
its temperature at the L1 point exceeds 2 × 105 K (in this
case a formation of the geometrically thick flow at rmin
can be treated in terms of its thermal expansion), or the
effective cross-section of the mass-transfer throat at the
L1 point is significantly larger than its canonical value. A
reason why these possibilities cannot be simply discarded
is that the normal component of AE Aqr is partly (by al-
most a half of its radius) situated inside the light cylinder
of the white dwarf. This situation is unique for all cur-
rently known close binaries and the mass-transfer process
which could be realized under these conditions is poorly
understood so far.
Another problem, which has to be addressed within the
above presented scenario, is a lack of any significant or-
bital Doppler modulation of the N and O Lyα lines (Itoh
et al. 2005). The flow effectively contributes to the X-ray
flux of the system only on a time scale of its expansion,
s,
7.7
texp ≃ 160 ℓ10 T −1/2
(11)
and therefore, would appear as a local source.
Indeed,
the temperature and density of an adiabatically expand-
ing fully ionized gas decrease as T ∝ l−2, and n ∝ l−3 and
the luminosity of the flow decreases on a time scale of texp
by more than an order of magnitude. This argues strongly
against an assumption that the hot gas fills the Roche lobe
of the white dwarf and spreads around the system (as it
has been suggested by Itoh et al. 2005). Under these con-
ditions, however, a modulation of the flow parameters with
the system orbital motion would be expected and the lack
of success in searching for this modulation can hardly be
interpreted in a simple way.
A modification of the currently adopted mass transfer
model is not required if the source of X-ray emission has a
multi-component nature. In particular, one can envisage a
situation in which a part of the detected X-rays are gener-
ated outside the system. The only available energy source
in this case is the spin-down power of the white dwarf (the
accretion power in this case can obviously be excluded).
Can this energy be converted into a 5 keV emission in a
region situated outside the system ?
The answer to this question depends on the ratio of the
spin-down power channeled into the mass outflow and par-
ticle acceleration respectively, which is a matter of serious
debate at present. It is rather negative if the spin-down
power is converted mainly into the kinetic energy of the
outflowing material. Indeed, the velocity dispersion of the
blobs leaving the system is limited to ∆V . 300 km s−1
(Wynn et al. 1997; Welsh et al. 1998; Ikhsanov et al. 2004).
This is a factor of 4 smaller than both the thermal veloc-
ity of a plasma heated up to the temperature 5 keV and
the velocity evaluated from the analysis of the width of
emission lines (see Introduction).
3
The answer might be, however, positive if the spin-
down power were released in a form of accelerated parti-
cles or/and MHD waves. As recently shown by Antonicci
& G´omez de Castro (2005), the temperature of X-ray pho-
tons emitted by dense (ne ∼ 1011 cm−3) blobs illuminated
by a shower of high-energy electrons (0.03-1 MeV) ranges
within an interval 1 -- 20 keV. The spectrum of this radia-
tion depends on the angle between the line of sight and the
velocity vector of electrons. In particular, the spectrum of
blobs situated between an observer and the source of accel-
erated particles differs from the spectrum emitted by blobs
situated behind the particle source. Therefore, an acceler-
ator of particles surrounded by the blobs would appear as
a multi-temperature source of 1 -- 20 keV emission.
Let us check whether such a situation can be realized in
AE Aqr. The radius of dense blobs ejected from the sys-
tem by the propeller action of the white dwarf increases
due to their thermal expansion and reaches the value of ℓx
on a time scale
τ =
ℓx
cs
≃ 104 ℓ10 c−1
6
s.
(12)
Here the speed of sound, c6 = cs/106 cm s−1, is normalized
by taking into account that the heating of the dense blobs
inside the Roche lobe of the white dwarf is insignificant
(see Sect. 2). On this time scale the blobs moving with an
average velocity Vout leave the system to a distance of
6 (cid:18)
r0 ∼ Voutτ ≃ 3 × 1011 ℓ10 c−1
Vout
300 km s−1(cid:19) cm (13)
and their number density decreases to
ℓx(cid:19)3
n0 = nb(cid:18) lb
≃ 1011 l3
10 (cid:16)
9 ℓ−3
nb
1014 cm−3(cid:17) cm−3
(14)
The number of blobs ejected by the white dwarf during
the time τ is
N ≃
3 M τ
4πnbmpl3
b
.
Combining Eqs. (5), (12), and (15) yields
r−1/2
10
N ≃ 103 M 1/2
0.8 c−1
6
ℓ10 l−3
9
(15)
.
(16)
Therefore, the stream of the expanded blobs will occupy
an area of
A ∼ πℓ2
xN ≃ 3 × 1023 (cid:18) ℓx
1010 cm(cid:19)2(cid:18) N
103(cid:19) cm2,
(17)
which constitutes ∼ 0.3 of the total area of a sphere with
a radius of r0. This indicates that 30% of the energy re-
leased in the system will be transferred through the stream
of the expanded blobs.
If now we assume that the spin-down power releases
predominantly in a form of accelerated particles (see e.g
Ikhsanov 1998; Meintjes & de Jager 2000) one finds the
total flux of the energy transferred through the area occu-
pied by the stream of the expanded blobs as (see Eqs. 5,
and 12 -- 17)
L1 = Lsd
A
4πr2
0
≃ 1033 erg s−1 ℓ10M 1/2
0.8 c6 l−1
9
r−1/2
10 ×
(18)
(cid:18)
Vout
300 km s−1(cid:19)−2(cid:18)
Lsd
6 × 1033 erg s−1(cid:19) .
4
Hence, a conversion of only 2% of this energy into X-rays
would be enough for the luminosity of the corresponding
source to be comparable to the X-ray luminosity observed
from the system.
The above estimates suggest that an interpretation of
the X-ray emission of AE Aqr in terms of expanded blobs
illuminated by a flux of accelerated particles might be
promising. A further analysis of this scenario, however,
requires information about the spectrum and geometry of
the wind of particles ejected by the white dwarf, which
is currently not available. Furthermore, a question about
the region of the wind formation also remains open.
If
this region (as in the case of spin-powered pulsars, see e.g.
Beskin, Gurevich & Istomin 1993) is situated at the light
cylinder, the surface of the normal companion would also
be partly affected by the wind. However, a contribution
of this additional source into the X-ray flux can unlikely
be significant since the area occupied by the normal com-
panion at the radius of the light cylinder is a factor of 20
smaller than A (see Eq. 17).
I thank James Pringle, Andrew Fabian and Christopher
Mauche for useful discussions and an anonymous referee
for suggesting improvements. I acknowledges the support
of the European Commission under the Marie Curie In-
coming Fellowship Program. The work was partly sup-
ported by Russian Foundation of Basic Research.
REFERENCES
Antonicci, A., G´omez de Castro, A.I. 2005, A&A, 432, 443
Beskin, V.S., Gurevich, A.F., Istomin, Ya.N. 1993, Physics of Pulsar
Magnitosphere, Cambridge Univ. Press, Cambridge
Beskrovnaya, N.G., Ikhsanov, N.R., Bruch, A., Shakhovskoy, N.M.
1996, A&A, 307, 840
Choi, C-S., Dotani, T., Agrawal, P.C. 1999, ApJ, 525, 399
Clayton, K.L., Osborne, J.P. 1995,
in Magnetic Cataclysmic
Variables, Eds. D. Buckley and B. Warner, ASP Conference Series
85, 379
de Jager, O.C., Meintjes, P.J., O'Donoghue, D., Robinson, E.L. 1994,
MNRAS, 267, 577
Eracleous, M., Horne, K. 1996, ApJ, 471, 427
Friedjung, M. 1997, New Astron., 2, 319
Ikhsanov, N.R. 1998, A&A, 338, 521
Ikhsanov, N.R., Neustroev, V.V., Beskrovnaya, N.G. 2004, A&A,
421, 1131
Itoh, K., Ishida, M., Kunieda, H. 2005, ApJ, accepted (astro-
ph/0412559)
King, A.R. 1993, MNRAS, 261, 144
Meintjes, P.J., de Jager, O.C. 2000, MNRAS, 311, 611
Meintjes, P.J., Venter, L.A. 2005, MNRAS, 360, 573
Mewe, R., Kaastra, J.S., Liedahl, D.A. 1995, Legacy, 6, 16
Warner, B. 1995, "Cataclysmic Variable Stars", Cambridge
University Press.
Welsh, W.F. 1999, "Annapolis Workshop on Magnetic Cataclysmic
Variables", Eds. C. Hellier and K. Mukai, ASP Conference Series
157, 357
Welsh, W.F., Horne, K., Gomer, R. 1995, MNRAS, 275, 649
Welsh, W.F., Horne, K., Gomer, R. 1998, MNRAS, 298, 285
Wynn, G.A., King, A.R., Horne, K. 1997, MNRAS, 286, 436
|
astro-ph/9806267 | 1 | 9806 | 1998-06-19T12:57:38 | A low-mass HI companion of NGC 1569? | [
"astro-ph"
] | High-sensitivity maps of the large-scale structure of atomic hydrogen in the starburst dwarf galaxy NGC 1569 show evidence for an HI cloud with a mass of 7*10**6 M_sun, at a projected distance of 5 kpc from the parent galaxy. This cloud may be a condensation in a low-column-density HI halo or a companion galaxy/HI-cloud. NGC 1569 and its companion are connected by a low surface brightness HI bridge. At the edge of NGC1569, the HI bridge coincides with H_alpha arcs, also detected in soft X-rays. | astro-ph | astro-ph | A&A manuscript no.
(will be inserted by hand later)
Your thesaurus codes are:
3 (09.01.2; 11.09.1 NGC 1569; 11.09.2; 11.09.5; 11.11.1; 11.19.3)
ASTRONOMY
ASTROPHYSICS
AND
8
9
9
1
n
u
J
9
1
1
v
7
6
2
6
0
8
9
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
A low-mass HI companion of NGC1569?
J.M. Stil1 and F.P. Israel1
Leiden Observatory, PO Box 9513, NL 2300 RA Leiden, The Netherlands
Received 4 March 1998; accepted 3 June 1998
Abstract. High-sensitivity maps of the large-scale structure of
atomic hydrogen in the starburst dwarf galaxy NGC 1569 show
evidence for an HI cloud with a mass of 7 · 106M⊙, at a pro-
jected distance of 5 kpc from the parent galaxy. This cloud
may be a condensation in a low-column-density HI halo or a
companion galaxy/HI-cloud. NGC 1569 and its companion are
connected by a low surface brightness HI bridge. At the edge of
NGC1569, the HI bridge coincides with Hα arcs, also detected
in soft X-rays.
Key words: ISM: atoms - Galaxies: individual: NGC 1569 -
interactions - irregular - kinematics and dynamics - starburst
The overall HI distribution consists of a high-column-den-
sity ridge with three peaks and an 'arm' extending to the west
(Reakes 1980, Israel & Van Driel 1990). Channel maps pre-
sented by Reakes (1980) also show weak emission south-east
of the bright HI ridge, likewise visible in the low resolution
map of Israel & Van Driel (1990). In this Letter, we concen-
trate on the nature of this emission and compare it with the ex-
tended HI haloes observed in the dwarf galaxies IC 10 (Shostak
& Skillman 1989) and NGC4449 (Bajaja et al. 1994). A more
complete analysis of the HI emission of NGC1569 is deferred
to a forthcoming paper.
1. Introduction
NGC 1569 is a small Im type galaxy at a distance of 2.2 ±
0.6 Mpc (Israel 1988), and a probable member of the IC 342
group. Due to its low galactic latitude, the members of this
group are heavily obscured. The number of known members
is growing rapidly with the discovery of Dwingeloo 1 (Kraan-
Korteweg et al. 1994), Dwingeloo 2 (Burton et al. 1996), MB 1
and 2 (McCall & Buta 1995) and MB 3 (McCall & Buta 1997).
With a diameter of 1.85 kpc, NGC 1569 is dominated by the
aftermath of a burst of star formation (e.g. Israel & de Bruyn
1988, Vallenari & Bomans 1996). It suffers an outflow of gas
(Waller 1991, Heckman et al. 1995 and references therein) and
contains three very luminous, compact star clusters and several
bright HII regions (Waller 1991).
The nearest neighbour to NGC 1569 is the dwarf irregular
galaxy UGCA 92 at a projected distance of 60 kpc (Karachent-
sev & Tikhonov 1993), at approximately the same radial veloc-
ity (vhel = −99 kms−1), and with an HI mass of 7 · 107M⊙
(de Vaucouleurs et al. 1991). Emissionline-dominated dwarf
galaxies such as NGC 1569 often, if not always, have nearby
companions (Taylor et al. 1995, but see Taylor et al. 1996).
The pair NGC 1569/UGCA 92 fits the selection criteria and
overall properties given by Taylor et al. (1995), but the orbital
timescale of this pair seems far too long to explain the recent
starburst in NGC 1569 in terms of interaction with UGCA 92.
Send offprint requests to: J.M. Stil
Correspondence to: [email protected]
2. Observations and data reduction.
HI in NGC 1569 was observed with the WSRT during four 12-
hour runs between November 1989 and January 1990. A total
of 160 baselines ranging from 36 m to 2772 m were sampled
at increments of 18 m. Because of missing short interferom-
eter spacings, structures larger than a few arcminutes become
increasingly diluted. However, the HI emission in individual
velocity channels generally does not extend that far, so that the
importance of this effect is minor.
The 2.5 MHz band was sampled by 128 independent fre-
quency channels of 4.12 km s−1 width. The Hanning-smoothed
velocity resolution is 8.2 km s−1. Maps at 27′′ and 60′′ resolu-
tion were produced by applying Gaussian weight functions of
843 m and 390 m width in the UV plane. The continuum was
subtracted in each pixel by fitting a first-order polynomial to
48 line-free channels at both edges of the frequency band. In
each channel map, the area containing line emission was iden-
tified manually and CLEANed down to half the r.m.s. noise.
The rms noise in each 27′′ channel map is 1.1 mJy per beam.
The channel maps have not been corrected for primary-beam
attenuation.
3. Results and analysis.
In the velocity range −160 km s−1 to −120 km s−1, the chan-
nel maps show emission from NGC 1569 itself, and also from
a second feature at α = 4h27m20s, δ = 64o43′.0, best iden-
tified in the vhel = −152 and vhel = −135km s−1 maps in
Fig. 1. Although the emission is weak (4 σ), we consider it
2
J.M. Stil & F.P. Israel: A low-mass HI companion of NGC1569?
Fig. 1. Velocity maps at 27′′ resolution, summed over four channels each. Heliocentric velocities marked are those of the third
channel in each set. Contours are at -10.5, 10.5 (3σ), 17.5, 24.5, 31.5, 38.5, 77.0 and 154.0 mJy per beam. The last panel also
contains Galactic foreground emission present in a velocity range of about 10 kms−1.
real. First, its location coincides with emission in Reakes's in-
dependent maps. Second, we eliminated the possibility of solar
interference by excluding all hour angles with the sun above
the horizon. We also extracted the line profile in the direction
of the feature from the raw data, i.e. before continuum sub-
traction and cleaning. In both cases, the emission persisted. We
will refer to this feature as NGC 1569-HI. There is further weak
emission between NGC 1569 and NGC 1569-HI in the velocity
range −130 km s−1 to −90 km s−1 (also visible in Reakes's
−100 km s−1 channel map.
A second filamentary structure is visible in the channel
maps with velocities -86 and -69 km s−1 near α = 4h25m30s,
δ = 64o44′. It appears as a westward extension of the bright-
est emission. It is considerably closer to NGC 1569 than
NGC 1569-HI and always connected to the brightest emission.
For this reason, we treat it as a part of NGC 1569 and refer to it
as the 'western arm'. This feature was first noticed by Reakes
(1980).
A map of the total HI column density is shown in Fig. 2,
together with line profiles summed over selected regions. The
emission of NGC 1569 is dominated by the high-column-den-
sity ridge and the western arm. NGC 1569-HI is 8′ east of
the centre of NGC 1569, and has a FWHM velocity width of
50 km s−1. The spatially-integrated emission of NGC 1569-
HI and of the bridge over the velocity range −190 km s−1 to
−60 km s−1, is 12.4 Jy km s−1, corrected for 16% primary-
beam attenuation. The line flux of NGC 1569-HI itself is 6.3
Jy km s−1, half of the total. If they are at a distance of 2.2 Mpc,
the masses of NGC 1569-HI and of the bridge are 7 · 106 M⊙
each.
On the side of NGC 1569, the apparent connection with
the bridge is located on the southern side of the galaxy. This
J.M. Stil & F.P. Israel: A low-mass HI companion of NGC1569?
3
100
0
-100
40
20
0
-20
-40
-60
-300
-200
-100
0
100
-300
-200
-100
0
100
50
0
-50
-100
40
20
0
-20
-40
-300
-200
-100
0
100
-300
-200
-100
0
100
Fig. 2. HI column density map of NGC1569. Contours are at -1 (dashed), 1, 2, 4, 6, 11, 16, 21, ... ·1020 HI cm−2. The luminous
star clusters A and B in the centre of NGC1569 are marked by dots. Thick lines mark the brightest Hα emission, Hα arcs and
the northern limit of filaments in a deep image by Hunter et al. (1993). The dashed curve shows the path of the position-velocity
diagram of Fig. 3. The synthesized beam is shown as a circle in the lower right corner. b). Line profiles of the areas labeled 1
through 4 in a, defined by the 2 · 1020 cm−2 contour. The line profile centroid shifts from -130 km s−1 to -80 km s−1 along the
HI bridge. Much narrower negative features are sidelobes of Milky Way foreground emission (see Fig. 1 and Fig. 3).
of NGC 1569-HI. The detection limit on the blue plates for a
smooth disk of 1′ diameter should be about 26 mag arcsec−2;
foreground stellar confusion may raise this somewhat. With 2.3
magnitudes foreground extinction (Israel 1988), we arrive at a
limit of about 23.5 mag arcsec−2. As this is close to the median
surface brightness of DDO dwarf galaxies (de Vaucouleurs et
al. 1981), the optical non-detection of NGC 1569-HI therefore
allows it to be a low-surface-brightness dwarf galaxy with a
foreground extinction similar to that of NGC 1569, an inter-
galactic HI cloud without any stars, or a Galactic high-velocity
cloud.
The central velocity of NGC 1569-HI is −130 ± 5 km s−1,
which is in the extreme blue wing of the line-of-sight Galactic
HI line profile (Hartmann & Burton 1997). The Galactic fore-
ground is clearly present at v > −70 km s−1 throughout the
field of view, but with a velocity width less than 10 km s−1.
However, we find no emission other than that of NGC 1569
and NGC 1569-HI at any velocity vhel < −120 km s−1, even
at 60′′ resolution. We have summed the spectra of background
point sources to look for absorption (Fig. 4). The 3 σ upper
limit to the optical depth per channel is 0.12 in the veloc-
ity range −300 km s−1 to −30 km s−1. The corresponding
upper limit to the HI column-density per channel of an ex-
tended component with spin temperature Ts, is NHI < 9.6 ·
1019(Ts/100K).
4. Discussion.
We first consider whether a Galactic origin is consistent with
both the large velocity range of the eastern emission and its
small angular size. The high-velocity-cloud complex closest to
Fig. 3. Position-velocity map along the HI bridge, the ridge,
and the arm extracted from the 60′′ resolution data cube.
Specific positions mentioned in the text have been idicated.
Contour
-10,-5,5,10,15,20,25,50,100,150,200,250
mJy/beam per 4.12 km s−1 channel. Negative contours are
drawn as dashed lines.
levels
:
region coincides with the location of two concentric Hα shells
described by Waller (1991).
Figure 3 shows a position velocity map along the curved
dashed line in Fig. 2. Going westwards from NGC 1569-HI,
'bridge' emission occurs at progressively less negative veloci-
ties. The rotation of NGC 1569 occurs in the opposite sense.
We have scanned the red and blue POSS II plates on the
Leiden ASTROSCAN, but failed to find an optical counterpart
4
J.M. Stil & F.P. Israel: A low-mass HI companion of NGC1569?
foreground emission inhibit the detection of an extended, low-
surface-brightness component (see e.g. Fig. 3). NGC 1569-HI
cannot be identified with the symmetric counterpart of the ex-
tended emission in the northwest, because its velocity is 80
km s−1 lower. An extended HI halo is sometimes inferred from
the difference in flux measured by an interferometer and a sin-
gle dish radio telescope. However, single dish observations are
heavily contaminated by the large column of Galactic HI. Fur-
thermore, there is no evidence for an HI halo in the maps pre-
sented by Reakes (1980).
The apparent retrograde rotation of NGC 1569-HI does not
exclude a prograde orbit, e.g. as a part of an extended HI disk.
High resolution observations of the haloes of IC 10 (Shostak &
Skillman 1989) and NGC 4449 (Bajaja et al. 1994) show a re-
versal of the velocity field. Such a reversal can be the result of a
warped HI disk bending through the plane of the sky. However,
the warp angle must then be larger than the inclination angle,
which is 60o for NGC 1569 (Israel 1988). There is no indica-
tion from the HI distribution that such an extreme warp exists
in NGC 1569.
As it is unlikely that NGC 1569-HI is an extension of the
rotating disk of NGC 1569, it is either primordial gas which
has not yet fallen into NGC 1569, or a separate object in a
close encounter with NGC 1569. The difference between these
interpretations is mainly semantic. No definite choice can be
made since there is no optical identification of NGC 1569-HI.
However, some general remarks can be made.
If NGC 1569-HI is a part of an HI halo which settles into
the inner disk of NGC 1569 on a cosmic timescale, its small
distance to NGC 1569 suggests that its is slowly spiraling in-
wards. This implies a relatively circular orbit, as opposed to
a parabolic or hyperbolic orbit expected for an encounter. As-
suming a circular orbit, a lower limit to the mass of NGC 1569
can be derived. As the projected distance, d, and radial veloc-
ity difference ∆v are lower limits to the actual distance and
relative velocity, the minimum mass of the system if it is to
be bound, must be Mmin ≈ G−1d∆v2 = 2 · 109M⊙. This
lower limit, which is absolute for a circular orbit, is an or-
der of magnitude higher than the estimated dynamical mass of
NGC 1569 within a radius of 1.3 kpc, where the rotation ve-
locity is maximal (Reakes 1980). The lower limit to the total
mass results in a strong upper limit to the HI mass to total mass
ratio MHI /Mtotal < 0.04 within 5 kpc, as opposed to the ob-
served MHI /Mtotal = 0.33 within a radius of 1.3 kpc. The
upper limit for the HI mass to total mass ratio is comparable to
the ratio in dark matter dominated dwarf irregular galaxies such
as DDO 154 (Carignan & Beaulieu 1989) at the last measured
point of the rotation curve. However, the HI mass to total mass
ratio is relatively constant with radius because the HI partial ro-
tation curve mimics the total rotation curve to a high accuracy
(Carignan & Beaulieu 1989, Broeils 1992). The combination
MHI /Mtotal = 0.33 within 1.3 kpc and MHI /Mtotal < 0.04
within 5 kpc, is exceptional. This contradiction is solved if the
system is not gravitationally bound.
The velocity difference between NGC 1569 and
NGC 1569-HI is 40 km s−1, their projected separation is
Fig. 4. Spectrum of background continuum sources normalized
to the mean of 91 mJy between −300 kms−1 and −60 kms−1;
rms fluctuations are 4%. The brightest continuum source con-
tributes approximately 50 % of the flux. Baselines shorter than
300 m have been excluded in order to suppress foreground
emission. No velocity smoothing has been applied.
NGC 1569 is complex H, at a distance of 5 degrees (Wakker
1991) and is in the relevant velocity range of −120 kms−1 to
−140 kms−1. In Table 1 we compare the properties of NGC
1569-HI to those of HVC cores, the dwarf galaxy M81 dwA,
and a sample of dwarf galaxies.
No distinction can be made from the column density. How-
ever, the linewidth of NGC 1569-HI is much higher than ex-
pected for small-scale structure in the Galactic foreground. The
timescale for free expansion to double its present size is short,
2 · 104dkpc year, where dkpc is the distance in kpc. Together
with the lack of other isolated HI clouds in the field of view,
the apparent connection (bridge) with NGC 1569, and the rela-
tively high contrast with the surroundings, leads us to conclude
that NGC 1569-HI is not a condensation in the Galactic fore-
ground.
Although NGC 1569 shows the signs of a global outflow
of gas (see e.g. Heckman et al. 1995), the mass of NGC 1569-
HI and the bridge (7 · 106 M⊙ each) as well as their location
projected along the major axis all rule out that they result from
this outflow.
Extended, low-column-density HI haloes have been detec-
ted in a number of late type galaxies (Huchtmeier & Richter
1982). NGC 1569-HI and the western arm are possibly high
density regions in a similar halo. The only evidence for ex-
tended emission with a column density NHI ∼ 1020cm−2 from
the present data is the emission in the north-west of NGC 1569.
Its possible counterpart, symmetric in position and in velocity
with respect to the centre of NGC 1569 should appear in the
channel maps near −60km s−1. Negative sidelobes of Galactic
J.M. Stil & F.P. Israel: A low-mass HI companion of NGC1569?
5
Table 1. Physical parameters of NGC 1569-HI compared with
HVC cores and low-mass dwarf galaxies.
Acknowledgements. We wish to thank R. Sancisi, W.B. Burton and P.
Van der Werf for their useful comments about this paper and E. Deul
for his help with the astroscan.
References
Bajaja E., Huchtmeier W.K., Klein U., 1994, A&A 285,385
Broeils A.H., 1992, A&A 256,19.
Burton W.B., Verheijen M.A.W., Kraan-Korteweg, R.C. et al., 1996,
A&A 309,687.
Carignan C., Beaulieu S., 1989, ApJ 347,760.
Della Ceca R., Griffiths R.E., Heckman T.H., Mackenty J.W., 1996,
ApJ 469,662
de Vaucouleurs G., de Vaucouleurs A., Buta R., 1981, AJ 86,1429
de Vaucouleurs G., de Vaucouleurs A., Corwin H.G. et al., 1991, Third
Reference Catalogue, Springer Verlag, N.Y.
Hartmann D., Burton W.B., 1997, Atlas of galactic neutral hydrogen,
Cambridge University Press
Heckman T.M., Dahlem M. Lehnert M.D., et al., 1995, ApJ 448,98
Huchtmeier W.K., Richter O.-G., 1982, A&A 109,331
Hunter D.A., Hawley W.N., Gallagher III J.S., 1993, AJ 106,1797
Israel F.P., 1988, A&A 194,24
Israel F.P., De Bruyn A.G., 1988, A&A 198,109
Israel F.P., Van Driel W., 1990, A&A 236,323
Karachentsev I.D. and Tikhonov N.A., 1993, In: Meylan G., Prug-
niel P. (eds.) proceedings ESO/OHP Workshop "dwarf galaxies"
p. 109
Kraan-Korteweg R.C., Loan A.J., Burton W.B., et al., 1994, Nat. 372,
77
Lo K.Y., Sargent W., Young K., 1993, AJ 106,507
McCall M.L., Buta R.J. 1995, AJ 109, 2460
McCall M.L., Buta R.J. 1997, AJ 113, 981
Reakes M., 1980, MNRAS 192,297
Sargent W.L.W., Sancisi R., Lo K.Y., 1983, ApJ 265,711
Shostak G.S., Skillman E.D., 1989, A&A 214,33
Taylor C.L., Brinks E., Grashuis R.M., et al. , 1995 ApJS 99,427
Taylor C.L., Brinks E., Grashuis R.M., et al. , 1996 ApJS 99,427
Vallenari A., Bomans D.J., 1996, A&A 313,713
Wakker B.P., Schwarz U.J., 1991, A&A 250,484
Waller W.H., 1991, ApJ 370,144
dMpc
0.6 dMpc
3 · 1020
5
-
0.76 - 5
0.8 - 3.4
2.5 - 42
20 - 36
23.4 - 25.2
N1569HIa HVC coreb M81 dwAc LSY dw.d
2.2
3.25
1.9 × 1.3
4 · 1020
19
24.4
dMpc
DHI (kpc) 2.6 × 1.3
3.5 · 1020
NHI,p
50
∆v
> 23.5
µB
Note : dMpc : distance in Mpc; DHI : HI diameter (kpc); NHI,p : peak
HI column density in 1020 HI cm−2; ∆v : HI FWHM line width
(km s−1); µB : mean surface brightness in B (mag arcsec−2); Ref-
erences: a. This work; b. Wakker & Schwarz (1991); c. Sargent et al.
(1983); d. range of the sample of 9 faint dwarfs in Lo et al. (1993).
5 kpc, and their HI mass ratio is about 0.06 (excluding the
bridge). Corresponding values for other pairs of dwarf galaxies
studied by Taylor et al. (1995) are 50 km s−1, 30 kpc, and 0.2
respectively. Thus, NGC 1569 and NGC 1569-HI appear to be
a close, but not an exceptional pair.
The virial mass of NGC 1569-HI is Mvir ∼ RHI∆v 2 /G =
6 · 108 M⊙. If the HI mass ratio is representative for the total
mass ratio, the ratio of the tidal radius to their actual separation
is 0.4. Judging from their projected separation, NGC 1569-HI
almost fills its tidal radius. Tidal forces exerted by NGC 1569
may thus explain the HI bridge. However, the western arm is
unlikely to be the result of tidal forces exerted by NGC 1569-
HI given the small mass and the apparent retrograde orbit of
NGC 1569-HI. This leaves the western arm unexplained. If
both are part of an outer halo which is not in equilibrium, as
was suggested for IC 10, this problem does not exist.
It is evident from Fig. 3 that the velocity structure of NGC
1569-HI and the bridge is well ordered. It is therefore a coher-
ent structure and not a chance superposition of multiple clouds.
The timescale on which this structure evolves is in the order
d/∆v = 108years. It is therefore a transient feature. Shostak
& Skillman (1989) found a structure similar to the HI bridge
and a disturbed location where this "plume" meets the disk of
IC 10. This was used as an argument in favour of interaction.
A similar argument can be made for NGC 1569.
The present data do not unequivocally show that the bridge
is falling into NGC 1569, but we may note some indications
that this is so. First, the presumed contact region exhibits
chaotic kinematics (see Heckman et al. 1995 for Hα and our
forthcoming paper for HI kinematics). Second, the Hα arcs
and the extended X-ray feature described by Heckman et al.
(1995) are found in the same direction. However, Hα filaments
are common in starburst galaxies, while a galactic wind is still
needed to obtain the high Mach number required to heat the gas
to X-ray temperatures. Finally, we note that the centre of the
starburst is offset to the south from the high column density HI
ridge, i.e. towards the bridge. It is tempting to speculate about
a possible connection between recent bursts of star formation
and infalling material.
|
0803.1832 | 2 | 0803 | 2008-03-31T00:01:44 | Extragalactic Zeeman Detections in OH Megamasers | [
"astro-ph"
] | We have measured the Zeeman splitting of OH megamaser emission at 1667 MHz from five (ultra)luminous infrared galaxies ([U]LIRGs) using the 305 m Arecibo telescope and the 100 m Green Bank Telescope. Five of eight targeted galaxies show significant Zeeman-splitting detections, with 14 individual masing components detected and line-of-sight magnetic field strengths ranging from ~0.5-18 mG. The detected field strengths are similar to those measured in Galactic OH masers, suggesting that the local process of massive star formation occurs under similar conditions in (U)LIRGs and the Galaxy, in spite of the vastly different large-scale environments. Our measured field strengths are also similar to magnetic field strengths in (U)LIRGs inferred from synchrotron observations, implying that milligauss magnetic fields likely pervade most phases of the interstellar medium in (U)LIRGs. These results provide a promising new tool for probing the astrophysics of distant galaxies. | astro-ph | astro-ph | Accepted for Publication in The Astrophysical Journal
Preprint typeset using LATEX style emulateapj v. 08/22/09
8
0
0
2
r
a
M
1
3
]
h
p
-
o
r
t
s
a
[
2
v
2
3
8
1
.
3
0
8
0
:
v
i
X
r
a
EXTRAGALACTIC ZEEMAN DETECTIONS IN OH MEGAMASERS
Timothy Robishaw, Eliot Quataert, & Carl Heiles
Department of Astronomy, University of California at Berkeley, 601 Campbell Hall, Berkeley, CA 94720-3411;
[email protected], [email protected], [email protected]
Accepted for Publication in The Astrophysical Journal
ABSTRACT
We have measured the Zeeman splitting of OH megamaser emission at 1667 MHz from five (ul-
tra)luminous infrared galaxies ([U]LIRGs) using the 305 m Arecibo telescope and the 100 m Green
Bank Telescope. Five of eight targeted galaxies show significant Zeeman-splitting detections, with
14 individual masing components detected and line-of-sight magnetic field strengths ranging from
≃ 0.5 − 18 mG. The detected field strengths are similar to those measured in Galactic OH masers,
suggesting that the local process of massive star formation occurs under similar conditions in (U)LIRGs
and the Galaxy, in spite of the vastly different large-scale environments. Our measured field strengths
are also similar to magnetic field strengths in (U)LIRGs inferred from synchrotron observations, imply-
ing that milligauss magnetic fields likely pervade most phases of the interstellar medium in (U)LIRGs.
These results provide a promising new tool for probing the astrophysics of distant galaxies.
Subject headings: galaxies: magnetic fields -- ISM: magnetic fields -- magnetic fields -- masers --
polarization -- radio lines: galaxies
1. INTRODUCTION
(Ultra)luminous infrared galaxies ([U]LIRGs) are a
population of galaxies that emit far-infrared (FIR) radia-
tion with energies comparable to those of the most lumi-
nous quasars (log(LFIR/L⊙) > 11 and 12 for LIRGs and
ULIRGs, respectively). Nearly every ULIRG appears to
have undergone a merger/interaction and contains mas-
sive star formation and/or an active galactic nucleus
(AGN) induced by gravitational interactions. Lo (2005)
details very long baseline interferometry (VLBI) obser-
vations of the 1667 MHz hydroxyl (OH) transition in the
nuclear regions in (U)LIRGs that have revealed multiple
masing regions with 1 < log(LOH/L⊙) < 4; these regions
are known as OH megamasers (OHMs). Each OHM has
a spectral linewidth of between 50 and 150 km s−1; when
viewed by a single dish, these spectral components are
superimposed. The 1667 MHz OHM flux density is al-
ways a few to many times that of the 1665 MHz tran-
sition, and in many cases the 1665 MHz line is absent
(Darling & Giovanelli 2002); this is an interesting con-
trast to the case of OH masers in the Galaxy in which the
1665 MHz transition is usually dominant (Reid & Moran
1988). The starbursts and AGNs in ULIRGs create
strong FIR dust emission as well as a strong radio con-
tinuum; the OHMs are generally believed to be pumped
by the FIR radiation field (e.g., Randell et al. 1995) al-
though collisional excitation may be important as well
(e.g., Lonsdale et al. 1998). Lockett & Elitzur (2008)
have recently suggested that the 53 µm OH pump lines
in addition to line overlap of large (& 10 km s−1) turbu-
lent linewidths can account for the observed dominance
of the 1667 MHz transition in OHMs. They further argue
that pumping due to FIR radiation can explain all ob-
served main-line OH masers, both those in Galactic star-
forming regions and those in OHM galaxies. Given the
conditions that exist in ULIRGs and considering that so
many OH masers in our Galaxy are associated with mas-
sive star-forming regions (Fish et al. 2003), it is therefore
not surprising that the entire OHM sample finds homes
in LIRGs, strongly favoring the most FIR-luminous, the
ULIRGs (Darling & Giovanelli 2002).
The high gas and energy densities in ULIRGs make
them natural locations to expect very strong magnetic
fields. Much of the radio emission in ULIRGs is re-
solved on scales of ∼ 100 pc with VLA observations
(Condon et al. 1991). High-resolution observations of
Arp 220 (Rovilos et al. 2003) show that the OHMs arise
in this region as well. With this size scale and the ob-
served radio flux densities, minimum energy arguments
suggest volume averaged field strengths of ≈ 1 mG (e.g.,
Condon et al. 1991; Thompson et al. 2006), which are
significantly larger than the ∼ 10 µG fields in normal
spirals. The field strengths in ULIRGs cannot be much
below a mG or else inverse Compton cooling would dom-
inate over synchrotron cooling, making it energetically
difficult to explain the radio flux densities from ULIRGs
and the fact that ULIRGs lie on the FIR-radio correla-
tion. The field strengths could, however, in principle be
larger than the minimimum energy estimate if, as in our
Galaxy, the magnetic energy density is in approximate
equipartition with the total pressure (Thompson et al.
2006). The latter can be estimated from the observed
surface density. CO observations of Arp 220 and several
other systems reveal ∼ 109M⊙ of molecular gas in the
central ∼ 100 pc (e.g., Downes & Solomon 1998) imply-
ing gas surface densities Σ ∼ 1 − 10 g cm−2, 103 − 104
times larger than in the Milky Way (MW). The equipar-
tition field scales as B ∝ Σ, implying that the mean field
in ULIRGs could approach ∼ 10 mG.
Motivated by the above considerations, we carried out
a survey of 8 (U)LIRGs searching for Zeeman splitting
in OHMs. This paper presents our results, which repre-
sent the first detections of extragalactic Zeeman splitting
from an emission line and the first extragalactic detec-
tions within an external galaxy proper. The only pre-
vious extragalactic detection was made by Kazes et al.
(1991) and confirmed by Sarma et al. (2005) via absorp-
2
Robishaw, Quataert, & Heiles
tion of 21 cm emission in a high-velocity system towards
NGC 1275 (Per A). § 2 outlines what is known about each
of our targets. In § 3 we describe the observations. In
§ 4 we discuss the data reduction and calibration method.
§§ 5 and 6 present a summary and discussion of the re-
sults, respectively.
2. SOURCE SELECTION
In selecting our sample of targets from the compilation
of all known OHMs by Darling & Giovanelli (2000, 2001,
2002), we chose the three simplest criteria possible. We
selected 12 (U)LIRGs: (a) with the largest OHM peak
flux densities; (b) whose discoverers did not regard the
OHM detection validity as suspicious; (c) that are ob-
servable from Arecibo, Puerto Rico, or Green Bank, West
Virginia. Our sample includes 2 of only 3 known OH
gigamasers (LOH > 104L⊙; Darling & Giovanelli 2002).
Here we summarize what is known about each source and
its OHM emission.1
IRAS F01417+1651: This LIRG is most commonly
known2 as III Zw 35 and has an optical heliocentric red-
shift of z = 0.0274.
It is a double galaxy system and
is classified as a Seyfert 2 galaxy. Staveley-Smith et al.
(1987) present a single-dish spectrum from the Jodrell
Bank Mk1A 76 m telescope showing emission from the
1667 MHz transition at a velocity of 8262 km s−1 with a
peak flux density of 240 mJy and a total velocity extent
of 270 km s−1 at the 10% flux density level. The line
profile can easily be seen to have at least three compo-
nents. The 1665 MHz line is also weakly detected (≃ 25
mJy) and completely separated from the 1667 MHz emis-
sion, with an estimated hyperfine line ratio (defined as
RH ≡ R f1667 dν/R f1665 dν, where the integrals repre-
sent the total flux density of each transition) of RH ≈ 9.
Killeen et al. (1996) observed III Zw 35 using the Aus-
tralia Telescope Compact Array (ATCA). Their goal was
an attempted detection of Zeeman splitting in the OHM
emission. The peak flux density was 247.6 mJy and their
sensitivity was 3.6 mJy. They observed no Zeeman split-
ting and their model-dependent estimate for a 3σ upper
limit on the line-of-sight magnetic field was 4.0 mG.
Diamond et al. (1999) present global VLBI observa-
tions of the OHM emission in III Zw 35. They label
two regions of 1667 MHz OHM emission in the south (S1
and S2) and three in the north (N1 -- N3), each region cov-
ering about 20 mas and separated by 90 mas. They re-
cover 60% of the single-dish flux density. Pihlstrom et al.
(2001) performed simultaneous high-resolution observa-
tions of the OHM emission in III Zw 35 using both the
European VLBI Network (EVN; baselines between 198 --
2280 km) and the Multi-Element Radio-Linked Interfer-
ometer Network (MERLIN; operated by Jodrell Bank
Observatory with baselines between 6.2 -- 217 km). The
map of the 1667 MHz emission shows two compact re-
gions coincident with the northern and southern sources
1 Unfortunately, we only obtained usable data for 2 of the 6
sources observed at Green Bank, therefore we only provide source
descriptions for the 8 (U)LIRGs for which we have presentable
results.
2 As a shorthand, when referring to source names in the text by
their IRAS designation, we shall henceforth use the right ascension
designator only; we shall refer to 01417, 10038, and 15327 by their
more common designators, III Zw 35, IC 2545, and Arp 220, re-
spectively. We shall retain the full IRAS designation in figure and
table captions as well as section headings.
of Diamond et al. (1999) connected by two bridges of
weaker, more diffuse emission. In total, 80% of the single-
dish flux density was recovered. A velocity gradient of
≃ 1.5 km s−1 pc−1 is observed from the southern to the
northern regions and is evident in the diffuse component.
The emission is modelled as a torus of multiple maser
clouds inclined at 60◦; the compact OHM emission would
be seen at the tangent points where a few clouds could
be superimposed in such a fashion that strong OHM
emission would be produced from the foreground clouds
amplifying those in the background. At the front and
back of the torus, the emission would be weak because
the path lengths through the torus are small and the
clouds are less likely to overlap (Pihlstrom et al. 2001;
Parra et al. 2005).
IRAS F10038−3338: Also known as IC 2545, this
LIRG is a set of interacting galaxies at z = 0.0341. A
single-dish spectrum made using the Parkes 64 m tele-
scope is presented by Staveley-Smith et al. (1992) show-
ing 1667 MHz OHM emission centered at 10093 km s−1
with a full width at half maximum (FWHM) velocity
range of 63 km s−1 and a peak flux density of 315 mJy.
Likely due to its low declination, there have been no
VLBI observations of this source despite its brightness.
Killeen et al. (1996) present an ATCA spectrum with
much better sensitivity (5.4 mJy in Stokes I) and veloc-
ity resolution than that of Staveley-Smith et al. (1992);
the 1667 MHz emission contains five narrow peaks super-
imposed on a broad emission component. The brightest
component has a peak flux density of 260 mJy and a
velocity of 10097 km s−1. Killeen et al. (1996) failed to
detect Zeeman splitting and estimated that the line-of-
sight field should be less than 4.3 mG.
IRAS F10173+0829: The only single-dish observa-
tions of the OHM line emission in this LIRG at z =
0.0480 were made with the 305 m Arecibo telescope and
are detailed in Mirabel & Sanders (1987). There are two
distinct peaks in the profile of the 1667 MHz emission
with a separation of about 100 km s−1, the dominant
peak having a velocity of 14720 km s−1 with a FWHM
of 39 km s−1 and a peak flux density of about 105 mJy.
The 1665 and 1667 MHz lines are well separated with a
hyperfine line ratio RH = 14.6.
MERLIN observations made by Yu (2004, 2005) show
roughly 50 maser spots distributed into three clumps, la-
belled east, central, and west, over an area of 1.′4 × 0.′6.
The spots within each clump are distributed along a line,
with each of the three lines having a different direction;
Yu (2005) proposes (without much justification) that the
spots may be distributed along a warped circumnuclear
torus seen edge-on. The OHM emission is seen only at
the 1667 MHz transition and is coincident with the in-
frared central position.
IRAS F11506−3851: Also known as ESO 320-30,
this LIRG is classified as an H II galaxy at z =
0.0108. A Parkes single-dish spectrum is presented by
Staveley-Smith et al. (1992) showing 1667 MHz OHM
emission centered at 3103 km s−1 with a FWHM ve-
locity extent of 87 km s−1 and a peak flux density of
105 mJy. There is neither enough sensitivity nor band-
width to clearly discern any 1665 MHz emission. There
are no interferometric observations of this source:
like
IC 2545, the low declination of 11506 would hinder any
Extragalactic Zeeman Detections in OH Megamasers
3
attempted VLBI observations.
IRAS F12032+1707: A gigamaser discovered at
Arecibo by Darling & Giovanelli (2001). The host ob-
ject, a ULIRG at z = 0.2170, has been classified as
a LINER-type AGN (Veilleux et al. 1999). The OHM
emission spans almost 2000 km s−1 with a redshifted
high-velocity tail and a mean flux density of roughly 9
mJy. The 1665 and 1667 MHz lines are impossible to dis-
tinguish and clearly blended. A very narrow and bright
component is seen at 64500 km s−1 with a peak flux
density of 16.3 mJy.
Pihlstrom et al. (2005) used the Very Long Baseline
Array (VLBA) to show that the OHM emission is con-
fined to an area of 25 × 25 mas. All the single-dish
flux density was recovered. They were able to clearly
identify five peaks in their Stokes I spectrum that cor-
responded with Darling's single-dish spectrum. By aver-
aging channels around each peak, they found the maser
components were spatially separated and aligned roughly
north-south, implying an ordered velocity gradient. No
continuum emission was detected, implying that the con-
tinuum emission is resolved out on scales less than 75
mas.
IRAS F12112+0305: This ULIRG is classified as a
LINER-type AGN and is an interacting pair of galax-
ies at z = 0.0730. The only information concerning the
OHM emission in this ULIRG is listed in tabular form
in Staveley-Smith et al. (1992); no spectrum has been
published. The 1667 MHz line was measured at a veloc-
ity of 5540 km s−1 and no information about the 1665
MHz transition is published. Its 1667 MHz flux density
is listed in Darling & Giovanelli (2002) as 45 mJy. There
are no VLBI observations of the 1667 MHz OHM emis-
sion for this source.
IRAS F14070+0525: Discovered by Baan et al.
(1992), this gigamaser is the most distant OHM at a
redshift of z = 0.2644. Darling & Giovanelli (2002)
redetected this source in their survey. The OH lines
are so wide and blended (1580 km s−1 at 10% peak
flux density) that it is impossible to identify any 1665
MHz emission. The spectral line profile measured by
Darling & Giovanelli (2002) has a peak flux density of 8.4
mJy and shows no significant changes since the original
detection by Baan et al. (1992). Darling & Giovanelli
(2002) suggest that many peaks in the profile are likely
the result of many masing nuclei within the host ULIRG,
which is classified as a Seyfert 2.
VLBA observations made by Pihlstrom et al. (2005)
recovered less than 10% of the single-dish flux density
and only two of the many single-dish spectral peaks were
detected. Most of the single-dish emission is therefore
diffuse. The spatial extent of the VLBA emission is con-
fined to 10 × 10 mas.
IRAS F15327+2340: This is perhaps the most well-
known ULIRG and is better known as Arp 220 or IC
4553. Baan et al. (1982) discovered the OHM emission
using Arecibo and list the single-dish properties of the
OHM emission as having a velocity of 5375 km s−1 and
FWHM velocity extent of 108 km s−1. The spectrum
clearly shows that the 1665 and 1667 MHz transitions
are distinct with a hyperfine line ratio RH = 4.2 and a
peak flux density of 320 mJy.
Smith et al. (1998) used global VLBI continuum imag-
ing at 18 cm to show that the high brightness tem-
perature core of Arp 220 is composed of multiple com-
pact sources, which they interpret as luminous radio
supernovae (RSNe). These RSNe are not coincident
with the compact 1667 MHz OHM spots discovered by
Lonsdale et al. (1998). More recently, Lonsdale et al.
(2006) have used high-sensitivity 18 cm VLBI observa-
tions of the nuclei to detect 4 previously unseen sources
in a one year period, supporting the RSNe interpretation.
Parra et al. (2007) have made the first multiwavelength
observations of these compact sources; they identify a
fraction of these sources to be supernova remnants.
Rovilos et al. (2003) present MERLIN maps of the
18 cm continuum and OHM line emission in Arp 220; two
components are seen roughly 1′′ apart, each coinciding
with a nucleus imaged in the infrared by Graham et al.
(1990). The OHM emission is resolved into one com-
ponent aligned with the eastern continuum feature and
two components that are aligned north to south strad-
dling the western nucleus. Lonsdale et al. (1998) and
Rovilos et al. (2003) present global VLBI spectral line
maps that show that the OHM emission is resolved
into multiple compact spots. The northernmost features
in both the eastern and western nuclei form elongated
ridges.
3. OBSERVATIONS
In 2006 February we used the L-band wide receiver of
the 305 m Arecibo3 telescope in full-Stokes mode in an
attempt to detect Zeeman splitting of the 1667 MHz OH
transition in the 6 positive-declination sources listed in
§ 2.
Since the spatial extent of each source is much smaller
than Arecibo's 3.′3 beam, our observing method was to
simply spend equal time at on-source and off-source po-
sitions. In this position-switching scheme, we alternated
between 4 minutes on source and 4 minutes at a reference
position having the same declination as the source and
a right ascension 4 minutes east of the source. In this
way, the hour angle ranges of the source and reference
observations were nearly identical. Our integration time
was one second, allowing us to remove short-term radio-
frequency interference (RFI). The total integration time
for each source is as follows: 5.4 hours for III Zw 35, 2.7
hours each for 10173 and 12032, 3.1 hours for 12112, 4.6
hours for 14070, and 5.9 hours for Arp 220. We config-
ured the correlator to produce 4 spectra per integration:
one 6.25 MHz bandpass centered on the mean of the 1665
and 1667 MHz transitions; 2 narrow bandwidths (either
3.125 or 12.5 MHz, depending on the velocity extent of
the source) centered on the 1665 and 1667 MHz transi-
tions, respectively; and one wide bandwidth (either 12.5
or 25 MHz) centered on the mean of the 1665 and 1667
MHz transitions. We calibrated the Mueller matrix for
Stokes parameters using the standard Arecibo technique
(Heiles et al. 2001; Heiles & Troland 2004) of observing
spider scans on the linearly-polarized continuum sources
3C 138 and 3C 286.
In 2005 December we used the L-band receiver of the
3 The Arecibo Observatory is part of the National Astronomy
and Ionosphere Center, which is operated by Cornell University
under a cooperative agreement with the National Science Founda-
tion.
4
Robishaw, Quataert, & Heiles
100 m Robert C. Byrd Green Bank Telescope4 (GBT) to
observe an additional six OHM galaxies. All but the two
sources in § 2 at negative declinations were affected by
insidious RFI that left our data corrupted beyond salva-
tion (For completeness, the sources that were obliterated
by RFI were IRAS F12540+5708, IRAS F13428+5608,
IRAS F17207−0014, and IRAS F20100−4156.) We used
two observing methods for each source: position switch-
ing (as described above) and least-squares frequency
switching (LSFS; see Heiles 2007 for the details of this ob-
serving method and its corresponding reduction scheme).
The LSFS method was used to accurately derive the gain
for each integration; the data were then combined in
the standard way using the off-source, position-switched
spectra. Our off-source positions were 23 minutes east
of each on-source position in order to cause the GBT to
track as closely as possible the path of our on-source ob-
servations. We used a 12.5 MHz bandwidth and 9-level
sampling for all observations.
We accumulated 4.0 hours of RFI-free, on-source inte-
gration time for IC 2545 and 5.8 hours for 11506. As we
did at Arecibo, we observed 3C 286 using spider scans
in order to calibrate the L-band Mueller matrix at the
GBT.
4. DATA REDUCTION
The complex Stokes I (see footnote5) line shape in each
of the maser sources is a composite of many narrow maser
lines at various velocities spread about the systemic ve-
locity of the system. Therefore, we chose to least-squares
fit each line profile with multiple Gaussian components.
Without VLBI observations, it is impossible to attribute
any particular velocity or width to a Gaussian compo-
nent within the profile. The only method available to us
for assessing a possible field strength from each Stokes V
spectrum was to decompose each I profile into the fewest
number of Gaussian components that would yield rea-
sonable residuals while also allowing enough components
to reproduce the multiple splittings in the V spectrum.
None of the parameters in our multiple-component Gaus-
sian fits were held fixed. We discuss our method in more
detail in § 4.4.
4.1. Calibration
The derived Mueller matrix was applied to all OHM
observations to correct the polarization products and ob-
tain the pure Stokes spectra for each observed source. We
converted from antenna temperature to flux density by
assuming the antenna gain to be 10.0 K Jy−1 at Arecibo
and 2.0 K Jy−1 at the GBT; these gains were estimated
from observations of standard flux density calibrators.
We follow the IAU definition for Stokes V , namely
V = RHCP − LHCP , where RHCP is the IEEE def-
inition of right-hand circular polarization.6 We deter-
mined the sense of Stokes V at Arecibo by observing
4 The National Radio Astronomy Observatory is a facility of
the National Science Foundation operated under cooperative agree-
ment by Associated Universities, Inc.
5 We use the classical definition of Stokes I, which is the sum
(not the average) of two orthogonal polarizations. Thus, stated
Stokes I flux densities are twice those listed in § 2 and in other
catalogs.
6 Defined as a clockwise rotation of the electric vector along the
direction of propagation.
the highly circularly polarized 1665 MHz Galactic maser
W49(OH); the result is consistent with the measurements
of Rogers et al. (1967) and Coles & Rumsey (1970). The
sense of Stokes V has not yet been determined for the
GBT Autocorrelation Spectrometer, which only began
functioning with full-Stokes capability months prior to
these observations.
4.2. RFI Removal
We examined each set of spectra, both off source and
on, for RFI and rejected suspicious-looking data, which
constituted only a few percent for only two sources, III
Zw 35 and Arp 220. The other sources were completely
free of RFI except for occasional monochromatic signals
whose topocentric frequencies are constant; fortunately,
most of these fall off the OHM lines.
For one source, 10173, the monochromatic RFI fell on
the OHM line. We observed this source over several days,
during which the changing Doppler shift moved the RFI
across part of the OH spectrum. For each day the RFI
was a sharp spike with the usual ringing sidelobes. We
Hanning smoothed each spectrum, which eliminated the
ringing, and interpolated across each day's spike, which
effectively removed the RFI.
The source III Zw 35, whose OH lines are centered
near 1622.5 MHz, was highly contaminated by RFI that
probably arises from the Iridium communications satel-
lites. The RFI consists of a spikey pattern that repeats
periodically across the spectrum at about a 0.33 MHz
interval. It was impossible to obtain reasonable results
by averaging data. However, by taking medians instead
of averaging, the spectra look quite good and the RFI
is reduced to levels of about 20 mJy in Stokes I and 3
mJy in Stokes V -- levels that are considerably smaller
(< 1%) than the OHM spectral features.
4.3. Bandpass and Gain Correction
We correct our spectra for the intermediate-frequency
bandpass. Both the Arecibo L-band wide receiver and
the GBT L-band receiver are dual-polarized feeds with
native linear polarization. For Stokes I and Q, we divide
each of the two linear polarization spectra (XX and Y Y )
by its associated off-source spectrum; then we add the
results to obtain Stokes I and subtract them to obtain
Q. To generate Stokes U and V , we combine the cross-
correlation spectra (XY and Y X) having divided by the
square root of the product of the off-source XX and Y Y
spectra.
We always show difference spectra: on-source minus
off-source. Normally, on- and off-source spectra are com-
bined by subtracting the latter from the former. If the
two spectra have equal noise σ, then the noise in the
difference is √2σ. Our off-source spectra have no fine-
scale frequency structure, so we can reduce the noise by
smoothing. We use a Fourier technique to smooth the off-
source average spectrum. By zeroing lags at high delays
in the autocorrelation function of the average off-source
spectrum and then Fourier transforming, we nearly elim-
inate the noise contribution from the off-source spectrum
while retaining the shape of the bandpass. This reduc-
tion in noise is particularly important for the polarized
Stokes parameters, which are weak.
As mentioned above, Stokes U and V are obtained via
Extragalactic Zeeman Detections in OH Megamasers
5
the cross-correlation products XY and Y X: this insu-
lates them from system gain fluctuations. However, since
Stokes Q is the difference between the two native linear
polarizations, it is susceptible to gain fluctuations. We
defer the discussion of linear polarization to § 5.2.
After gain and bandpass correction, the on- and off-
source spectra were averaged separately and combined
to yield the final average Stokes spectra.
4.4. Fitting Gaussian Components to Stokes I Profiles
Fitting multiple Gaussians to complicated spectral
profiles carries a significant degree of subjectivity because
the fits are nonlinear. Generally, nonlinear fits require
beginning from initial "guessed" parameters and letting
the fit converge with successive iterations (Press et al.
1992). Nonlinear fits usually have multiple χ2 minima
and the particular minimum selected depends on the
initial guesses, which in turn depend on the subjective
judgement of the person doing the fitting. Therefore, we
outline the following guidelines that were used for select-
ing the initial guesses for Gaussian components in the
fits to the Stokes I spectra:
1. For each peak (i.e., local maximum) in I, we in-
cluded a single Gaussian component whose three
parameters (flux density, central frequency, and fre-
quency width) were visually estimated.
2. Many I peaks are distinctly asymmetric. We fitted
these asymmetries by including one or two Gaus-
sian components with visually-estimated parame-
ters in addition to the central component of guide-
line (1).
3. The Gausssian components estimated in guidelines
(1) and (2) are usually fairly narrow and lie on top
of one or two underlying broader lines -- core-halo
structure. We included one or two broad Gaussian
components to represent these broader lines.
4. With all of the above, our goal was to use the fewest
number of Gaussian components that would yield
reasonable residuals.
It was straightfoward to apply the above guidelines to
the sources III Zw 35, IC 2545, 11506, 12112, 14070, and
even Arp 220, for which we fitted 18 components. The
sources 10173 and 12032 are somewhat more complex.
In §§ 5.1.1 -- 5.1.8 we describe how we applied the above
selection guidelines when appropriate.
We stress that our Gaussian-component representa-
tions are not unique. Particular problems include:
1. In guideline (1) above, noise prevents us from iden-
tifying weak components. This introduces a sensi-
tivity cutoff. Noise also prevents us from distin-
guishing two or more closely-spaced blended real
components from a single broader component. Be-
cause we favor choices with the fewest number of
components, this introduces a bias towards wider
components.
2. In guidelines (2) and (3) above, whether to rep-
resent a peak needing multiple Gaussians by an
asymmetry or core-halo structure can be extremely
subjective. For example, the combination of two
narrow Gaussians separated by a fraction of their
FWHMs can closely mimic the combination of a
broad and a narrow Gaussian with roughly the
same centers.
In summary, for most sources our Gaussian fits fol-
low our fitting guidelines in a reasonably straightforward
fashion; if the fits were done by other people who followed
these guidelines, the components would be mostly repro-
duced. However, Gaussian component fitting has uncer-
tainties as mentioned above, particularly when compo-
nents are blended and signal-to-noise (S/N) is low.
5. RESULTS
For each source, the Stokes I spectrum exhibits a fairly
broad, relatively smooth underlying profile for the OHM
emission on top of which small bumps from individual
masers can be seen. VLBI studies show that the un-
derlying profile arises from spatially extended OH emis-
sion and, sometimes, an assembly of many masers that
are not individually recognizable (Pihlstrom et al. 2005;
Diamond et al. 1999). We fit the Stokes I spectrum for
each source with a series of Gaussian profiles. For Arp
220, 12112, and 12032 we also fit a first-degree polyno-
mial (12032 required a second-degree polynomial in ad-
dition), since these profiles exhibit broad wings. A few
sources have a large number of discernable individual
masers: for example, we used 18 Gaussians components
for Arp 220.
We examined circular polarization for each source, and
linear polarization for the six Arecibo sources only. Five
of the eight sources exhibit significant circular polariza-
tion that is interpretable as Zeeman splitting, particu-
larly for the recognizable individual maser components.
For four of the sources, there is evidence that the mag-
netic field reverses direction between OHM spots within
the source.
We see detectable linear maser polarization in two
sources (possibly four) and are able to estimate Faraday
rotation in both. We present all spectra as a function
of frequency as viewed in the heliocentric frame. Since
all OHMs are extragalactic sources, OHM spectra are al-
most always presented versus optical heliocentric velocity
v⊙, which is conventionally defined as
v⊙
c ≡
ν0
ν − 1 ≡ z⊙ ,
(1)
where c is the speed of light, ν0 is the rest frequency
(which is taken to be 1667.359 MHz for OHMs since this
transition always dominates the 1665.4018 MHz transi-
tion), ν is the observed frequency, and z⊙ is the redshift
of the maser.
First, in § 5.1 we present the circular polarization re-
sults for each source in addition to describing the total
intensity properties. We present the linear polarization
results for each source observed with Arecibo in § 5.2.
For sources in which the 1665 and 1667 MHz emission
are separable, we calculate the hyperfine ratio RH .
5.1. Circular Polarization and Line-of-Sight Magnetic
Fields
For the usual case in which the Zeeman splitting is
small compared to the linewidth, the Stokes V spectrum
6
is given by
Robishaw, Quataert, & Heiles
V = (cid:18) ν
ν0(cid:19)(cid:18) dI
dν(cid:19) bBk ,
(2)
where Bk is the line-of-sight component of the magnetic
field at the OHM and b is known as the splitting coeffi-
cient7 (Heiles et al. 1993), equal to 1.96 Hz µG−1 for the
OH 1667.359 MHz transition;8 the factor ν/ν0, equiva-
lent to (1 + z⊙)−1, accounts for the frequency compres-
sion of redshifted lines.
In order to derive a magnetic
field strength, we need to least-squares fit the Stokes V
spectrum with the functional form of equation (2). As
is the custom in radio Zeeman work, we add a term on
the right that is linear in Stokes I to account for leakage
of I into the measured V . For Bk > 0 (by convention a
positive magnetic field points away from the observer), if
Stokes V is plotted as a function of frequency, V will be
positive on the low-frequency side of a Stokes I emission
line.
We solved equation (2) in two ways. In one, we simul-
taneously fitted Stokes V for multiple Gaussian compo-
nents (selected as outlined in § 4.4) to derive separate,
independent magnetic fields for each Gaussian.
In the
other, we chose a limited range in frequency ν, either 0.1
or 0.25 MHz, and fitted for Bk for the center of this range,
positioning the center sequentially at each spectral chan-
nel to obtain Bk as a function of frequency; we refer to
this as the B(ν) fit. The former method is appropriate
for individual masers, while the latter is more suitable for
the broad component. We plot the B(ν) results only for
sources for which the results provide additional insight.
Performing these fits requires the calculation of dI/dν.
The I profiles are somewhat noisy and the frequency
derivative is often very noisy. This means that traditional
least-squares fitting cannot be used, because it assumes
no error in the independent variables. Sault et al. (1990)
discuss this and suggest using a generalized maximum
likelihood technique. We choose the much simpler ap-
proach of using our multiple-Gaussian fit to the Stokes I
spectrum as the independent variable: it has no noise, so
it satisfies the requirements of the conventional method
of least squares.
We present two vertically stacked plots for each source
(Figures 1 -- 8). In the top panel, the Stokes I spectrum
is plotted as a solid line and the profiles of the Gaus-
sian components are plotted as dashed lines. The resid-
uals (the difference between the data and the composite
Gaussian fit) are plotted with enhanced vertical scale as
a solid line near the middle height of the panel. Scale
bars are plotted on both the residuals and the baseline
of the spectrum to the right of the OHM emission; the
height of each scale bar corresponds to the labelled flux
density. The bottom axes of both plots show heliocen-
tric frequency and the top axis of the top plot displays
the optical heliocentric velocity.
In the bottom panel,
the Stokes V spectrum is plotted as a solid line and the
7 The splitting coefficient is directly proportional to the Land´e
g-factor for the transition: b = 2gµ0/h, where µ0 is the Bohr
magneton and h is Planck's constant.
8 Modjaz et al. (2005) made a valiant effort to detect Zeeman
splitting of 22.2 GHz H2O megamasers in NGC 4258 using the
VLA and the GBT, but the splitting coefficient for this hyperfine
transition is nearly 1000 times weaker than that of the 1667 MHz
OH transition.
dashed line represents the best fit to equation (2). The
integers located between the top and bottom panels label
the number of each Gaussian component as assigned in
the corresponding tabular summary and are positioned
at the central frequencies of each component.9 The dis-
played spectra and residuals are smoothed over 7 chan-
nels for every source.
We assume that each Gaussian represents an emission
component for the 1667 MHz transition. Arp 220 and
12032 have line profiles that are too complex for the 1665
and 1667 MHz lines to be distinguished. This introduces
some uncertainty in our Zeeman splitting interpretations
in §§ 5.1.5 & 5.1.8. For sources where the 1665 MHz
transition is not blended with the 1667 MHz emission,
we present the spectra showing both transitions in § 5.2
and calculate the hyperfine line ratios. In all cases the
1665 MHz transition was too weak for Zeeman splitting
to be detected even if observed in the 1667 MHz line.
5.1.1. IRAS F01417+1651 (III Zw 35)
As we mentioned in § 4.2, our observations of III Zw
35 suffered severe RFI that we were able to greatly re-
duce by combining the data using medians instead of
averaging. There remains a spikey pattern that repeats
periodically across the spectrum at an interval of ≃ 0.33
MHz. Remarkably, this spikey pattern is restricted to an
8 MHz wide interval centered almost exactly on the OHM
lines. The spikey pattern appears in both the on-source
and off-source spectra, so we regard this as terrestrial
interference. Despite the RFI, Figure 1 shows that both
I and V are well-detected. In fitting Gaussians we are
conservative because we realize that the RFI may have
contaminated the line shape. In particular, the 0.33 MHz
intervals happen to fall close to the two peaks in Stokes
I.
Table 1 lists the parameters of the Gaussian compo-
nents that best fit the Stokes I spectrum. Column (1)
lists the zero-based component number. Column (2) lists
the peak flux density of each component in mJy and the
corresponding uncertainty. Column (3) lists the central
heliocentric frequency of each component in MHz and the
corresponding uncertainty. Column (4) lists the FWHM
of each component in MHz and the corresponding uncer-
tainty. Column (5) lists the optical heliocentric velocity
corresponding to the central frequency of each compo-
nent. Column (6) lists the derived line-of-sight magnetic
field in mG for each component and the corresponding
uncertainty.
Appying our guidelines from § 4.4 yielded the 5 Gaus-
sian components shown in Figure 1 and listed in Table
1. There are 2 distinct peaks in the OHM emission. The
peak nearest 1622.8 MHz is asymmetric in such a way
that 2 narrow components needed to be added near this
peak in order to minimize the residuals. The overall pro-
file has a core-halo structure, with the 4 narrow compo-
nents lying on top of a broader component.
The Stokes V spectrum (Srms = 0.97 mJy) in Figure 1
shows prominent features that are fitted reasonably well
by the 5 Gaussian components, with Zeeman splitting
yielding significant fields in three Gaussians: Gaussian 0
9 Where multiple labels overlap, the font size has been reduced
and the labels stacked corresponding to their associated flux den-
sities.
Extragalactic Zeeman Detections in OH Megamasers
7
has Bk = 2.9±0.2 mG, and Gaussians 3 and 4 have fields
of −2.7 ± 0.1 and −3.6 ± 0.3 mG, respectively. Thus the
field reverses from one peak to the other. Pihlstrom et al.
(2001) present 13 spectra from the 1667 MHz OHM emis-
sion of III Zw 35 that they mapped using the EVN. These
data show clearly that the 8215 and 8240 km s−1 compo-
nents (Gaussians 3 and 4) arise from the southern peak,
while the 8312 km s−1 component (Gaussian 0) is as-
sociated with the northern peak. This provides clear
evidence that the reversal is arranged with the magnetic
field pointing away from us in the north and towards us
in the south.
The panels on the right side of Figure 1 show results
relevant to the B(ν) fit. The top panel shows the compos-
ite Gaussian-fitted (noise-free) Stokes I spectrum. The
center panel shows the measured Stokes V spectrum as a
solid line; the dashed line represents the Stokes V spec-
trum that would be produced by a uniform line-of-sight
magnetic field of 1 mG -- this is obtained from equa-
tion (2) by setting Bk = 1 mG and using the derivative
of the composite Gaussian shown in the top panel. The
bottom panel shows the B(ν) fit -- the derived Bk as a
function of frequency -- as described in § 5.1. There is
a clear systematic pattern, with the field reversing sign
from one peak to the other. The estimated field strengths
are also consistent with the Gaussian fits.
Parra et al. (2005) and Pihlstrom et al. (2001) present
models of the OHM emission in III Zw 35 as a clumpy,
rotating starburst ring at an inclination of 60◦, with an
inner radius of 22 pc and a radial thickness of 3 pc.
Both the Gaussian and B(ν) analyses suggest that an az-
imuthal magnetic field is embedded within this starburst
ring such that the field points towards us at the south-
ernmost tangent point and away from us at the north-
ernmost tangent point. Parra et al. (2005) estimate that
the OHM clouds would be magnetically confined by a
magnetic field of order ∼ 10 mG.
Killeen et al. (1996) used the ATCA to observe the
OHM emission in III Zw 35. Their sensivity of Srms = 3.6
mJy in Stokes V was not sufficient to detect the Zeeman
splitting of the 1667 MHz line.
5.1.2. IRAS F10038−3338 (IC 2545)
As the residuals in Figure 2 show, the 1667 MHz OHM
emission from IC 2545 is fitted extremely well by 5 nar-
row Gaussian components and one broad one. Using our
prescription from § 4.4, we see that there are 4 distinct
peaks. The peak nearest 1613.1 MHz has an asymmetry
that can be represented with a single extra narrow com-
ponent. A broad component represents the evident core-
halo structure of the OHM emission profile. It is unclear
if the emission feature at 1611.7 MHz corresponds to 1665
MHz emission or redshifted 1667 MHz emission. The
Stokes I flux density and line profile have not changed
since the observations of Killeen et al. (1996). Table 2
lists the Gaussian fit parameters shown in Figure 4. The
Stokes V spectrum has an rms noise of Srms = 0.7 mJy.
There are three clear detections: Gaussian 1 probes a
field of −1.8 ± 0.3 mG; Gaussian 2 is fitted by a field
of −11.3 ± 1.2 mG; and Gaussian 5 shows a reversal in
sign with a field of 1.7± 0.3 mG. Since no VLBI observa-
tions exist for this LIRG, nothing can be said about the
structure of the field reversal.
5.1.3. IRAS F10173+0829
As mentioned in § 4.2, we used Hanning-smoothed
spectra when least-squares fitting this source because of
RFI. We increased the derived uncertainties in Table 3
by the appropriate factor of p8/3 (Killeen et al. 1996,
Table A1).
We fit Stokes I with 7 Gaussians as shown in Figure
3 and listed in Table 3. Using the selection guidelines of
§ 4.4, we required 4 narrow components to sufficiently fit
the extremely asymmetric peak near 1589.3 MHz. The
fit to Stokes I yielded reasonable residuals by including
a single extremely broad component. This source repre-
sents a case where the profile structure is too complex
to be modelled by our straightforward fitting guidelines:
we regard the derived components as highly suspect re-
gardless of the quality of the fit.
Detectable portions of the Stokes V spectrum are re-
stricted to the strongly peaked I line, where V oscillates
rapidly. Due to the asymmetry in this profile, we found
it impossible to obtain a set of Gaussians that gives good
fits to both I and V . Our final fit reflects a compromise
between large residuals and a larger number of Gaus-
sians. While it is clear that Zeeman splitting has been
detected in this source, we would need to add many more
narrow, weak Gaussian components to our Stokes I fit
in order to obtain statistically significant magnetic field
derivations from the fit to Stokes V (which has an rms
noise of Srms = 0.8 mJy). Without any VLBI observa-
tions or any other physical motivation for adding such
components, the best we can do is present the evidence
for Zeeman splitting without estimating field strengths
for individual maser components.
Figure 3, right panel, shows plots relevant to the B(ν)
fit. The dashed line in the middle panel shows the Stokes
V spectrum that would be expected for a uniform Bk = 2
mG. The bottom panel shows the B(ν) fit. There is
a clear systematic pattern, with the field reversing sign
from one side of the peak to the other. This reversal is
not revealed by the Gaussian fits because there are no
narrow Gaussians on either side of the peak and because
the peak itself is represented by a single Gaussian (com-
ponent 2).
Since no spectral information is presented in the MER-
LIN maps of Yu (2004, 2005), it is impossible to asso-
ciate any of our Gaussian components with OHM spots
in 10173.
5.1.4. IRAS F11506−3851
The top panel of Figure 4 shows that the 1667 MHz
Stokes I emission is fitted quite well by 6 narrow Gaus-
sian components. The parameters for each component
are listed in Table 4. Gaussian components 0 and 3 have
derived magnetic fields that look significant when judged
by their formal errors. However, given the quality of the
Stokes V spectrum (Srms = 0.4 mJy), we have no confi-
dence in either result. The 1665 MHz emission is clearly
completely separated from the 1667 MHz emission, yield-
ing a hyperfine ratio RH = 4.9.
5.1.5. IRAS F12032+1707
The OHM emission from this gigamaser has an ex-
tremely wide extent; it is impossible to distinguish the
1665 and 1667 MHz emission. We fit its Stokes I spec-
8
Robishaw, Quataert, & Heiles
trum with 13 Gaussians as shown in Figure 5 with pa-
rameters as listed in Table 5. Due to the complexity of
this profile, this is an extremely difficult case to apply our
component selection guidelines to. We first identify 8 lo-
cal maxima marked in Figure 5 by components 0, 2, 3, 5,
6, 8, 10, and 12. The peaks near 8 and 10 display shoul-
ders and are clearly blended with narrower components:
we added one additional component to each (components
7 and 9, respectively). We represent a broad shoulder,
not quite intense enough to produce a local maximum,
near 1373 MHz by component 11. Finally, we add two
broad halo components, numbers 1 and 4, to minimize
the residuals. While the prescription of § 4.4 allows us to
select these 13 components somewhat straightforwardly
-- even producing respectable residuals -- visual inspec-
tion of Figure 5 suggests that our model simplifies and
glosses over the innate complexity of this OHM profile.
Without VLBI observations, there is little that can be
done to improve upon our model.
Detectable signals in the V spectrum (Srms = 0.9 mJy)
are restricted to the stronger of the two narrow peaks.
This narrow peak is somewhat asymmetric and requires
two Gaussians for a good fit. These two components
(numbers 7 and 8) have field strengths of 10.9 ± 1.7 and
17.9±0.9 mG, respectively. Three other Gaussians, num-
bers 1, 6, and 11, have fields that are nearly 3σ; however,
visual examination of the V spectrum shows bumps and
wiggles throughout at the ≈ 1 mJy level, and we have
no confidence in these purported fields.
Since the significant splittings occur where the Stokes
I profile has the highest flux density, and since the 1667
MHz transition dominates in all OHMs, we feel comfort-
able assuming that the emission is from the 1667 MHz
transition. Of course, because of the blending of the 1665
and 1667 MHz lines, there is an unresolvable ambiguity
that could affect the derived field strengths.
The Stokes I line profile has changed since the source's
discovery by Darling & Giovanelli (2001). The flux den-
sity of the broad component and the narrow component
near 1372.3 MHz have remained the same. However, the
narrow component near 1371.3 MHz, which used to be
roughly 6 mJy (using our classical definition of Stokes
I) weaker than their 32.5 mJy peak at 1372.3 MHz, has
flared and is now the strongest component with a flux
density of 44 mJy. This time-variable component is the
same one that exhibits the largest splitting and there-
fore probes the strongest field; in § 6 we compare this re-
sult with newly-observed strong field detections in time-
variable Galactic OH maser components.
5.1.6. IRAS F12112+0305
We were able to fit the Stokes I OHM emission quite
nicely with 5 Gaussians as shown in Figure 6. The fit
parameters are listed in Table 6. There is no detectable
signal in Stokes V (Srms = 1.2 mJy). This is the first
published spectrum of OHM emission in 12112.
5.1.7. IRAS F14070+0525
Table 7 lists the parameters for the 7 Gaussian com-
ponents used to fit the Stokes I OHM emission in 14070.
Since the 1665 and 1667 MHz lines are clearly blended
in this source, we assume that all of the components rep-
resent 1667 MHz emission. As seen in Figure 7, this
decomposition provides a decent fit but there is no de-
tectable signal in Stokes V (Srms = 0.5 mJy). Gaussian
number 1 shows a nearly 3σ detection of magnetic field;
however, the associated feature in the Stokes V spectrum
appears to be no more significant than the other features
of mJy-strength intensity. We have no confidence in this
near detection.
5.1.8. IRAS F15327+2340 (Arp 220)
The line profile of the Stokes I OHM emission in Arp
220 is very complex. Our fit required 18 Gaussian com-
ponents, as seen in Figure 8, to obtain reasonable resid-
uals. An 18-component fit might seem overwhelming,
but the components are easily obtained using our selec-
tion guidelines from § 4.4. There are 9 distinct peaks
(i.e., local maxima), represented in Figure 8 by Gaussian
components 0, 4, 5, 7, 9, 13, 14, 16, and 17. There are 6
narrow or fairly narrow bumps or shoulders that are not
intense enough to produce local maxima, represented by
Gaussian components 1, 2, 3, 11, 12, and 15. Compo-
nent 10 was needed to represent the asymmetry in the
brightest peak near 1638.15 MHz. Finally, components
6 and 8 were needed to represent core-halo structure in
the overall profile. This 18-component fit reproduces all
of the visually obvious narrow, weak bumps, as well as
the overall profile shape. However, the residuals exhibit
a different signature in the line from that off the line,
which means that our fit does not represent the I profile
perfectly. We expended considerable effort making sure
that each of the 18 Gaussians listed in Table 8 is actually
needed for the fit by inspecting the residuals for different
combinations of omitted Gaussians.
Six narrow Gaussians exhibit visually obvious signa-
tures in Stokes V (which has an rms noise of Srms = 1.16
mJy) and provide good fits for Zeeman splitting. The ab-
solute values of the derived field strengths range from 0.7
to 4.7 mG, with 4 negative and 2 positive fields. Gaus-
sians 2 and 3 have opposite field strengths.
Gaussian numbers 9 and 11 are strong (a few hundred
mJy), have comparable FWHMs of about 0.1 MHz, and
are separated by about the FWHM. This makes them
easily distinguishable. They have opposite field direc-
tions as given by the least-squares fit, but the reversal
in sign is also visually apparent. Zeeman splitting pro-
duces a Stokes V pattern that looks like the frequency
derivative of the line, with amplitude and sign scaled
by Bk. Thus, for a single Gaussian component, the V
pattern looks like the letter "S" lying on its side, with
an inevitable negative and positive part; the integral over
Stokes V must be zero. However, the V pattern for these
Gaussians in Figure 8 doesn't look like this; instead it is
positive on both sides of the line and negative in the mid-
dle. The only way to obtain positive V on each side of a
spectral bump is for the field to have different signs on
the two sides (cf. Verschuur 1969, Figure 3, the first radio
detection of Zeeman splitting in Cas A). The integral of
Stokes V must again be zero, with the central negative
portion balanced by the two positive ones on the sides.
The reversed field is not only a result of the fits, but is
also visually apparent.
We can compare our single-dish spectrum with the se-
lected global VLBI spectra presented by Rovilos et al.
(2003) and Lonsdale et al. (1998). There are a number
of Gaussian components that appear to be directly asso-
Extragalactic Zeeman Detections in OH Megamasers
9
ciated with the resolved OHM spots: component 11 at
5334 km s−1 originates in a southwestern spot tracing a
positive field; component 6 at 5393 km s−1 originates in
the southeast and traces a positive field; component 4
at 5425 km s−1 originates in the center of the northeast
ridge and traces a negative field; component 0 at 5533 km
s−1 originates in one of the southwestern spots tracing
a negative field. Three other features are more ambigu-
ous: components 2 and 3 could be associated with either
the northwestern or northeastern OHM ridges, while the
brightest component, number 9 at 5351 km s−1, appears
to contain emission from both the northern ridges as well
as the southwestern maser spots; these ambiguities pre-
vent any possible field associations. The picture painted
by the possible associations is for a field reversal from
positive to negative from the southern to the northern
features of the eastern OHM spots; there is no obvious
reversal in the western region, but it is possible given the
associations above.
5.2. Linear Polarization
For all observations, we used dual-polarized feeds with
native linear polarization. This means that the observed
Stokes Uobs and Vobs come from cross-correlation prod-
ucts, which insulates them from system gain fluctuations.
However, Stokes Qobs comes from the difference between
the two native linear polarizations, so it is susceptible
to time-variable, unpredictable gain fluctuations. This
leads to coupling between Stokes I and Stokes Qobs; in
other words, a scaled replica of the I profile appears in
the Qobs profile, with a random and unknown scaling
factor, so Qobs is unreliable.
Normally, when deriving linear polarization, one com-
bines Stokes Qobs and Uobs in the standard ways to obtain
polarized intensity and position angle for the astronomi-
cal source. However, since Qobs is unreliable for our mea-
surements, we derived Stokes Qsrc and Usrc for the source
from Uobs alone by least-squares fitting its variation with
parallactic angle. This is quite feasible at Arecibo be-
cause all sources pass within 20◦ of the zenith, so track-
ing for a reasonably long time provides a wide spread in
parallactic angle. This makes the least-squares fit robust
and provides good sensitivity and low systematics. For
the source III Zw 35, which was plagued by serious inter-
ference, we performed a minimum-absolute-residual-sum
(MARS) fit.
As with the Stokes I and V spectra, the least-squares
derived Qsrc and Usrc spectra are displayed after sub-
tracting both the off-source position and the continuum.
We then use these baseline-subtracted Stokes spectra to
derive spectra for polarized intensity and position angle.
We do this because, even for the position-switched spec-
tra, the continuum linear polarization is usually domi-
nated by the diffuse Galactic synchrotron background.
Although this prevents us from reliably deriving linear
polarization for the (U)LIRG continuum radiation, the
frequency-variable polarization is reliable.
For two sources below, we least-squares fit for the Fara-
day rotation measure RM . Performing this fit requires
some care because the RM is derived from the position
angle ψ, which in turn is obtained by combining Qsrc
and Usrc, which combine nonlinearly through the arctan
function [ψ = 0.5 arctan(Qsrc/Usrc)]. The channel-by-
channel data are too noisy to produce a good-looking
spectral plot of ψ, so on our plots we boxcar smooth by
an appropriate number of points. One cannot linearly
fit the unsmoothed values of ψ to frequency because the
arctan function produces nonlinear noise in ψ. To avoid
this problem, we performed a nonlinear fit to the un-
smoothed arctan(Qsrc/Usrc); this extra complication en-
sures that the derived values and errors are unaffected by
smoothing.
We were unable to analyze the linear polarization for
the two sources we observed using the GBT (IC 2545 and
11506) because of inadequate parallactic angle coverage.
We report the linear polarization for the Arecibo results
here and discuss their interpretation in § 6.3.
We present three vertically stacked plots for each
source below (Figures 9 -- 14). The top panel shows the
position-differenced, baseline-subtracted Stokes I spec-
trum over 12.5 MHz, therefore including both the 1665
and 1667 MHz transitions for each source. The middle
panel presents the linear polarization intensity and the
bottom panel displays the derived position angle ψ as a
function of heliocentric frequency.
5.2.1. IRAS F01417+1651 (III Zw 35)
The linear polarization results for III Zw 35 are pre-
sented in Figure 9. The top panels exhibit the position-
differenced, baseline-subtracted Stokes I spectrum. The
1665 MHz transition is clearly visible and the hyperfine
line ratio is RH = 6.0. The middle panels clearly display
that the spectrum of linearly polarized intensity is ex-
tremely spikey. Although higher S/N would help, the
spikes might be real and possibly correspond to indi-
vidual masers that are too weak to be seen clearly in
the Stokes I spectrum. The polarized intensity shows a
seemingly real peak centered near 1622.8 MHz, which is
also the center of the Stokes I peak. The polarized in-
tensity is about 5 mJy and the Stokes I peak is roughly
If the
500 mJy, so the fractional polarization ≈ 1%.
other spikes are real, then their fractional polarizations
are much higher.
The bottom panels display the position angle ψ. Posi-
tion angles exhibit less scatter than intensities, and the
angle looks well-defined for the 1622.8 MHz peak. Also,
it seems to show a gradual change across the line, which
is about 1 MHz wide. The dashed line displays the result
of a least-squares fit to the frequency dependence of the
position angle, using only those points that are marked
as diamonds: RM = −21900 ± 3700 rad m−2. The ex-
trapolated dashed line goes through the clusters of points
associated with spikes centered near 1623.7 and 1624.0,
and moreover, even the slope of the line matches the
data for these spikes. The slope also seems to match the
1624.4 MHz polarized-intensity spike, but the data are
offset by about 60◦. We speculate that: (1) these three
polarized-intensity spikes come from individual OHMs
that are too weak to see in the top panels of Figure 9; (2)
they all suffer the same Faraday rotation of ≃ −21900
rad m−2 as the central peak; and (3) the intrinsic po-
sition angle for the 1624.4 MHz maser differs from the
other two by about 60◦.
5.2.2. IRAS F10173+0829
Figure 10 displays the linear polarization results for
10173. The polarized intensity shows a low-S/N spike
that is centered on the Stokes I line: the polarization
10
Robishaw, Quataert, & Heiles
fraction is about 1% and the position angle about 60◦.
The spike is too narrow to fit for Faraday rotation. The
hyperfine line ratio for 10173 is RH = 10.7.
OHM radiation from both OH lines suffers a common
Faraday rotation of RM ≈ 1250 rad m−2; this is ∼ 20
times smaller than the value derived for III Zw 35.
5.2.3. IRAS F12032+1707
The linear polarization results for 12032 are shown in
Figure 11. The polarized intensity shows multiple spikes
that might be real. The most significant is centered at
≃ 1372 MHz, with a peak flux density of ≃ 5 mJy, and
has ψ ≃ 15◦; near this frequency, I varies from ≃ 10
to ≃ 40 mJy, so if this peak is real then the fractional
polarization is huge, ≈ 50% to ≈ 10% -- unheard of for
OH masers of any stripe.
5.2.4. IRAS F12112+0305
Figure 12 shows the linear polarization results for
12112. The lower frequencies are plagued by RFI, which
remarkably disappears at the low-frequency boundary of
the 1667 MHz line (centered at 1554.5 MHz). Accord-
ing to the National Telecommunications and Information
Administration Manual of Regulations and Procedures
for Federal Radio Frequency Management, this RFI is
likely attributable to space-to-Earth aeronautical mobile
satellite communications operated by Inmarsat. There
is no trace of any detectable linear polarization for this
source. This is the first detection of the 1665 MHz tran-
sition for 12112; the hyperfine line ratio is RH = 4.0.
5.2.5. IRAS F14070+0525
Figure 13 displays the linear polarization results for
14070. The linear polarization intensity is approximately
4 mJy across the entire 12.5 MHz bandwidth with an
estimated position angle of −48◦.
5.2.6. IRAS F15327+2340 (Arp 220)
The top panels of Figure 14 show the Stokes I profile
for Arp 220 including both the 1665 and 1667 MHz tran-
sitions. The hyperfine line ratio is RH = 3.5. The middle
panels show the linear polarization intensity, which has
a well-defined peak centered at 1638 MHz and peaks at
about 2 mJy. This is only ≈ 0.3% of the total intensity
at this frequency. This is a very small fractional polar-
ization but is very well-detected.
The bottom panels show that the position angle of
linear polarization is well-defined in two regions of low
noise, one centered near 1638 MHz and the other near
1636 MHz. The former region corresponds to the 1667
MHz line and the latter is aligned with the 1665 MHz
transition. This 1665 MHz line is unconvincingly visible
in the polarized intensity spectrum, but the low noise in
its position angle spectrum is unmistakable.
We fit the frequency variation of ψ to obtain the Fara-
day rotation measure RM using those points marked
as diamonds in the bottom right panel of Figure 14.
For the 1638 MHz component alone, we obtain RM =
5230 ± 7930 rad m−2. For the combination of the 1636
and 1638 MHz components, we obtain RM = 1250±1040
rad m−2. These errors are considerable and make the for-
mal result only marginally significant. The dashed line
in the bottom-right panel displays the result of the fit
for both components together; visual inspection shows
that not only is it an acceptable fit for both components
together, but it is also acceptable for the 1638 MHz com-
ponent alone. It is not unreasonable to conclude that the
6. DISCUSSION
6.1. OH Maser Zeeman Pairs in the Milky Way
While our results are the very first in situ Zeeman
detections in external galaxies, OH masers in the MW
have been used as Zeeman magnetometers for well over
a decade. In contrast to OHMs, Galactic OH maser emis-
sion lines are so narrow (∼ 0.5 km s−1) that fields of ≈ 1
mG are sufficient to completely split the left and right
circular σ components into pairs. More than 100 of these
Zeeman pairs have been compiled by Fish et al. (2003)
and Reid & Silverstein (1990) with a distribution whose
mean is consistent with 0 mG and whose standard de-
viation is 3.31±0.09 mG. Typical densities in OH maser
regions are n ∼ 106 − 107 cm−3; for a field strength of
∼ 10 µG in gas at ∼ 1 − 100 cm−3, the fields probed by
Galactic OH masers are consistent with the enhancement
of B ∝ n1/2 (Fish et al. 2003). The linear polarization
of the σ components is often measured in addition to the
π component, but the π components, which are in the-
ory 100% linearly polarized, are rarely measured to be
purely so.
Unlike in the OH masers in our Galaxy, the flux density
of the 1667 MHz transition in all OHMs is larger than
that of the 1665 MHz transition and, until now, no polar-
ization has been detected (Lo 2005). There is no defini-
tive explanation for the dominance of the 1667 MHz tran-
sition, but recent work suggests that this probably arises
because the extragalactic lines are wider than the Galac-
tic maser lines (P. Goldreich, private communication;
M. Elitzur, private communication; Lockett & Elitzur
2008).
Our detections yield a median line-of-sight magnetic
field strength of ≃ 3 mG in OHMs in (U)LIRGs, which is
comparable to the field strengths measured in OH masing
regions in the MW. This strongly suggests that the local
process of massive star formation occurs under similar
conditions in (U)LIRGs, galaxies with vastly different
large-scale environments than our own.
The magnetic field strengths we find in the OHMs
in (U)LIRGs (∼ 3 mG) are comparable to the volume-
averaged fields of & 1 mG inferred from synchrotron ob-
servations. These results imply that mG magnetic fields
likely pervade most phases of the interstellar medium in
(U)LIRGs. It is unclear, however, how to physically re-
late the two different magnetic field strengths in more
detail given the possibility that each may probe rather
different phases of the ISM. Some models of OHMs invoke
radiative pumping in molecular clouds with gas densities
∼ 103.5 − 104 cm−3 (e.g., Randell et al. 1995). This is
similar to the mean gas density in the central ∼ 100 pc
in (U)LIRGs, in which case our observations likely probe
the mean ISM magnetic field (whether the synchroton ra-
diation also arises from gas at this density is unclear; up-
coming GLAST observations of neutral pion decay may
help assess this; see Thompson et al. 2007). It is also pos-
sible, however, that the OHMs arise in somewhat denser
gas (n ∼ 106 − 107 cm−3; e.g., Lonsdale et al. 1998), as
appears to be true in the MW (e.g., Fish et al. 2003). In
this case, the magnetic field probed by OHMs is likely
Extragalactic Zeeman Detections in OH Megamasers
11
stronger than that in the bulk of the ISM. If we as-
sume the B ∝ n1/2 scaling often assumed in the MW
(Mouschovias 1976; Fish et al. 2003), the field strengths
in the masing regions in (U)LIRGs are probably within
a factor of ∼ 3 of the mean ISM field (rather than a
factor of several 100 in the MW), given the large mean
gas densities in (U)LIRGs. This is still reasonably con-
sistent with the mean field strength of & 1 mG inferred
from synchrotron observations. Without a better under-
standing of the physical conditions in the masing regions,
however, it is difficult to provide a more quantitative
connection between our inferred field strengths and ei-
ther the mean ISM field or the magnetic field probed by
synchrotron emission. Ultimately doing so is important
because it will allow stringent constraints to be placed
on the dynamical importance of magnetic fields across a
wide range of physical conditions in (U)LIRGs.
Fish et al. (2003), with their comprehensive survey
of Galactic OH masers and the accompanying statisti-
cal discussion, strongly support several previous sugges-
tions that the field direction in OH masers usually mir-
rors that of the large-scale field in the vicinity of the
masers. MERLIN observations of OH masers in Cep A
by Bartkiewicz et al. (2005) also present Zeeman detec-
tions corroborating the field's alignment with the ambi-
ent ISM field direction. Thus, measuring the direction
of the field in an OH maser reveals the field direction
not only in the maser, but also outside and in the vicin-
ity of the OH maser. For the MW, this aids us to infer
the large-scale magnetic field morphology. To directly
compare the star formation processes in the MW and
(U)LIRGs, it will be necessary to increase the sample
of magnetic field strengths in (U)LIRGs and to directly
map the Zeeman splitting of individual OHM spots us-
ing VLBI in order to probe whether reversals occur at
smaller angular scales.
6.2. Strong Fields and Time Variability
Slysh & Migenes (2006) and Fish & Reid (2007) both
observed fields of 40 mG using Zeeman observations of
OH maser spots in W75N; these are the highest field
strengths measured in Galactic OH masers and are an
order of magnitude larger than the typical OH maser
field. These OH maser spots also happen to have been
flaring based on multi-epoch VLBA observations; per-
haps time variability in OH masers is correlated with
strong magnetic fields.
Interestingly, our strongest de-
tection, Bk ∼ 18 mG in the gigamaser 12032, occurs in
an OHM component that has increased in flux density by
a factor of two since its previous published observation
(Darling & Giovanelli 2001).
These results strongly support the development of an
observational program to monitor both the time vari-
ability of the Stokes I flux density and magnetic field
strength in OHMs as well as the necessity of observing
the circular polarization of time-variable Galactic masers
in hopes of detecting strong magnetic fields.
6.3. Linear Polarization and Faraday Rotation
Our measured rotation measures of RM ≃ 21900 rad
m−2 for III Zw 35 and RM ≃ 1250 rad m−2 for Arp
220 are large by most standards, but are not unreason-
able for (U)LIRGs. As mentioned in § 1, the magnetic
field strength throughout the ULIRG ISM should be & 1
mG from synchrotron observations. Electron densities
are estimated to be ∼ 1 − 10 cm−3 in the hot ion-
ized plasma, both from observations of X-ray emission
(e.g., Grimes et al. 2005) and from theoretical models of
supernova-driven galactic winds (e.g., Chevalier & Clegg
1985). Over a path length ∼ 100 pc in the central por-
tions of ULIRGs, ne ∼ 1 cm−3 and B ∼ 1 mG imply
hneBkLi ∼ 0.1 G cm−3 pc, or RM ∼ 80000 radians
m−2. This is a factor of 4 -- 60 larger than our measured
values.
It is reasonable for this simple estimate to overestimate
the measured RM . This is because the RM depends
only on the line-of-sight field component. The proba-
bility density function for the line-of-sight component of
a randomly oriented magnetic field is flat between zero
and the perfectly-oriented case; thus, for a set of sources
with randomly-oriented fields, the observed line-of-sight
field component is reduced by a factor of two, and 1/4
of the sources have the observed component less than
1/4 the perfectly-aligned value. In addition, and prob-
ably more importantly, the observed Faraday rotation
responds only to the systematic line-of-sight field com-
ponent, while the synchrotron radiation and the total
magnetic energy depend on the total field, systematic
plus random. Our estimate of RM ∼ 80000 is for the
total field, not the systematic field, because the latter is
much harder to predict.
The measured RM might also be reduced by finite
source-size effects and/or propagation through an inho-
mogenous medium (Burn 1966). First, suppose that the
magnetic field is everywhere uniform but that the Fara-
day rotation is produced in the same region where the
maser radiation is produced, and that this region is ex-
tended along the line of sight.
In this case, different
line-of-sight depths of the maser are rotated by different
amounts. This washes out the linear polarization and
can reduce the apparent Faraday rotation. In the other
extreme, think of the field as primarily random except for
a small uniform component. Maser radiation observed at
a given frequency might come from more than one maser
located at different positions on the sky or at different
distances into the source. In the former case, the RM
might change with position on the sky; in the latter, it
might change along the line of sight. In either case, its
average value can be small. In addition, for an individ-
ual maser the field might fluctuate along the line of sight,
reducing the total RM .
The interpretation of the linear polarization and RM
data is thus currently difficult and non-unique. Obser-
vations of more systems would be helpful and may ulti-
mately provide unique constraints on the thermal elec-
tron density and/or magnetic field structure (e.g., rever-
sals) in the nuclei of (U)LIRGs.
It is a pleasure to acknowledge Phil Perillat, who per-
formed the Mueller matrix calibration observations and
reductions, measured the antenna gain, and wrote the
online data acquisition software at Arecibo. We thank
Karen O'Neil, Amy Shelton, and Mark Clark for helping
us institute LSFS observing at the GBT. This research
benefited from helpful discussions with Jeremy Darling,
Vincent Fish, Peter Goldreich, Bill Watson, Fred Lo,
12
Robishaw, Quataert, & Heiles
Moshe Elitzur, Willem Baan, and Loris Magnani. TR
appreciates the technical guidance of grammarian Elena
Cotto. This research was supported in part by NSF grant
AST-0406987. Support for this work was also provided
by the NSF to TR through awards GSSP 05-0001, 05-
0004, and 06-0003, from the NRAO. EQ was supported in
part by NASA grant NNG06GI68G and the David & Lu-
cile Packard Foundation. This research has made use of
NASA's Astrophysics Data System Abstract Service and
the SIMBAD database, operated at CDS, Strasbourg,
France.
Facilities: Arecibo, GBT
REFERENCES
Baan, W. A., Rhoads, J., Fisher, K., Altschuler, D. R., &
Haschick, A. 1992, ApJ, 396, L99
Baan, W. A., Wood, P. A. D., & Haschick, A. D. 1982, ApJ, 260,
Parra, R., Conway, J. E., Diamond, P. J., Thrall, H., Lonsdale,
C. J., Lonsdale, C. J., & Smith, H. E. 2007, ApJ, 659, 314
Parra, R., Conway, J. E., Elitzur, M., & Pihlstrom, Y. M. 2005,
L49
Bartkiewicz, A., Szymczak, M., Cohen, R. J., & Richards,
A. M. S. 2005, MNRAS, 361, 623
Burn, B. J. 1966, MNRAS, 133, 67
Chevalier, R. A., & Clegg, A. W. 1985, Nature, 317, 44
Coles, W. A., & Rumsey, V. H. 1970, ApJ, 159, 247
Condon, J. J., Huang, Z.-P., Yin, Q. F., & Thuan, T. X. 1991,
ApJ, 378, 65
Darling, J., & Giovanelli, R. 2000, AJ, 119, 3003
-- 2001, AJ, 121, 1278
-- 2002, AJ, 124, 100
Diamond, P. J., Lonsdale, C. J., Lonsdale, C. J., & Smith, H. E.
1999, ApJ, 511, 178
Downes, D., & Solomon, P. M. 1998, ApJ, 507, 615
Fish, V. L., & Reid, M. J. 2007, ApJ, 656, 952
Fish, V. L., Reid, M. J., Argon, A. L., & Menten, K. M. 2003,
Graham, J. R., Carico, D. P., Matthews, K., Neugebauer, G.,
Soifer, B. T., & Wilson, T. D. 1990, ApJ, 354, L5
Grimes, J. P., Heckman, T., Strickland, D., & Ptak, A. 2005,
ApJ, 596, 328
ApJ, 628, 187
Heiles, C. 2007, PASP, 119, 643
Heiles, C., Goodman, A. A., McKee, C. F., & Zweibel, E. G.
1993, in Protostars and Planets III, ed. E. H. Levy & J. I.
Lunine (Tucson: Univ. Arizona Press), 279
Heiles, C., et al. 2001, PASP, 113, 1274
Heiles, C., & Troland, T. H. 2004, ApJS, 151, 271
Kazes, I., Troland, T. H., & Crutcher, R. M. 1991, A&A, 245, L17
Killeen, N. E. B., Staveley-Smith, L., Wilson, W. E., & Sault,
R. J. 1996, MNRAS, 280, 1143
Lo, K. Y. 2005, ARA&A, 43, 625
Lockett, P., & Elitzur, M. 2008, ApJ, in press
(astro-ph/0801.2937)
Lonsdale, C. J., Diamond, P. J., Thrall, H., Smith, H. E., &
Lonsdale, C. J. 2006, ApJ, 647, 185
Lonsdale, C. J., Lonsdale, C. J., Diamond, P. J., & Smith, H. E.
1998, ApJ, 493, L13
Mirabel, I. F., & Sanders, D. B. 1987, ApJ, 322, 688
Modjaz, M., Moran, J. M., Kondratko, P. T., & Greenhill, L. J.
2005, ApJ, 626, 104
Mouschovias, T. C. 1976, ApJ, 207, 141
Pihlstrom, Y. M., Baan, W. A., Darling, J., & Klockner, H.-R.
A&A, 443, 383
2005, ApJ, 618, 705
Pihlstrom, Y. M., Conway, J. E., Booth, R. S., Diamond, P. J., &
Polatidis, A. G. 2001, A&A, 377, 413
Press, W. H., Teukolsky, S. A., Vetterling, W. T., & Flannery,
B. P. 1992, Numerical Recipes in FORTRAN: The Art of
Scientific Computing (2nd ed.; Cambridge: Cambridge Univ.
Press)
Randell, J., Field, D., Jones, K. N., Yates, J. A., & Gray, M. D.
1995, A&A, 300, 659
Reid, M. J., & Moran, J. M. 1988, in Galactic and Extragalactic
Radio Astronomy, ed. G. L. Verschuur & K. I. Kellermann (2nd
ed.; Berlin: Springer), 255
Reid, M. J., & Silverstein, E. M. 1990, ApJ, 361, 483
Rogers, A. E. E., Moran, J. M., Crowther, P. P., Burke, B. F.,
Meeks, M. L., Ball, J. A., & Hyde, G. M. 1967, ApJ, 147, 369
Rovilos, E., Diamond, P. J., Lonsdale, C. J., Lonsdale, C. J., &
Smith, H. E. 2003, MNRAS, 342, 373
Sarma, A. P., Momjian, E., Troland, T. H., & Crutcher, R. M.
Sault, R. J., Killeen, N. E. B., Zmuidzinas, J., & Loushin, R.
Slysh, V. I., & Migenes, V. 2006, MNRAS, 369, 1497
Smith, H. E., Lonsdale, C. J., Lonsdale, C. J., & Diamond, P. J.
2005, AJ, 130, 2566
1990, ApJS, 74, 437
1998, ApJ, 493, L17
Staveley-Smith, L., Cohen, R. J., Chapman, J. M., Pointon, L., &
Unger, S. W. 1987, MNRAS, 226, 689
Staveley-Smith, L., Norris, R. P., Chapman, J. M., Allen, D. A.,
Whiteoak, J. B., & Roy, A. L. 1992, MNRAS, 258, 725
Thompson, T. A., Quataert, E., & Waxman, E. 2007, ApJ, 654,
219
Thompson, T. A., Quataert, E., Waxman, E., Murray, N., &
Martin, C. L. 2006, ApJ, 645, 186
Veilleux, S., Kim, D.-C., & Sanders, D. B. 1999, ApJ, 522, 113
Verschuur, G. L. 1969, ApJ, 156, 861
Yu, Z.-Y. 2004, ChA&A, 28, 287
-- 2005, ChJA&A, 5, 159
Extragalactic Zeeman Detections in OH Megamasers
13
IRAS F01417+1651 (III Zw 35) Gaussian Fit Parameters
TABLE 1
Gaussian
(1)
0 . . . . . . .
1 . . . . . . .
2 . . . . . . .
3 . . . . . . .
4 . . . . . . .
S
(mJy)
(2)
ν
(MHz)
(3)
∆ν
(MHz)
(4)
v⊙
(km s−1)
(5)
Bk
(mG)
(6)
1622.3743 ± 0.0049 0.2041 ± 0.0081
190.71 ± 6.77
255.68 ± 22.16 1622.6881 ± 0.0277 0.2786 ± 0.0514
99.26 ± 6.28
1622.7237 ± 0.0103 0.9694 ± 0.0295
176.13 ± 34.75 1622.7604 ± 0.0020 0.0905 ± 0.0080
1.17 ± 59.50 1622.8864 ± 0.0142 0.1544 ± 0.0239
8312.6
8253.0
8246.2
8239.3
8215.3
2.94 ± 0.18
−0.47 ± 0.18
1.73 ± 0.78
−2.73 ± 0.13
−3.59 ± 0.26
Heliocentric Velocity (km s−1)
8400
8300
8200
8100
01417+1651
46 mJy
500
400
300
200
100
)
y
J
m
(
I
Heliocentric Velocity (km s−1)
8400
8300
8200
8100
1621.5
1622.0
1622.5
0
1623.0
123 4
1623.5
1624.0
1621.5
1622.0
1622.5
1623.0
1623.5
1624.0
)
y
J
m
(
V
5
0
−5
1621.5
1622.0
1622.5
1623.0
1623.5
1624.0
500
400
300
200
100
0
)
y
J
m
(
I
)
y
J
m
(
V
10
5
0
−5
−10
1621.5
1622.0
3
2
1
0
−1
−2
)
G
m
(
B
1623.5
1624.0
−3
1621.5
1622.0
1622.5
1623.0
Heliocentric Frequency (MHz)
1622.5
1623.0
Heliocentric Frequency (km s−1)
1623.5
1624.0
Fig. 1. -- Total intensity and circular polarization results for IRAS F01417+1651 (III Zw 35). (Top left) The Stokes I spectrum (solid
line; twice the conventionally defined flux density) is plotted as a function of heliocentric frequency (bottom axis) and optical heliocentric
velocity (top axis). The profile of each Gaussian component is plotted as a dashed line with the corresponding component number shown
below the frequency axis at the corresponding central frequency. Residuals from the composite Gaussian fit are plotted through the center
of the panel (solid, light-weight line) and are expanded by a factor of 4. The scale bars near the right edge of the plot correspond to the
labelled flux density range. (Bottom left) The Stokes V spectrum (solid line) and its fit (dashed line). (Top right) Composite Gaussian fit
to Stokes I. (Middle right) Measured Stokes V (solid line); hypothetical Stokes V (dashed line) produced by a uniform Bk = 1 mG using
the derivative of the composite I profile above. (Bottom right) Derived Bk (crosses and solid line) and uncertainty (error bars) from the
B(ν) fit. All spectra and residuals are smoothed with a boxcar of 7 channels.
14
Robishaw, Quataert, & Heiles
IRAS F10038−3338 (IC 2545) Gaussian Fit Parameters
TABLE 2
S
(mJy)
(2)
ν
(MHz)
(3)
∆ν
(MHz)
(4)
v⊙
(km s−1)
(5)
Bk
(mG)
(6)
1612.3870 ± 0.0044 0.1505 ± 0.0125
47.06 ± 3.10
179.73 ± 5.15
1612.8661 ± 0.0032 0.1747 ± 0.0077
83.85 ± 5.50
1613.0090 ± 0.0119 0.9548 ± 0.0405
227.46 ± 36.43 1613.0362 ± 0.0052 0.1222 ± 0.0095
2.85 ± 14.04 1613.1570 ± 0.0089 0.1703 ± 0.0157
1613.3556 ± 0.0035 0.1136 ± 0.0075
97.85 ± 4.75
1.58 ± 0.75
10221.0
− 1.76 ± 0.26
10128.9
10101.4 −11.28 ± 1.16
10096.2
− 0.07 ± 0.18
10073.0
− 0.13 ± 0.16
1.67 ± 0.33
10034.9
Gaussian
(1)
0 . . . . . . .
1 . . . . . . .
2 . . . . . . .
3 . . . . . . .
4 . . . . . . .
5 . . . . . . .
Extragalactic Zeeman Detections in OH Megamasers
15
10400
Heliocentric Velocity (km s−1)
10200
10000
9800
10038−3338
50 mJy
1612
0
1613
1 2
3 4 5
1614
1615
1612
1613
Heliocentric Frequency (MHz)
1614
1615
400
300
200
100
)
y
J
m
(
I
0
1611
6
4
2
0
−2
−4
)
y
J
m
(
V
−6
1611
Fig. 2. -- Total intensity and circular polarization results for IRAS F10038−3338 (IC 2545). See caption for Figure 1. (Top) Residuals
are expanded by a factor of 4.
16
Robishaw, Quataert, & Heiles
IRAS F10173+0829 Gaussian Fit Parameters
TABLE 3
Gaussian
(1)
0 . . . . . . .
1 . . . . . . .
2 . . . . . . .
3 . . . . . . .
4 . . . . . . .
5 . . . . . . .
6 . . . . . . .
S
(mJy)
(2)
ν
(MHz)
(3)
∆ν
(MHz)
(4)
v⊙
(km s−1)
(5)
Bk
(mG)
(6)
30.62 ± 6.34 1589.2420 ± 0.0037 0.0518 ± 0.0113
16.98 ± 5.95 1589.2862 ± 0.0041 0.0244 ± 0.0106
161.91 ± 7.80 1589.3190 ± 0.0024 0.1786 ± 0.0050
15.54 ± 7.63 1589.3258 ± 0.0073 0.0463 ± 0.0240
0.42 ± 1.58 1589.5315 ± 0.0194 0.5661 ± 0.0257
3.50 ± 2.82 1589.6383 ± 0.0154 0.0411 ± 0.0412
16.62 ± 2.46 1589.8772 ± 0.0072 0.1670 ± 0.0279
14735.9
2.34 ± 2.50
14727.1
4.19 ± 3.42
0.25 ± 0.89
14720.7
14719.3 −2.95 ± 5.15
14678.6
0.93 ± 5.54
14657.5 −2.90 ± 17.78
0.80 ± 7.57
14610.2
14900
Heliocentric Velocity (km s−1)
14600
14700
14800
14500
14800
14750
14700
14650
14600
14550
Heliocentric Velocity (km s−1)
10173+0829
200
150
100
50
)
y
J
m
(
I
8 mJy
1588.5
1589.0
1589.5
4 5
012
3
1590.0
6
1590.5
)
y
J
m
(
V
)
G
m
(
B
1589.0
1589.2
1589.4
1589.6
1589.8
1590.0
1589.0
1589.2
1589.4
1589.6
1589.8
1590.0
3
2
1
0
−1
−2
1.0
0.5
0.0
1588.5
1589.0
Heliocentric Frequency (MHz)
1589.5
1590.0
−0.5
−1.0
1590.5
1589.0
1589.2
1589.4
1589.8
Heliocentric Frequency (km s−1)
1589.6
1590.0
)
y
J
m
(
I
)
y
J
m
(
V
200
150
100
50
0
6
4
2
0
−2
−4
−6
Fig. 3. -- Total intensity and circular polarization results for IRAS F10173+0829. See caption for Figure 1. (Top left) Residuals are
expanded by a factor of 5. (Middle right) Dashed line shows the hypothetical Stokes V produced by a uniform Bk = 2 mG using the
derivative of the composite I profile above.
Extragalactic Zeeman Detections in OH Megamasers
17
IRAS F11506−3851 Gaussian Fit Parameters
TABLE 4
Gaussian
(1)
0 . . . . . . .
1 . . . . . . .
2 . . . . . . .
3 . . . . . . .
4 . . . . . . .
5 . . . . . . .
S
(mJy)
(2)
ν
(MHz)
(3)
∆ν
(MHz)
(4)
v⊙
(km s−1)
(5)
Bk
(mG)
(6)
64.73 ± 20.75 1650.0094 ± 0.0428 0.1910 ± 0.0456
102.55 ± 26.92 1650.1891 ± 0.0240 0.1810 ± 0.1018
86.37 ± 83.80 1650.3232 ± 0.0231 0.1358 ± 0.0467
116.30 ± 24.97 1650.4685 ± 0.0257 0.2693 ± 0.0976
0.31 ± 7.25
1650.7843 ± 0.1171 0.3616 ± 0.2594
23.79 ± 7.46
1651.0731 ± 0.0102 0.1260 ± 0.0368
3152.3
3119.3
3094.7
3068.0
3010.1
2957.1
1.21 ± 0.27
0.45 ± 0.20
0.36 ± 0.17
0.73 ± 0.20
0.68 ± 0.75
1.03 ± 0.50
18
Robishaw, Quataert, & Heiles
Heliocentric Velocity (km s−1)
3600
3400
3200
3000
2800
150
11506−3851
100
)
y
J
m
(
I
50
0
1647
1648
1649
16 mJy
1650
1651
0 1 2 3 4 5
1652
1653
)
y
J
m
(
V
1.5
1.0
0.5
0.0
−0.5
−1.0
−1.5
1647
1648
1649
Heliocentric Frequency (MHz)
1650
1651
1652
1653
Fig. 4. -- Total intensity and circular polarization results for IRAS F11506−3851. See caption for Figure 1. (Top) Residuals are expanded
by a factor of 2.
Extragalactic Zeeman Detections in OH Megamasers
19
IRAS F12032+1707 Gaussian Fit Parameters
TABLE 5
Gaussian
(1)
0 . . . . . . .
1 . . . . . . .
2 . . . . . . .
3 . . . . . . .
4 . . . . . . .
5 . . . . . . .
6 . . . . . . .
7 . . . . . . .
8 . . . . . . .
9 . . . . . . .
10 . . . . .
11 . . . . .
12 . . . . .
S
(mJy)
(2)
ν
(MHz)
(3)
∆ν
(MHz)
(4)
v⊙
(km s−1)
(5)
Bk
(mG)
(6)
4.31 ± 0.54 1367.1002 ± 0.0169 0.4082 ± 0.0370
3.95 ± 0.96 1367.6739 ± 0.0457 0.7915 ± 0.1793
8.42 ± 0.99 1368.8016 ± 0.0094 0.6264 ± 0.0448
3.00 ± 0.34 1369.5922 ± 0.0071 0.1460 ± 0.0207
0.21 ± 0.36 1369.6903 ± 0.0452 2.6434 ± 0.1574
3.23 ± 0.54 1369.9216 ± 0.0036 0.0450 ± 0.0089
6.67 ± 0.34 1370.7194 ± 0.0064 0.2954 ± 0.0165
21.67 ± 0.65 1371.2516 ± 0.0030 0.4769 ± 0.0115
16.99 ± 0.46 1371.3316 ± 0.0010 0.0849 ± 0.0029
16.19 ± 3.38 1372.1576 ± 0.0568 0.4872 ± 0.0531
25.02 ± 4.51 1372.3299 ± 0.0027 0.2795 ± 0.0165
9.56 ± 0.51 1372.7780 ± 0.0412 0.7315 ± 0.0588
2.02 ± 0.21 1373.8371 ± 0.0179 0.3807 ± 0.0500
65844.0 − 6.79 ± 7.94
65690.6
30.42 ± 12.43
65389.5
6.91 ± 4.86
65178.7 − 1.05 ± 6.48
65152.6 − 3.25 ± 3.98
65091.0
3.10 ± 3.41
11.64 ± 4.31
64878.6
64737.0
10.90 ± 1.72
64715.8
17.92 ± 0.89
1.78 ± 2.48
64496.4
64450.6 − 1.45 ± 1.12
64331.7 −11.69 ± 4.98
64051.0
20.61 ± 15.49
20
Robishaw, Quataert, & Heiles
Heliocentric Velocity (km s−1)
66000
65500
65000
64500
64000
63500
)
y
J
m
(
I
)
y
J
m
(
V
12032+1707
4 mJy
40
30
20
10
0
1366
1368
1
2
0
1370
4 5
3
1372
9
10 11
6
7
8
1374
12
1376
4
2
0
−2
−4
1366
1368
1370
1372
Heliocentric Frequency (MHz)
1374
1376
Fig. 5. -- Total intensity and circular polarization results for IRAS F12032+1707. See caption for Figure 1. (Top) Residuals are expanded
by a factor of 2.
Extragalactic Zeeman Detections in OH Megamasers
21
IRAS F12112+0305 Gaussian Fit Parameters
TABLE 6
Gaussian
(1)
0 . . . . . . .
1 . . . . . . .
2 . . . . . . .
3 . . . . . . .
4 . . . . . . .
S
(mJy)
(2)
ν
(MHz)
(3)
∆ν
(MHz)
(4)
v⊙
(km s−1)
(5)
Bk
(mG)
(6)
16.19 ± 0.75 1554.1822 ± 0.0013 0.1467 ± 0.0054
58.16 ± 0.51 1554.2767 ± 0.0048 0.5188 ± 0.0098
48.63 ± 0.97 1554.5634 ± 0.0007 0.1893 ± 0.0028
13.83 ± 0.59 1554.5641 ± 0.0006 0.0309 ± 0.0017
0.32 ± 0.34 1554.8544 ± 0.0037 0.4359 ± 0.0063
21831.1 −0.20 ± 1.99
21811.6
0.25 ± 1.17
1.02 ± 0.77
21752.3
21752.1
0.27 ± 1.03
3.09 ± 1.92
21692.1
22
Robishaw, Quataert, & Heiles
Heliocentric Velocity (km s−1)
21900
21800
21700
21600
12112+0305
6 mJy
)
y
J
m
(
I
100
80
60
40
20
0
1553.5
1554.0
0 1
1554.5
2
3
1555.0
4
1555.5
4
2
0
−2
)
y
J
m
(
V
−4
1553.5
1554.0
1554.5
Heliocentric Frequency (MHz)
1555.0
1555.5
Fig. 6. -- Total intensity and circular polarization results for IRAS F12112+0305. See caption for Figure 1. (Top) Residuals are expanded
by a factor of 4.
Extragalactic Zeeman Detections in OH Megamasers
23
IRAS F14070+0525 Gaussian Fit Parameters
TABLE 7
Gaussian
(1)
0 . . . . . . .
1 . . . . . . .
2 . . . . . . .
3 . . . . . . .
4 . . . . . . .
5 . . . . . . .
6 . . . . . . .
S
(mJy)
(2)
ν
(MHz)
(3)
∆ν
(MHz)
(4)
v⊙
(km s−1)
(5)
Bk
(mG)
(6)
3.50 ± 0.22 1315.2359 ± 0.0105 0.4088 ± 0.0317
8.95 ± 0.19 1316.0759 ± 0.0715 2.1625 ± 0.0861
5.70 ± 0.30 1316.2886 ± 0.0047 0.2040 ± 0.0134
11.26 ± 0.68 1316.9061 ± 0.0100 0.6771 ± 0.0302
0.08 ± 0.50 1317.5519 ± 0.0118 0.4878 ± 0.0271
2.78 ± 0.26 1318.8802 ± 0.0106 0.2664 ± 0.0307
10.76 ± 0.16 1319.1605 ± 0.0113 1.2737 ± 0.0199
80262.3 −10.42 ± 10.49
80019.7
26.65 ± 9.57
79958.4
1.76 ± 4.65
79780.3 − 3.31 ± 4.67
79594.2 − 3.16 ± 5.59
79212.1
12.72 ± 10.82
79131.6 −12.92 ± 6.16
24
Robishaw, Quataert, & Heiles
83000
82000
81000
80000
79000
78000
Heliocentric Velocity (km s−1)
14070+0525
3 mJy
20
15
10
5
0
1305
1310
1315
0 1
2 3 4
1320
6
5
1325
)
y
J
m
(
I
)
y
J
m
(
V
2
1
0
−1
−2
1305
1310
1320
Heliocentric Frequency (MHz)
1315
1325
Fig. 7. -- Total intensity and circular polarization results for IRAS F14070+0525. See caption for Figure 1. (Top) Residuals are expanded
by a factor of 2.
Extragalactic Zeeman Detections in OH Megamasers
25
IRAS F15327+2340 (Arp 220) Gaussian Fit Parameters
TABLE 8
Gaussian
(1)
0 . . . . . . .
1 . . . . . . .
2 . . . . . . .
3 . . . . . . .
4 . . . . . . .
5 . . . . . . .
6 . . . . . . .
7 . . . . . . .
8 . . . . . . .
9 . . . . . . .
10 . . . . .
11 . . . . .
12 . . . . .
13 . . . . .
14 . . . . .
15 . . . . .
16 . . . . .
17 . . . . .
S
(mJy)
(2)
ν
(MHz)
(3)
∆ν
(MHz)
(4)
v⊙
(km s−1)
(5)
Bk
(mG)
(6)
14.52 ± 0.98 1637.1424 ± 0.0008 0.0243 ± 0.0020
10.92 ± 0.61 1637.3105 ± 0.0018 0.0709 ± 0.0050
11.80 ± 1.02 1637.5074 ± 0.0009 0.0231 ± 0.0024
10.12 ± 1.00 1637.5736 ± 0.0011 0.0241 ± 0.0029
0.90 ± 0.77 1637.7198 ± 0.0002 0.0590 ± 0.0006
51.49 ± 1.06 1637.8723 ± 0.0005 0.0621 ± 0.0016
324.77 ± 3.96 1637.8916 ± 0.0010 0.3362 ± 0.0028
97.84 ± 2.60 1638.0196 ± 0.0008 0.0800 ± 0.0023
293.07 ± 4.06 1638.0313 ± 0.0011 1.0064 ± 0.0075
386.20 ± 5.88 1638.1189 ± 0.0007 0.0882 ± 0.0012
81.13 ± 2.84 1638.1375 ± 0.0002 0.0292 ± 0.0008
237.99 ± 4.75 1638.2098 ± 0.0011 0.1039 ± 0.0024
46.95 ± 2.66 1638.3468 ± 0.0051 0.1894 ± 0.0103
18.22 ± 1.02 1638.4199 ± 0.0008 0.0342 ± 0.0024
19.79 ± 0.78 1638.6066 ± 0.0011 0.0769 ± 0.0036
7.62 ± 0.60 1638.8445 ± 0.0033 0.1038 ± 0.0103
11.23 ± 0.82 1638.9714 ± 0.0014 0.0367 ± 0.0034
13.18 ± 0.57 1639.0663 ± 0.0018 0.0908 ± 0.0051
5533.2
5501.9
5465.2
5452.9
5425.6
5397.2
5393.6
5369.7
5367.6
5351.2
5347.8
5334.3
5308.8
5295.2
5260.4
5216.1
5192.5
5174.9
−4.78 ± 0.53
−0.11 ± 1.21
−2.78 ± 0.64
7.77 ± 0.76
−2.78 ± 0.13
0.33 ± 0.25
0.26 ± 0.12
−0.15 ± 0.18
0.14 ± 0.21
−0.76 ± 0.06
−0.24 ± 0.11
0.66 ± 0.10
−1.03 ± 0.59
0.20 ± 0.50
0.22 ± 0.69
1.78 ± 2.10
1.42 ± 0.86
−0.46 ± 1.15
26
Robishaw, Quataert, & Heiles
Heliocentric Velocity (km s−1)
5500
5400
5300
5200
800
15327+2340
20 mJy
1638.0
1011 1213
9
6
5
1638.5
14
8
7
1637.0
0
1637.5
2 3
1
4
)
y
J
m
(
I
)
y
J
m
(
V
600
400
200
0
15
10
5
0
−5
−10
−15
1637.0
1637.5
1638.0
1638.5
Heliocentric Frequency (MHz)
1639.0
15 1617
1639.0
Fig. 8. -- Total intensity and circular polarization results for IRAS F15327+2340 (Arp 220). See caption for Figure 1. (Top) Residuals
are expanded by a factor of 16.
Extragalactic Zeeman Detections in OH Megamasers
27
8800
Heliocentric Velocity (km s−1)
8600
8400
8200
8000
01417+1651
1620
1621
1622
1623
1624
1625
1620
1621
1622
1623
1624
1625
RM
≈
−21900 rad m−2
500
400
300
200
100
0
)
y
J
m
(
I
)
y
J
m
(
n
o
i
t
a
z
i
r
a
l
o
P
r
a
e
n
L
i
1619
12
10
8
6
4
2
0
1619
90
)
g
e
d
(
e
l
g
n
A
n
o
i
t
i
s
o
P
45
0
−45
−90
Heliocentric Velocity (km s−1)
9500
9000
8500
8000
7500
01417+1651
500
400
300
200
100
0
)
y
J
m
(
I
1616
1618
1620
1622
1624
1626
1628
)
y
J
m
(
n
o
i
t
a
z
i
r
a
l
o
P
r
a
e
n
L
i
12
10
8
6
4
2
0
1616
1618
1620
1622
1624
1626
1628
)
g
e
d
(
e
l
g
n
A
n
o
i
t
i
s
o
P
90
45
0
−45
−90
1616
1618
1620
1622
1624
1626
1628
1619
1620
Heliocentric Frequency (MHz)
1621
Heliocentric Frequency (MHz)
1622
1623
1624
1625
Fig. 9. -- Linear polarization results for IRAS F01417+1651 (III Zw 35). (Left) Top panel shows Stokes I, middle panel shows the linearly
polarized intensity, and bottom panel shows the position angle over the entire 12.5 MHz bandwidth. (Right) Same as left panels, but with
the frequency range narrowed to 6 MHz; the bottom right plot also shows the fitted Faraday rotation as a dashed line whose slope was
determined by fitting to the points marked as diamonds. All spectra are plotted as a function of heliocentric frequency (bottom axis). The
top panels show the optical heliocentric velocity (top axis). All spectra are smoothed by a boxcar of 23 channels.
28
Robishaw, Quataert, & Heiles
16000
Heliocentric Velocity (km s−1)
14500
15000
15500
14000
10173+0829
1584
1586
1588
1590
1592
1594
1584
1586
1588
1590
1592
1594
200
150
100
50
0
3.0
2.5
2.0
1.5
1.0
0.5
0.0
90
45
0
−45
−90
)
y
J
m
(
I
)
y
J
m
(
n
o
i
t
a
z
i
r
a
l
o
P
r
a
e
n
L
i
)
g
e
d
(
e
l
g
n
A
n
o
i
t
i
s
o
P
1584
1586
Heliocentric Frequency (MHz)
1590
1588
1592
1594
Fig. 10. -- Linear polarization results for IRAS F10173+0829. See caption for Figure 9. All spectra are smoothed by a boxcar of 17
channels.
Extragalactic Zeeman Detections in OH Megamasers
29
66500
66000
Heliocentric Velocity (km s−1)
64500
65500
65000
64000
63500
12032+1707
1364
1366
1368
1370
1372
1374
1376
1364
1366
1368
1370
1372
1374
1376
40
30
20
10
0
10
8
6
4
2
0
)
y
J
m
(
I
)
y
J
m
(
n
o
i
t
a
z
i
r
a
l
o
P
r
a
e
n
L
i
)
g
e
d
(
e
l
g
n
A
n
o
i
t
i
s
o
P
90
45
0
−45
−90
1364
1366
1368
1370
Heliocentric Frequency (MHz)
1372
1374
1376
Fig. 11. -- Linear polarization results for IRAS F12032+1707. See caption for Figure 9. All spectra are smoothed by a boxcar of 31
channels.
30
Robishaw, Quataert, & Heiles
23000
Heliocentric Velocity (km s−1)
22500
21500
22000
21000
12112+0305
1548
1550
1552
1554
1556
1558
1548
1550
1552
1554
1556
1558
100
80
60
40
20
0
10
8
6
4
2
0
80
60
40
20
0
)
y
J
m
(
I
)
y
J
m
(
n
o
i
t
a
z
i
r
a
l
o
P
r
a
e
n
L
i
)
g
e
d
(
e
l
g
n
A
n
o
i
t
i
s
o
P
1548
1550
1552
1556
Heliocentric Frequency (MHz)
1554
1558
Fig. 12. -- Linear polarization results for IRAS F12112+0305. See caption for Figure 9. All spectra are smoothed by a boxcar of 11
channels.
Extragalactic Zeeman Detections in OH Megamasers
31
Heliocentric Velocity (km s−1)
81000
80000
79000
14070+0525
1310
1312
1314
1316
1318
1320
1310
1312
1314
1316
1318
1320
20
15
10
5
0
7
6
5
4
3
2
1
)
y
J
m
(
I
)
y
J
m
(
n
o
i
t
a
z
i
r
a
l
o
P
r
a
e
n
L
i
)
g
e
d
(
e
l
g
n
A
n
o
i
t
i
s
o
P
−15
−30
−45
−60
−75
1310
1312
1314
1318
Heliocentric Frequency (MHz)
1316
1320
Fig. 13. -- Linear polarization results for IRAS F14070+0525. See caption for Figure 9. All spectra are smoothed by a boxcar of 11
channels.
Robishaw, Quataert, & Heiles
6500
Heliocentric Velocity (km s−1)
6000
5500
5000
4500
Heliocentric Velocity (km s−1)
5400
5600
5800
5200
32
)
y
J
m
(
I
)
y
J
m
(
n
o
i
t
a
z
i
r
a
l
o
P
r
a
e
n
L
i
)
g
e
d
(
e
l
g
n
A
n
o
i
t
i
s
o
P
800
600
400
200
0
3.0
2.5
2.0
1.5
1.0
0.5
0.0
90
45
0
−45
−90
15327+2340
1632
1634
1636
1638
1640
1642
15327+2340
800
600
400
200
0
)
y
J
m
(
I
)
y
J
m
(
n
o
i
t
a
z
i
r
a
l
o
P
r
a
e
n
L
i
1632
1634
1636
1638
1640
1642
)
g
e
d
(
e
l
g
n
A
n
o
i
t
i
s
o
P
1635
3.0
2.5
2.0
1.5
1.0
0.5
0.0
1635
−20
−30
−40
−50
−60
1636
1637
1638
1639
1636
1637
RM
1638
≈
1250 rad m−2
1639
1639
1632
1634
1636
1638
1640
1642
1635
Heliocentric Frequency (MHz)
1636
1637
1638
Heliocentric Frequency (MHz)
Fig. 14. -- Linear polarization results for IRAS F15327+2340 (Arp 220). See caption for Figure 9. (Right) The frequency range has been
narrowed to 4 MHz. All spectra are smoothed by a boxcar of 9 channels.
|
astro-ph/0204246 | 1 | 0204 | 2002-04-15T16:56:25 | Observation of periodic variable stars towards the galactic spiral arms by EROS II | [
"astro-ph"
] | We present the results of a massive variability search based on a photometric survey of a six square degree region along the Galactic plane at ($l = 305^\circ$, $b = -0.8^\circ$) and ($l = 330^\circ$, $b = -2.5^\circ$). This survey was performed in the framework of the EROS II (Exp\'erience de Recherche d'Objets Sombres) microlensing program. The variable stars were found among 1,913,576 stars that were monitored between April and June 1998 in two passbands, with an average of 60 measurements. A new period-search technique is proposed which makes use of a statistical variable that characterizes the overall regularity of the flux versus phase diagram. This method is well suited when the photometric data are unevenly distributed in time, as is our case. 1,362 objects whose luminosity varies were selected. Among them we identified 9 Cepheids, 19 RR Lyrae, 34 Miras, 176 eclipsing binaries and 266 Semi-Regular stars. Most of them are newly identified objects. The cross-identification with known catalogues has been performed. The mean distance of the RR Lyrae is estimated to be $\sim 4.9 \pm 0.3$ kpc undergoing an average absorption of $\sim 3.4 \pm 0.2$ magnitudes. This distance is in good agreement with the one of disc stars which contribute to the microlensing source star population.Our catalogue and light curves are available electronically from the CDS, Strasbourg and from our Web site http://eros.in2p3.fr. | astro-ph | astro-ph | Astronomy & Astrophysics manuscript no.
(will be inserted by hand later)
Observation of periodic variable stars towards the Galactic spiral
arms by EROS II ⋆
F. Derue1,2 ⋆⋆, J-B. Marquette3, S. Lupone3, C. Afonso2,4 ⋆⋆⋆, C. Alard5, J-N. Albert1, A. Amadon2,
J. Andersen6, R. Ansari1, ´E. Aubourg2, P. Bareyre2,4, F. Bauer2, J-P. Beaulieu3, G. Blanc2†, A. Bouquet4,
S. Char7‡, X. Charlot2, F. Couchot1, C. Coutures2, R. Ferlet3, P. Fouqu´e8,9 J-F. Glicenstein2,
B. Goldman2,4 ⋆⋆⋆, A. Gould4,10, D. Graff2,11, M. Gros2, J. Haıssinski1, J-C. Hamilton4§, D. Hardin2¶, J. de
Kat2, A. Kim4†, T. Lasserre2, L. Le Guillou2, ´E. Lesquoy2,3, C. Loup3, C. Magneville2, B. Mansoux1,
´E. Maurice12, A. Milsztajn2, M. Moniez1, N. Palanque-Delabrouille2, O. Perdereau1, L. Pr´evot12,
N. Regnault1†, J. Rich2, M. Spiro2, A. Vidal-Madjar3, L. Vigroux2, S. Zylberajch2
The EROS collaboration
1 Laboratoire de l'Acc´el´erateur Lin´eaire, IN2P3-CNRS-Universit´e de Paris-Sud, B.P. 34, 91898 Orsay Cedex,
France
2 CEA, DSM, DAPNIA, Centre d'´Etudes de Saclay, 91191 Gif-sur-Yvette Cedex, France
3 Institut d'Astrophysique de Paris, INSU-CNRS, 98 bis Boulevard Arago, 75014 Paris, France
4 Coll`ege de France, LPCC, IN2P3-CNRS, 11 place Marcelin Berthelot, 75231 Paris Cedex, France
5 DASGAL, INSU-CNRS, 77 avenue de l'Observatoire, 75014 Paris, France
6 Astronomical Observatory, Copenhagen University, Juliane Maries Vej 30, 2100 Copenhagen, Denmark
7 Universidad de la Serena, Facultad de Ciencias, Departamento de Fisica, Casilla 554, La Serena, Chile
8 Observatoire de Paris, DESPA, 92195 Meudon Cedex, France
9 European Southern Observatory (ESO), Casilla 19001, Santiago 19, Chile
10 Ohio State University, Department of Astronomy, Columbus, OH 43210, U.S.A.
11 University of Michigan, Department of Astronomy, Ann Arbor, MI48109, U.S.A.
12 Observatoire de Marseille, INSU-CNRS, 2 place Le Verrier, 13248 Marseille Cedex 04, France
Received <date> / Accepted <date>
Abstract. We present the results of a massive variability search based on a photometric survey of a six square
degree region along the Galactic plane at (l = 305◦, b = −0.8◦) and (l = 330◦, b = −2.5◦). This survey was
performed in the framework of the EROS II (Exp´erience de Recherche d'Objets Sombres) microlensing program.
The variable stars were found among 1,913,576 stars that were monitored between April and June 1998 in two
passbands, with an average of 60 measurements. A new period-search technique is proposed which makes use of a
statistical variable that characterizes the overall regularity of the flux versus phase diagram. This method is well
suited when the photometric data are unevenly distributed in time, as is our case. 1,362 objects whose luminosity
varies were selected. Among them we identified 9 Cepheids, 19 RR Lyrae, 34 Miras, 176 eclipsing binaries and 266
Semi-Regular stars. Most of them are newly identified objects. The cross-identification with known catalogues
has been performed. The mean distance of the RR Lyrae is estimated to be ∼ 4.9 ± 0.3 kpc undergoing an
average absorption of ∼ 3.4± 0.2 magnitudes. This distance is in good agreement with the one of disc stars which
contribute to the microlensing source star population. Our catalogue and light curves are available electronically
from the CDS, Strasbourg and from our Web sitek.
2
0
0
2
r
p
A
5
1
1
v
6
4
2
4
0
2
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
Send offprint requests to: J-B. Marquette : [email protected]
⋆ This work is based on observations made with the MARLY
telescope of the EROS collaboration at the European Southern
Observatory, La Silla, Chile.
⋆⋆ Presently at Centre de Physique des Particules de
Marseille, IN2P3-CNRS, 163 avenue de Luminy, case 907,
13288 Marseille Cedex 09, France
⋆⋆⋆ Presently at Department of Astronomy, New Mexico State
University, Las Cruces, NM88003-8001, U.S.A.
† Presently at University of California, Department of
Physics, Berkeley, CA 97720, U.S.A
1. Introduction
In 1996, the EROS II collaboration started an observa-
tion program towards the Galactic Spiral Arms (GSA)
‡ deceased
§ Presently at
ISN,
Joseph-
Fourier, 53 avenue des Martyrs, 38026 Grenoble Cedex, France
¶ Presently at LPNHE, IN2P3-CNRS-Universit´es Paris VI
et VII, 4 place Jussieu, F-75252 Paris Cedex 05, France
IN2P3-CNRS-Universit´e
2
The EROS collaboration: Observation of periodic variable stars
dedicated to microlensing events. Since then, four regions
of the Galactic plane located at large angles with respect
to the Galactic Centre are being monitored to disentangle
the disc, bar and halo contributions to the microlensing
optical depth. Seven microlensing event candidates have
already been published, based on three years (1996-98) of
observations (Derue et al. 1999, 2001)[hereafter papers I
and II]. The distance of the source stars used in these pa-
pers to compute the expected optical depths was deduced
from a detailed study of our colour-magnitude diagrams. It
was thus found that the source star population is located
∼ 7 kpc away, undergoing an interstellar extinction A(V )
of about 3 magnitudes (see Mansoux (1997) for more de-
tails). This distance estimate is in rough agreement with
the distance to the spiral arms obtained by Georgelin et
al. (1994) and Russeil et al. (1998), but its uncertainty is
limiting further interpretation of our microlensing optical
depth estimates. It was therefore desirable to seek more
information on the distance distribution of the source stars
– whether these stars belong to the disc or to the spiral
arms – and on the reddening along our observation line
of sights. This led us to perform a dedicated variable star
search between April and June 1998, on a subset of our
Galactic plane fields. The analysis was restricted to the
brightest stars of this subset.
Among the wide variety of variable stars, periodic ones
are of particular interest. The properties of Cepheids make
them well suited to trace the Galactic spiral arms. Their
reddening is measurable as well as their distance via the
period-luminosity (PL) relation. RR Lyrae stars are old
stars, well suited to trace the disc population. One can
infer their mean dereddened magnitude and their abso-
lute magnitude (Gould and Popowski 1998). The infrared
PL(K) relation for Miras and Semi-Regular variable stars
can be calibrated using a comparison of DENIS and
EROS LMC giant stars (Cioni et al. 2001). Finally de-
tached eclipsing binaries also offer the opportunity to mea-
sure their stellar parameters and their distance (Paczy´nski
1996).
This paper presents the results of this particular cam-
paign which led to a catalogue containing a large number
of new variable objects in the Galactic plane. Sect. 2 gives
the basic features of the observational setup, Sect. 3 gives
details on a new algorithm used to search for periodic vari-
ations of the luminosity. Sect. 4 describes the catalogue
and the cross-identification process. In Sect. 5 we use the
selected RR Lyrae to estimate the mean reddening of our
fields and we give the distance distribution of these stars.
2. Experimental setup and observations
The MARLY telescope and its two cameras, the way we
carry out our observations, as well as our data reduction
sequence are described in Paper I and references therein.
The two EROS passbands are non standard. The so-
called EROS-red passband RE is centred on ¯λ = 762 nm,
close to IC Cousins, with a full width at half maximum
∆λ ≃ 85 nm, and EROS-visible passband VE is centred
(1)
on ¯λ = 600 nm, close to VJ Johnson, with ∆λ ≃ 78 nm.
The EROS II colour magnitude system is defined as fol-
lows : a zero colour star with VE − RE = 0 (a main
sequence A0 star) will have its RE magnitude numeri-
cally equal to its Cousins RC magnitude and its VE mag-
nitude numerically equal to its Johnson VJ magnitude.
The colour transformation between the EROS II system
(VE,RE) and the standard Johnson-Cousins (VJ ,IC ) sys-
tem is then :
IC = RE + 0.01 × (VE − RE)
VJ = VE + 0.39 × (VE − RE) .
The colour coefficients are obtained from the study of
our passbands based on Landolt standards and on one
of the EROS II LMC fields observed simultaneously in
BVJ RC IC with the Danish 1.54 m (at ESO-La Silla) and
with the MARLY. The zero points are established with
tertiary standards in BV RI taken with the Danish 1.54
(Regnault 2000). We have cross-checked our RE photom-
etry with the IC one of DENIS (Fouqu´e et al. 2000).
Furthermore, using Eqs. (1), the mean magnitudes of the
LMC red-giant clump stars agree within 0.1 magnitude
with determinations made by Harris & Zaritsky (1999)
and Udalski et al. (1998). We thus estimate that the pre-
cision of the zero points of the MARLY calibration is
∼ 0.1mag.
Table 1. Characteristics of the 6 one square degree
fields monitored for this study. The equatorial (J2000)
and galactic coordinates of the field centres, the number of
photometric measurements per light curve per colour Nm
and the number of analysed light curves Nobs (in millions)
are given for each field. The observation duration was ∼
100 days.
Field
α
δ
b◦
l◦
Norma (γ Nor)
gn450
gn453
gn455
gn459
16:09:45
16:22:28
16:26:52
16:15:51
-53:07:03
-52:06:20
-52:21:02
-54:48:45
-1.17
-1.69
-2.35
-2.86
330.49
332.24
332.54
329.82
Musca (θ Mus)
tm550
tm551
13:27:04
13:31:18
-63:02:18
-63:34:41
-0.47
-1.07
306.98
307.37
Nm Nobs
59
1.30
0.28
63
0.36
52
0.30
57
61
0.36
0.61
60
0.27
61
0.34
60
Total
1.91
Among the 29 fields of the EROS GSA microlensing
program, the six fields considered here were monitored
about once per night between April and June 1998. They
represent 2 square degrees towards θ Mus and 4 towards
γ Nor (named after the closest bright star). Table 1 gives
their coordinates, the number of images Nm taken in each
direction and the number of analysed light curves Nobs.
To avoid CCD saturation by the brightest stars (IC ∼ 9),
by Cepheids or Miras in particular, we have reduced the
exposure time to 15 s instead of the 120 s used in the
The EROS collaboration: Observation of periodic variable stars
3
and VE. The photometric accuracy is ∼15% at RE∼18,
and about 2% for the brightest stars. Finally, using the
k
e
e
w
/
s
t
n
e
m
e
r
u
s
a
e
m
f
o
r
e
b
m
u
N
16
14
12
10
8
6
4
2
16
14
12
10
8
6
4
2
0
114
116
118
120
122
124
126
128
130
132
134
Week number since 01/01/96
Fig. 1. Average time sampling for the 6 fields monitored
(2 towards γ Nor, upper panel, and 4 towards θ Mus, lower
panel), in number of measurements per week and per field.
microlensing survey. As a consequence the catalogue is
incomplete as far as faint stars are concerned, but it could
be updated later by using the total set of available GSA
images. Fig. 1 shows the average time sampling. Three
gaps can be seen in our data : the first two (around weeks
119 and 123) were due to bad weather conditions while the
third one (around week 127) corresponds to the annual
maintenance of our setup.
3. Search for periodic stars
3.1. Reconstruction of the light curves
Since the EROS photometry is described in detail
in
Ansari (1996) only the main features of the PEIDA++
package are summarised below. For each field, a tem-
plate image is first constructed using one exposure of very
good quality. A reference star catalogue is set up with
this template using the CORRFIND star finding algo-
rithm (Palanque–Delabrouille et al. 1998). For each sub-
sequent image, after geometrical alignment with the tem-
plate, each identified star is fitted together with its neigh-
bours, using a PSF determined on bright isolated stars
and imposing the position from the reference catalogue.
A relative photometric alignment is then performed, as-
suming that most stars do not vary. Photometric errors
are computed for each measurement, assuming again that
most stars are stable, and parametrised as a function of
star brightness and image sequence number. Fig. 2 shows
the mean point-to-point relative dispersion of the mea-
sured fluxes along the light curves as a function of RE
Fig. 2. Average value of the relative frame to frame dis-
persion of the luminosity measurements versus RE (upper
panel) and VE (lower panel), for stars with at least 30
reliable measurements in each colour. This dispersion is
taken as an estimator of the mean photometric precision.
The superimposed hatched histograms show the EROS
magnitude distribution of monitored stars.
PEIDA++ photometric package, we reconstruct the light
curves of 1,913,576 stars.
Table 2. Impact of each selection criterion on the data.
For each cut, the number of remaining light curves is given.
Cut Criterion
Total analysed
Nm > 30
RE < 17
Pre-filtering
Period search
Aliasing
Visual inspection
1,913,576
1,299,690
330,089
41,545
2,553
2,424
1,362
1
2
3
4
5
6
3.2. Pre-selection
Each one of the light curves is subjected to a series of
selection criteria in order to isolate a small sub-sample
on which we will apply the time consuming period search
algorithms. These analysis cuts are briefly described here-
4
The EROS collaboration: Observation of periodic variable stars
after (see Derue (1999) for more details) and their effect
on the data is summarised in Table 2 :
cut 1 : At least 30 measurements should be available in both
passbands and the base flux must be positive;
cut 2 : The search is restricted to stars whose magnitude is
RE<17 which corresponds to a photometric accuracy
in RE better than ∼10%.
cut 3 : A non specific pre-filter is applied which retains most
variable stars. It selects light curves satisfying one or
both of the following criteria :
– the relative dispersion of the flux measurements is
25% larger than the average one for the set of stars
having the same magnitude;
– the distribution of the deviations with respect to
the base flux is incompatible with the one expected
from a stable source with Gaussian errors during
the observation period (Kolmogorov-Smirnov test).
These cuts are tuned to select ∼10% of the light curves.
We have checked that this procedure allows one to
retrieve the previously known Cepheids observed by
EROS in the Magellanic Clouds. We also keep a ran-
domly selected set of light curves (∼2%) to produce
unbiased colour-magnitude diagrams, for comparison
purposes;
At this stage a set of 41,545 light curves remains which is
then subjected to a periodicity search.
3.3. Light curve selection
We use three independent methods to extract periodic
light curves. The first two are classical methods al-
ready described in the literature : method 1 is based on
the Lomb-Scargle periodogram (see Scargle 1982) while
method 2 makes use of the One Way Analysis of Variance
algorithm (see Schwarzenberg-Czerny 1996). Both provide
the probability for false periodicity detection. In method
1 one computes the Fourier power over a set of frequen-
cies. It is therefore well adapted to identify sinusoidal light
curves. It can be improved by incorporating higher har-
monics in order to detect any kind of variability such as
eclipsing binaries (Grison 1994; Grison et al. 1995).
We developed a new method, the third one, in order to
extract periodic light curves in a way which is insensitive
to the particular shape of the variation. This method also
provides a probability for false periodicity detection.
It consists in searching for a frequency such that the
corresponding phase diagram, i.e the series of fluxes Fi
versus phases ϕi in increasing order of ϕ, displays a reg-
ular structure significantly less scattered than for other
frequencies. Let Tobs be the observation duration (∼ 100
days in this analysis). We span the frequency domain from
T −1
obs to (0.2 days)−1 with a constant step of (4 × Tobs)−1.
The total number of test-frequencies is thus Ntest ∼ 2000.
This sampling ensures that the total phase increment over
Tobs is π/2 for two adjacent test-frequencies.
For each value of the test-frequency we compute the
corresponding phase diagram. We calculate a χ2 from the
Fig. 3. Distributions of χ2 (left panels) and variation of x
with the value of the test-period in days (middle panels)
obtained with the third method for a stable star and for
a typical Cepheid candidate. The bold dots are pointing
to the actual period of the star. The right panels show
the light curve (in RE) obtained once the period has been
folded in.
weighted differences of Fi and the fluxes interpolated be-
tween Fi−1 and Fi+1 :
χ2 =
Nm
Xi=1
(cid:18) Fi − (1 − Ri) × Fi−1 − Ri × Fi+1
σi
(cid:19)2
(2)
i = ǫ2
i ǫ2
i−1 + R2
i + (1 − Ri)2ǫ2
where Ri = (ϕi − ϕi−1)/(ϕi+1 − ϕi−1) and Nm is the
number of measurements. The uncertainty σi takes into
account the errors ǫi on the flux Fi and on the interpo-
lated flux : σ2
i+1. Expression
(2) can be interpreted as the χ2 of the set of differences
between the odd measurements with respect to the line
joining even ones, added to the χ2 of the set of differ-
ences between even measurements with respect to the line
joining odd ones. If the star is measured in both colours,
we add the χ2 obtained in each colour (Nm is then twice
as large). For a given stable star with Gaussian errors,
each phase diagram can be considered as a random real-
isation of the light curve. When the test-frequency spans
the search domain, the distribution of the χ2 parameter
defined by Eq. (2) is the one of the standard χ2 with Nm
degrees of freedom. Since Nm is large enough, this distri-
bution is close to a Gaussian with average Nm and vari-
ance 2Nm (see upper left panel of Fig. 3). For a periodic
variable star, the χ2 distribution displays a main cluster,
when the test-frequency results in a phase diagram with
non-correlated point to point variations, and a few lower
The EROS collaboration: Observation of periodic variable stars
5
10 4
10 3
10 2
10
1
0
10 4
10 3
10 2
10
1
0
10 4
10 3
10 2
10
1
0
10
20
30
40
50
60
70
80
90
100
10
20
30
40
50
60
70
80
90
100
10
20
30
40
50
60
70
80
90
100
panels of Fig. 3), and the probability for false detection
is then extremely small. Fig. 4 displays the probability
distribution obtained with the three methods for the set
of filtered light curves1.
We apply the three algorithms which all give a prob-
ability for no periodicity. A star is accepted only if se-
lected by all three methods with the following thresh-
olds : P (method 1) < 10−15, P (method 2) < 10−10,
P (method 3) < 10−5, tuned in order to allow one to re-
trieve the previously known Cepheids observed by EROS
in the Magellanic Clouds (see Derue (1999) for more de-
tails). This procedure (cut 4) selects a sample of 2,553
stars.
Ten times more stars would have been selected if we
had used method 1 or method 2 only (with the same
thresholds), most of them being spurious variables. The
Fig. 4. Distribution of − log(P rob) for method 1 (top), 2
(middle) and 3 (bottom). The histograms show the distri-
bution for all 41,545 stars on which the periodicity search
is done (in white) and for the 2,553 selected light curves
(in black).
values when the test-frequency corresponds to a phase di-
agram with a regular structure (see lower left and middle
panels of Fig. 3). In practice, instead of using the param-
eter defined by Eq. (2), we use the reduced variable :
x =
χ2− < χ2 >
< χ2 > /Nm ×
1
√2Nm
(3)
6
5
4
3
2
1
0
where < χ2 > is the average of the realisations of χ2 for all
test-frequencies. For a stable star, the distribution of this
variable x is a Gaussian centred at zero, with unit vari-
ance. If the errors are correctly determined < χ2 > /Nm
is close to unity; if the errors are all systematically over-
estimated (or underestimated), then including this term
ensures a global renormalisation of the errors in Eq. (3),
and the distribution of our reduced variable x is also a
normal distribution. Let xmin be the smallest value of
x calculated among all test-frequencies for a given star.
Under the hypothesis that the light curve is produced by
a stable star, the probability to obtain at least one value
x ≤ xmin in a series of Ntest realisations is P rob(x <
xminNtest) = 1 − h1 − 1
. If this
√2π R xmin
probability is small, then :
−∞
P rob(x < xminNtest) ≃
If the light curve exhibits periodic variations, then there
exist test-frequencies for which x is significantly smaller
than typical values of this variable (see middle and right
(4)
2
e−x2/2dxiNtest
erfc(cid:18)−xmin√2 (cid:19) .
Ntest
-1
10
-1
1
10
2
10
Fig. 5. Period-colour diagram (P in days vs VJ -IC ) of the
2,553 selected candidates before cut 5.
third method would have added far less candidates if used
alone, but still a factor 2 more. Combining independent
methods has thus the advantage of considerably reducing
the noise background (mostly due to aliases of one day or
noisy measurements) while giving redundant information
about the period. To obtain individual periods we perform
Fourier fits with five harmonics :
flux = flux0 +
5
Xl=1
al cos(
2π
P
l(t − t0) + φl),
(5)
1 The third method gives a probability which is always larger
than 10−30. Indeed, the smallest value of x which can be ob-
tained takes place when χ2 = 0, then x ∼ √Nm < 10.
6
The EROS collaboration: Observation of periodic variable stars
where P is the period, φl the phase and al the amplitude.
We define the amplitude ratios Rkl = ak/al, and the phase
differences (defined modulo 2π) Φkl = φk − kφl, with k >
l. Objects with non-significant harmonic amplitudes (i.e
with almost sinusoidal light curves) have R21 ∼ 0 and
their Φ21 is ill defined.
The selection of periodic variable stars is complicated
by aliases. Some of the stars with periods equal to a sim-
ple fraction or a low multiple of one day may be badly
phased because of the nightly sequence of measurements.
These aliased periods are seen in Fig. 5 as vertical groups
of dots at 2/3, 2 and 3 days. To eliminate them we de-
mand (cut 5) that the fitted periods are not within ±1%
of these values. One can also notice some vertical groups
of points around 25 days which correspond to data gaps
in our sample (see Fig. 1). Once these objects are re-
moved, 2,424 stars remain. The flux values of the remain-
ing stars are folded using each period obtained with the
three methods. The resulting phase diagrams are visually
inspected. Some of them display an obvious spurious peri-
odic or quasi-periodic variability due to a low photometric
quality. After this final visual selection (cut 6) the list of
variable stars includes 1,362 candidates which exhibit un-
ambiguous periodic variability.
3.4. Type of variability
The classification of the selected stars among different
types of variability cannot be based on the position of the
objects in the colour-magnitude diagram since the spread
in distance of these stars entails a spread in magnitude
and colour. It is desirable however to classify the vari-
ous light curves according to some physical parameters.
In the following we mainly use criteria based on the pe-
riod P of the luminosity variations and on the amplitude
ratio ∆VE/∆RE. For each selected type of variable star
the phase diagram of a typical candidate is displayed in
Fig. 6. Three groups are distinguished depending on their
period :
1st : stars with a period larger than 60 days (543 objects) :
For these objects an entire period has not been ob-
served. There is thus no warranty that these objects
are periodic ones. 34 display a nearly linear light curve
and are thus catalogued as Miras candidates. The other
509 objects are catalogued as Long Period Variable
stars (LPVs).
2nd : stars with a period between 30 and 60 days (387 ob-
jects) :
264 Semi-Regular variable stars are selected by requir-
ing : ∆VE/∆RE > 1.2 to select pulsating stars (see be-
low), VJ -IC > 2.5 to discriminate from bluer variable
stars and ∆RE < 1.0 to avoid possible Miras or LPVs
wrongly phased. The long term stability of these stars
is not known. Some of the reported periods may change
from season to season, as a result of their semi-regular
behaviour. The remaining 123 objects are catalogued
as miscellaneous variable stars.
Table 3. Number of selected objects for each type of vari-
ability.
Period range
P > 60 d
60 > P > 30 d
Type
LPV
Miras
P < 30 d
Semi-Regular
miscellaneous
pulsating
RRc
RRab
classical-Cepheids
s-Cepheids
miscellaneous
non-pulsating
EA
EB
EW
miscellaneous
Number of objects
543
509
34
387
264
123
432
60
14
5
6
3
32
372
130
35
11
196
3rd : stars with a period smaller than 30 days (432 objects) :
The colour change for a Cepheid in standard passbands
is ∆VJ /∆IC >1.3 (Madore et al. 1991) which corre-
sponds to ∆VE/∆RE > 1.2 in the EROS system. Two
sets are thus distinguished based on this criterion :
- The pulsating variable stars (60 objects) :
For stars with period P < 1 day, two samples of
RR Lyrae are identified : the RRc have R21 < 0.4
(14 objects) and the RRab have R21 > 0.4 (5 ob-
jects). We adopt the morphological classification
proposed by Antonello et al. (1986) and classify
as s-Cepheids the stars that lie in the lower part
of the R21 − P plane, and as classical Cepheids
the remaining stars. s-Cepheids pulsate in the first
overtone and classical Cepheids in the fundamen-
tal mode (see e.g Beaulieu et al. (1995); Beaulieu
& Sasselov (1996)). We use the empirical function
Rcut
21 (P ) = 0.4 − (P/30 days) in order to sepa-
rate these pulsation modes (see Fig. 7). Among
the five objects that pass the s-Cepheid cut, only
three belong to the R21 − P and Φ21 − P distri-
butions of galactic s-Cepheids and their phase pa-
rameter Φ21 is poorly constrained. Besides most
of classical Cepheids have amplitudes larger than
∆RE > 0.4mag (see e.g Afonso et al. (1999)). For
stars with period P > 1 day, three samples are then
identified : the classical Cepheids have R21 > Rcut
21
and ∆RE > 0.4mag (6 objects); the s-Cepheids
have R21 < Rcut
21 and 3 rad < Φ21 < 6 rad (3
objects); the remaining 32 objects are catalogued
as miscellaneous pulsating stars.
- The non-pulsating variable stars (372 objects) :
The remaining objects have similar amplitudes in
both passbands. We classify them according to the
The EROS collaboration: Observation of periodic variable stars
7
classical Cepheid
s-Cepheid
miscellaneous pulsating
RR Lyrae ab
RR Lyrae c
eclipsing binary EA
eclipsing binary EB
eclipsing binary EW
miscellaneous
Semiregular
Mira
long period variable LPV
Fig. 6. Phase diagrams (or light curves for the two lowest panels) for typical variable stars of our catalogue in RE
magnitude.
following criteria. Algol systems (130 stars, type
EA) display well-defined eclipses whose secondary
one has a depth lower than half the primary one,
and possibly flat light curve between them. EB
type objects (35 in total) show a secondary eclipse
equal to half the primary one. The EW type (11
objects) is characterised by similar depths of the
two eclipses. The members of a residual sample of
196 objects do not look like convincing eclipsing bi-
naries and are catalogued as miscellaneous variable
stars.
As emphasised by Udalski et al. (1999a) a large num-
ber of variable objects show small amplitude sinusoidal
variations, such as ellipsoidal binary variable stars.
A contamination of the sample of pulsating stars by
eclipsing binaries is thus possible.
Figures 8 and 9 show the location of the selected vari-
ables in the colour-magnitude diagrams. Also plotted are
10,000 stars located in the central part of the two fields
8
The EROS collaboration: Observation of periodic variable stars
1.2
1
0.8
0.6
0.4
0.2
0
8
7
6
5
4
3
2
1
0
1
1
10
10
Fig. 7. Distinction between classical Cepheids, s-Cepheids
and miscellaneous pulsating stars in the R21 − P and
Φ21− P planes. The individual uncertainties are reported.
The cloud of points represent the Cepheids observed in the
LMC (adapted from Afonso et al. (1999)). The curve cor-
responds to the empirical function Rcut
21 (P ) = 0.4− (P/30
days) used to distinguish between s and classical Cepheids.
tm550 and gn450. Most of the Cepheids are much brighter
than our magnitude threshold (cut 2); this is not so for
RR Lyrae (see Fig. 9, lower panel). Our catalogue is thus
not complete for this type of variable stars, as already
mentionned.
4. The catalogue
The catalogue is composed of two tables containing ob-
jects with periods P smaller or larger than 30 days, respec-
tively. The identifier of each star is given according to the
recommendations of the IAU Commission 5 in The Rules
and Regulations for Nomenclature (see the Annual Index
of A&A). The general acronym used in the catalogue is
EROS2 GSA followed by J2000 equatorial coordinates
in the format JHHMMSS±DDMMSS. The remainder of
the identifier in parentheses gives some information relat-
ing to the internal organisation of the EROS database :
gnnnn or tmnnn is the name of the field, followed by
the CCD number and the location on the image follow-
ing the EROS II nomenclature. The remaining number
is the star identifier used in the EROS database. As an
example, J132630-630945(tm5504m12359) is the name of
the 12359th star observed in quarter m of CCD 4 in the
field tm550. The J2000 equatorial coordinates of this star
are 13:26:30.11, -63:09:45.66.
The equatorial coordinates (J2000) of individual stars
have been obtained as follows. First, we have inserted the
suitable WCS keywords into the header of the EROS II
reference images using the WCSTools package (Mink
1999). Whenever possible, the cross-identification of each
star with previously known objects within a 10′′ search
radius has been done using the Simbad and Vizier
databases available at the CDS, Strasbourg.
The tables contain the following information :
1. Identifier
2. Right ascension α (J2000)
3. Declination δ (J2000)
4. hREi mean magnitude in EROS-red
5. ∆RE amplitude peak to peak in RE
6. hVEi mean magnitude in EROS-visible
7. ∆VE amplitude peak to peak in VE
8. Period in days. Note that periods longer than 30 days
are given with less accuracy since the time span of the
measurements does not allow a precise determination.
Measured periods which are longer than 60 days (i.e
2/3 of the observation period) are flagged by writing
"P > 60 d" and have no warranty to be true periodic
variable stars. The peak to peak amplitude of these
stars could be meaningless and the mean magnitude is
determined with low accuracy.
For Cepheids and RR Lyrae the results of the Fourier fit
are given :
9. Fourier coefficient ratio R21;
10. Fourier coefficient ratio R31;
11. Phase difference φ21 (in rad);
Also given when possible :
13. Type of variability (C=classical Cepheids, S=s-
Cepheids, puls.= miscellaneous pulsating
stars,
EA,EB,EW=eclipsing binaries, misc=miscellaneous
variable stars, SR=Semi-Regular variables, M=Miras,
LPV=long period variables);
14. Name of cross-identified object(s) within a search ra-
dius of 10′′.
The catalogue is planned to be installed at the CDS (see
also our Web site http://eros.in2p3.fr/).
A by-product of such a survey is the possibility to
update the coordinates given in older catalogues. As
an example, it was found that the well known classical
Cepheid OO Cen with the SIMBAD identifier V*OO Cen
was 20" away from an EROS object, namely J132630-
630945(tm5504m12359), as shown in the finding chart of
Fig. 10. The period and coordinates we report are compat-
ible with the ones given by the General Catalogue of
Variable Stars (GCVS)2 (Kholopov 1985) (see Table
4). It seems that an error occurred when the coordinates
of this particular star were filled in the SIMBAD. The
light curve of this object is shown in Fig 6.
We have performed several cross-identifications of
our catalogue with those previously available, namely
2 Queries
on
this
catalogue
are
possible
at
http://www.sai.msu.su/cgi-bin/wdb-p95/gcvs/stars/form
The EROS collaboration: Observation of periodic variable stars
9
Table 4. Period and coordinates obtained on OO Cen by
SIMBAD, GCVS and this study.
Source
object
P (days)
α (B1950)
δ (B1950)
GCVS
SIMBAD
V* OO Cen OO Cen
12.8805
n/a
13:23:09
13:23:06.85
-62:54:24.9
-62:54.0
This study
J132630-630945(tm5504m12359)
12.894
13:23:08.8
-62:54:11
was not operational at the time of the observations.
Therefore the comparison can only be carried out on 0.7
square degree. Furthermore, as pointed above, this com-
parison is restricted to stars with magnitude IC < 14. A
total of 118 EROS objects, 6 GCVS and 23 CKS91 ob-
jects lie in this region. Three objects are common to the
GCVS and CKS91 catalogues one. The overlap between
the three catalogues represents only 19 objects. Among
them one finds the most interesting ones, such as Cepheids
OO Cen and V881 Cen (see Table 5). Some objects show
a large difference between the two magnitude determina-
tions. These are long period variables for which the mean
magnitude is measured on only a part of the whole period,
and thus ill determined in both surveys. Only 7 known
variable stars are not recovered by our analysis (see Table
6). Conversely, 99 objects of our catalogue are not listed
by CKS91. All of them are labelled Long Period Variable
stars (LPV).
Table 6. Previously known variable stars not recovered
by this survey. For each of them we give (where possible)
their type of variability, their period and the reason for
their absence in the catalogue.
Object
HQ Nor
HY Nor
Type P(days)
EB
Mira
90.9
236
catalogue
GCVS
GCVS
8.4860
UW Nor
13214-6256
13218-6254
13248-6249
13232-6249
EA
LPV
LPV
LPV
LPV
GCVS
CKS91
CKS91
CKS91
CKS91
comment
too long period
close to
the CCDs gap
fails cut 6
fails cut 6
fails cut 6
fails cut 6
fails cut 6
Fig. 10. Finding chart of the star OO Cen - J132630-
630945(tm5504m12359). The position given by the
SIMBAD database is shown by the square which is 10"
wide, while the EROS position is shown by the circle.
the IRAS Point Source Catalog (Beichman et al.
1998), the MSX5C Infrared Astrometric Catalog
(Egan et al. 1996), the Two Micron All Sky Survey
(2MASS) (Skrutskie et al. 1997), the CKS91 catalogue
(Caldwell et al. 1991) and the General Catalogue of
Variable Stars (GCVS) (Kholopov 1985). A total of 38
IRAS sources and 220 MSX5C objects have been thus re-
trieved. The overlap with the available 2MASS catalogue
exists only for the EROS field gn459, representing 255
stars. A total of 233 2MASS objects have been thus re-
trieved among them 37 objects classified as Semi-Regular
variables.
An overlap exists with the CKS91 catalogue (Caldwell
et al. 1991). These authors have searched for bright
Cepheids and other variable stars with I < 14, to-
wards Crux and Centaurus, during 42 days with less than
10 measurements per star. A small overlap exists be-
tween this survey and the GSA fields tm550 and tm551.
Unfortunately this overlap involves our CCD #2 which
10
The EROS collaboration: Observation of periodic variable stars
7
9
11
13
15
17
19
21
-1
7
9
11
13
15
17
19
21
-1
0
1
2
3
4
5
6
7
8
0
1
2
3
4
5
6
7
8
Fig. 8. Colour-magnitude diagrams (IC vs VJ -IC ) for the Miras (represented by dots •), Semi-Regular variable stars
(diamonds ⋄) and eclipsing binaries (open circles ◦). The dotted line corresponds to cut 2 on the luminosity of the
stars.
The EROS collaboration: Observation of periodic variable stars
11
7
9
11
13
15
17
19
21
-1
7
9
11
13
15
17
19
21
-1
0
1
2
3
4
5
6
7
8
0
1
2
3
4
5
6
7
8
Fig. 9. Colour-magnitude diagrams (IC vs VJ -IC ) for the classical Cepheids (represented by stars ⋆), s-Cepheids (dots
•), and RR Lyrae (open circles ◦).
12
The EROS collaboration: Observation of periodic variable stars
Table 5. Previously known variable stars recovered by this survey. For each star we give its name, the original
catalogue, the type of variability, the period and the I magnitude given (if any) in the catalogue and in this survey.
Object
OO Cen
Catalogue Type P(days)
12.8805
GCVS
C
I
9.96
V881 Cen
GCVS
C
-
10.57
V608 Cen
GCVS
EB
1.6287
12.95
CKS91
13246-6259
CKS91
13297-6244
CKS91
13206-6305
CKS91
13211-6300
CKS91
13214-6306
CKS91
13214-6256
CKS91
13215-625
CKS91
13219-6301
CKS91
13229-6252
CKS91
13230-6251
CKS91
13238-6227
CKS91
13238-6254
CKS91
13246-6306
CKS91
13247-6244
CKS91
13264-6256
CKS91
13289-6308
CKS91
EA
CKS91
E
CKS91
SR
CKS91
LPV
CKS91
CKS91
SR
SR
CKS91
LPV
CKS91
LPV
CKS91
LPV
CKS91
LPV
CKS91
LPV
CKS91
LPV
CKS91
CKS91
M
SR
CKS91
LPV
CKS91
M
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
13.29
13.05
1
12.35
11.73
10.51
12.85
10.43
11.81
12.51
12.48
12.63
12.21
12.46
12.66
12.37
12.65
EROS ID
J132630-630945
(tm5504m12359)
J132721-630110
(tm5503l3554)
J132948-630634
(tm5505m13922)
J132758-631449
(tm5505l9598)
J133311-630023
(tm5511m5801)
J132400-632057
(tm5504l1003)
J132427-631542
(tm5504k5591)
J132444-632224
(tm5504l7786)
J132447-631201
(tm5504k8841)
J132454-631437
(tm5504k9938)
J132519-631720
(tm5504l13138)
J132433-632201
(tm5504m10330)
J132625-630645
(tm5504m11603)
J132714-634338
(tm5507l1058)
J132719-633644
(tm5507l1752)
J132803-632200
(tm5505l10106)
J132813-630013
J132813-630013
J132948-631227
(tm5510n4312)
J133222-632339
(tm5513k12129)
P(days)
12.8941
Ic
10.20
15.2278
10.97
1.7601
12.05
5.4918
13.03
0.2141
13.20
45.83
12.34
P > 60
11.75
55.00
10.79
52.39
12.77
P > 60
10.23
P > 60
13.00
58.46
11.96
P > 60
12.41
P > 60
10.78
P > 60
11.91
P > 60
13.99
44.80
12.72
P > 60
12.56
P > 60
12.65
The EROS collaboration: Observation of periodic variable stars
13
5. Discussion
The motivation of our search was to improve our knowl-
edge of the distance distribution and of the extinction of
the microlensing source stars used in papers I and II. In
the following we use the RR Lyrae which are well-known
distance indicators and have been observed in all six direc-
tions that we investigated. The GSA fields having a high
non-uniform absorption, we give only an average redden-
ing towards our fields, based on EROS data alone.
0
0
For each selected RR Lyrae we estimate the extinc-
tion A(V ) = 2.4 × E(V − I) using the standard extinc-
tion coefficients (Schlegel et al. 1998; Stanek 1996). The
colour excess E(V − I) is derived from the colour that we
measured and their intrinsic colour : (V − I)RRab
= 0.4
and (V − I)RRc
= 0.2 (Alcok et al. 1998). The uncer-
tainty on individual extinctions is estimated to be 0.5mag.
This error includes the 0.2mag uncertainty on the EROS
colour measurement (see Eq. (1)) and a 0.1mag magni-
tude uncertainty on the intrinsic colour. The absolute
magnitude of RR Lyrae is M RR
V ∼ 0.71 with a precision
of 0.07mag (Gould and Popowski 1998). Finally, we esti-
mate the distance to each star simply by using the relation
m− M = 5× (log(d) − 1) + A. The typical uncertainty on
the distance is 20%. The left panels of Fig. 11 show the
obtained extinction A(V ) versus the calculated distance.
The mean extinction of the RR Lyrae is A(V ) ∼ 3.6 ± 0.2
magnitude towards θ Mus and 3.3 ± 0.1 towards γ Nor,
with a dispersion of 1 magnitude. The mean distance of
the RR Lyrae is ∼4.7± 0.3 kpc towards θ Mus and 5.3±0.2
kpc towards γ Nor. The dispersion of the values is 1.4 kpc
which reflects the spread in distance of disc stars. The disc
population contributes to the sources of the microlensing
events that we observe in the Galactic plane. A model
based on the distribution of matter in the disc and the
luminosity function of neighbooring stars has been used
in Derue (1999) to estimate the distance of the star popu-
lation from the disc. The distance of RR Lyrae is in good
agreement with the one obtained with this model.
6. Conclusion
In the course of our program dedicated to microlens-
ing events, we have devoted a fraction of observing time
to the search for variable stars in six directions of the
Galactic plane. This exploratory campaign, that lasted
three months, led to the discovery of 1,362 variable stars.
Among them we identified 9 Cepheids, 19 RR Lyrae, 34
Miras, 176 eclipsing binaries and 266 Semi-Regular vari-
able stars. We have set up a catalogue of all of the
1,362 stars and cross-identified it with several other cat-
alogues. In particular a comparison with the GCVS and
the CKS91 catalogues shows that only a small fraction
(∼15%) of the objects that we have identified appear in
those two. Among the stars most appropriate to be used
as distance indicators, the Cepheids turned out to be too
few to warrant a particular study. As far as RR Lyrae are
concerned, we have determined their mean distance and
7
6
5
4
3
2
1
7
6
5
4
3
2
1
1
0.8
0.6
0.4
0.2
0
2
1.75
1.5
1.25
1
0.75
0.5
0.25
0
Mean
RMS
4.687
1.425
2
4
6
8
10
Mean
RMS
5.333
1.403
2
4
6
8
10
2
4
6
8
10
2
4
6
8
10
Fig. 11. Extinction A(V ) versus the distance of RR
Lyrae(in kpc) towards θ Mus and γ Nor. The right panels
show the inferred distance distribution.
found it to be ∼5 kpc. Yet the statistics being quite lim-
ited, we are considering pursuing this effort by launching
a longer search for variable stars based on the results of
this first campaign.
Acknowledgements. The WCSTools package was made avail-
able to us thanks to the work of Doug Mink, NASA/GSFC,
Harvard. The Skycat/Gaia tool is the result of a joint effort
by the computer staff of ESO Garching and from the Starlink
Project, UK.
References
Afonso, C., Albert,
J.N, Amadon, A.
CEA Saclay
1999,
et
al.
Report
http://www-
coll.),
(EROS
Dapnia–SPP
dapnia.cea.fr/Spp/Experiences/EROS/Cepheides/catalog cep.html.
99/25.
also
See
Alcock, C., Allsman, R.A., Alves, D.R. (MACHO coll.),
1998, ApJ, 492, 190.
Ansari, R., 1996, Vistas in Astronomy 40, 519.
Antonello, E., Poretti, E., 1986, A&A, 169, 149.
Beaulieu J-P., Grison P., Tobin W. et al. (EROS coll.),
1995, ApJ 303, 137.
Beaulieu J-P. & Sasselov D., 1996, Variable Stars and the
Astrophysical Returns of Microlensing Surveys, eds. R.
Ferlet, J.P. Maillard, B. Raban, p. 193.
Beichman, C.A., Neugebauer, G.A., Habing, H.J. et al.,
Explanatory Supplement to the IRAS Catalogs and
Atlases, Eds. 1988.
Caldwell, J., Keane, M., Schechter, P., 1991, AJ 101, 1763.
Cioni M-R.L., Marquette J-B., Loup C. et al., 2001, ac-
cepted by A&A, astro-ph/0109014.
14
The EROS collaboration: Observation of periodic variable stars
Derue, F., 1999, Ph.D. thesis, CNRS/IN2P3, LAL re-
port 99-14, Universit´e Paris XI Orsay. also available at
the URL http://www.lal.in2p3.fr/presentation/ biblio-
theque/publications/Theses99.html
Derue, F., Afonso, C., Alard, C. et al. (EROS coll.), 1999,
A&A 351, 87.
Derue, F., Afonso, C., Alard, C. et al. (EROS coll.), 2001,
A&A 373, 126.
Egan, M.P., Price, S.D., Moshir, M.M. et al., 1999, AFRL-
VS-TR-1999-1522.
Fouqu´e, P., Chevallier, L., Cohen, M. et al., 2000, A&AS
141, 313.
Georgelin, Y.M, Amram, P., Georgelin, Y.P. et al., 1994,
A&AS 108, 513.
Gould, A., Popowski, P., 1998, ApJ, 508, 844.
Grison, P., 1994, A&A, 289, 404.
Grison P., Beaulieu J-P., Pritchard J-D. et al. (EROS
coll.), 1995 A&AS 109, 447.
Harris J., Zaritsky D., 1999, AJ 117, 2831.
Kholopov, P.N, 1985 In : General Catalogue of Variable
Stars, vol. 1, Moscow Nauka Pub. House.
Madore, B.F. & Freedman, W.L., 1991, PASP 103, 933.
Mansoux B., 1997, Ph.D. thesis, CNRS/IN2P3, LAL re-
port 97-19, Universit´e Paris 7.
Mink D.G., Astronomical Data Analysis Software and
Systems VIII, A.S.P. Conference Series, 1999, Dave
Mehringer, Ray Plante, Doug Roberts, eds., p. 498.
Paczy´nski, B., 1996, astro-ph/9608094, invited talk at the
Extragalactic Distance Scale STScI Symposium, May
7-10, 1996, Baltimore, Maryland, USA.
Palanque-Delabrouille, N., Afonso, C., Albert, J-N. et al.
(EROS coll.), 1998, A&A 332, 1.
Regnault, N., 2000, PhD thesis, Universit´e Paris VII,
LAL-CNRS/IN2P3 Report 00/65. also available at
the URL http://www.lal.in2p3.fr/presentation/ biblio-
theque/publications/Theses00.html
Russeil, D., Amram, P., Georgelin, Y.P. et al., 1998,
A&AS 130, 119.
Scargle, J.D., 1982, ApJ 263, 835.
Schwarzenberg-Czerny, A., 1996, ApJL 460, 107.
Schlegel, D.J., Finkbeiner, D.P., Davies, M., 1998, AJ,
500, 525.
Skrutskie, M.F., Schneider, S.E., Stiening, R. et al.
(2MASS Coll), 1997, Proc. Workshop "The Impact of
Large Scale Near-IR Sky Surveys", 25.
Stanek, K.Z., 1996, ApJL, 460, 37.
Udalski, A., Szyma´nski, M., Kubiak, M. et al. (OGLE
Coll.), 1998, Act. Astr. 48, 1.
Udalski, A., Soszynski, I., Szyma´nski, M. et al. (OGLE
Coll.), 1999, Act. Astr. 49, 45.
|
astro-ph/9412058 | 1 | 9412 | 1994-12-16T13:47:56 | On the Origin of Low-Surface-Brightness Galaxies | [
"astro-ph"
] | We reconsider the problem of the origin of Low-Surface-Brightness (LSB) galaxies within the standard Cold Dark Matter (CDM) cosmological model ($\Omega =1, h=0.5$). Taking into account the effect of substructure on the collapse times of low overdensity peaks, we show that the abundance of these peaks is larger than previously expected because of the dragging caused by dynamical friction induced by the presence of small-scale substructure. The number density of these shallow, low-density peaks turns out to be in good agreement with the observed overdensity of the blue LSB galaxies found during recent surveys. | astro-ph | astro-ph | ON THE ORIGIN OF
LOW-SURFACE-BRIGHTNESS
GALAXIES
V. ANTONUCCIO-DELOGU
Osservatorio Astro(cid:12)sico di Catania
Citt(cid:18)a Universitaria - Viale A. Doria
Catania ITALY
Theoretical Astrophysics Center
Blegdamsvej , DK Copenhagen, DENMARK
Accepted for publication in Astrophys. Letters and Comm.
Abstract { We reconsider the problem of the origin of low-surface-brightness
(LSB) galaxies within the standard Cold Dark Matter (CDM) cosmological model
((cid:10) = ; h = :). Taking into account the e(cid:11)ect of substructure on the collapse times
of low overdensity peaks, we show that the abundance of these peaks is larger than
previously expected because of the dragging caused by dynamical friction induced
by the presence of small-scale substructure. The number density of these shallow,
low-density peaks turns out to be in good agreement with the observed overdensity
of the blue LSB galaxies found during recent surveys.
key-words { galaxies { galaxy formation { cosmology
. INTRODUCTION
The problem of the existence of a signi(cid:12)cant amount of LSB galaxies (here-
after LSBs'), has been around since almost thirty years. Hubble ( ) and De
Vaucouleur ( ) (cid:12)rst pointed out that photographic surveys could be be bi-
ased against low-surface-brightness ob jects. The argument was taken up more
recently by Disney ( ), who added new estimates of the relevanece of this
bias on the completeness of the existing galaxy surveys. After these claims
many people tried to develop new techniques to increase the sensitivity of
photographic materials toward low-surface brightness levels. A new technique
introduced by D. Malin allowed the serendipiditous discovery of two spectac-
ular ob jects, namely Malin I and II (Bothun et al., , ). The total HI
mass of Malin is about
M
, and its disk scale-length is (cid:25) Kpc, making
(cid:12)
it one of the largest and shallowest spiral galaxy within the local neighborhood.
Nonetheless, it could have been hardly detected at optical wavelengths due to
its very low central surface brigthness (only : B-mag/arcsec
, more than
(cid:27) away from the Freeman's value of : which characterizes bright spirals).
Although Malin I and II are exceptional ob jects under many respects, more
than years of e(cid:11)orts due to the Cardi(cid:11) group (M. Disney, S. Phillips, D.
Davies and collaborators) and by few other people in the USA and in Australia
(most notably D. Malin, G. Impey, G. Bothun, S. McGaugh, J. Schombert)
has led to the discovery of a large amount of LSBs in clusters like Hydra and
in the (cid:12)eld. These surveys have shown that, although ob jects as large as Malin
I and II are rare, LSBs' with photometric properties more similar to those of
normal galaxies but with a much smaller central surface brightness are very
common. Due to the intrinsic faintness of these ob jects it has become possible
to measure the redshifts and metallicity for some of them only recently, after
the development of CCD detectors (van der Hulst et al., ). These (cid:12)rst
results show that these galaxies have lower-than-solar metallicities and almost
no color gradients across the disc. Moreover, their HI surface density is well
below the critical value for star formation in normal spirals found observa-
tionally by Kennicut ( ): this could explain the low luminosity of these
ob jects. Although redhifts are known only for few LSBs', the intrinsic faint-
ness of these ob jects strongly suggests that they cannot be detected if they
were at very high redshifts. The low metallicity suggests that any interpreta-
tion based on a fading mechanism could meet serious theoretical di(cid:14)culties,
so we are left with the problem of explaining how galaxy formation has been
delayed until recently, compared to normal spiral galaxies.
. DENSITY PEAKS IN STANDARD CDM.
In this paper we will look at the predictions of a very popular cosmological
model, namely the standard CDM, concerning the origin of LSBs'. This model
explains quite sucessfully the observed clustering properties of the Universe
from cluster up to MWBR scales but, as many other cosmological models, it
runs into trouble when it tries to predict the observed photometric properties
of galaxy populations. Stellar formation mechanisms, largely unknown by to-
day, can hinder a signi(cid:12)cant fraction of the information contained in the initial
conditions.
It has been recently suggested that environmental e(cid:11)ects like, for instance,
tidal (cid:12)elds, can play a signi(cid:12)cant role on star formation e(cid:14)ciencies, and ulti-
mately on the observed properties of normal, bright galaxies (Lacey and Silk,
) The low star formation rate of LSBs' suggests that these ob jects could
have been poorly a(cid:11)ected by their environment, so that their photometric
properties could be more directly related to the statistical properties of the
density peaks from which they originated. Moreover, LSBs are found in the
(cid:12)eld and within the central regions of rich clusters like Coma, within envi-
ronemnts spanning at least two order of magnitudes in density, and they seem
to have the same clustering properties as bright normal spirals, although with
a smaller amplitude (Mo et al., ). All these facts point to suggest that
LSBs could be much lesser biased tracers of the post-recombination density
(cid:12)eld than normal bright spirals.
One of the main predictions of the idealized spherical collapse model (Gunn
and Gott, ) is that only density perturbations having an initial overden-
sity after recombination larger than a critical value (cid:14)
((cid:25) : for a stan-
c
(GG)
dard CDM) can turn around within lesser than one Hubble time. However,
this model does not take into account the e(cid:11)ect of the presence small-scale
substructure. One of the distinctive features of the standard CDM model
is a very (cid:13)at power spectrum over a wide range of subgalactic mass scales
(
(cid:20) M (cid:20)
M
). Physically this means that the Universe is very lumpy,
(cid:12)
and (cid:12)lled with many small-scale subclumps on these scales. In a recent series of
papers (Antonuccio-Delogu, Colafrancesco and Del Popolo, ; Antonuccio-
Delogu and Colafrancesco, ) we have shown that the presence of substruc-
ture induces a delay in the collapse, so that the critical overdensity for collapse
is rather given by:
(GG)
(cid:21)
(cid:16)
(cid:17)
(cid:14)
= (cid:14)
+
()
c
c
(GG)
(cid:0) (cid:22)
(cid:14)
C
Here (cid:21)
is a coe(cid:14)cient which ultimately depends on the power spectrum, and
(cid:22)(x) is a known function. The dynamical friction a(cid:11)ects more seriously the
collapse of the small overdensity peaks, and it alters the statistical properties
of the density (cid:13)uctuation (cid:12)eld.
This latter point can be understood by looking at the properties of the selection
function t ((cid:23) ). Following Bardeen et al. ( ) we have de(cid:12)ned this as the
probability that an average density (cid:13)uctuation, selected with some prescribed
(cid:12)lter on the scale R
, will collapse on a cosmological time scale:
f
Z
(cid:23)
LSB
h
i
(cid:22)
t ((cid:23) ) =
d(cid:26)p
(cid:26); h
(cid:14) i (R
)
()
f
(cid:14)
c
where the probability distribution is given by the standard CDM (see e.g.
Table : Number densities of LSBs for di(cid:11)erent values of the parameters of
the standard CDM. Columns are as follows: ) (cid:12)ltering radius (in units of
)
Kpc, ) (cid:12)ltering mass (in
M
), ) normalization constant, ) upper limit
(cid:12)
of central overdensity, ) number density (Mpc
), ) ratio of LSBs to total
(cid:0)
peaks' density.
R
M
b
(cid:23)
n
n
=n
f
f
LSB
LSB
LSB
tot
:
:
:
: (cid:2)
:
(cid:0)
"
"
"
:
: (cid:2)
:
(cid:0)
"
"
.
: (cid:2)
:
(cid:0)
"
"
.
:
: (cid:2)
:
(cid:0)
:
:
:
: (cid:2)
:
(cid:0)
"
"
"
:
: (cid:2)
:
(cid:0)
:
:
: (cid:2)
:
(cid:0)
"
"
"
:
: (cid:2)
:
(cid:0)
Lilje and Lahav, ). It is easy to verify that the selection function has the
correct asymptotic behaviour in both limits (t ((cid:23) ) ! for (cid:23) ! , t ((cid:23) ) !
for (cid:23) ! ).
If we now identify the progenitors of LSBs' with peaks of the density (cid:12)eld
having (cid:23) (cid:20) (cid:23)
, their number density will be given by:
LSB
Z
Z
(cid:23)
(cid:23)
LSB
LSB
(cid:23)
"
#
n
=
t ((cid:23) ) N
((cid:23) ) =
d(cid:23) t ((cid:23) ) (cid:1)
exp
(cid:0)
G ((cid:13) ; (cid:13) (cid:23) )
()
LSB
peak
(cid:25)
R
(cid:3)
where the quantities in the last identity are standard within the CDM model
(see e.g. Bardeen et al., ).
. RESULTS AND CONCLUSIONS.
In Table I we show the results of applying eq. () for di(cid:11)erent values of R
; b
f
(the normalization constant of standard CDM) and (cid:23)
, the maximum as-
LSB
sumed central overdensity of those peaks giving origin to LSBs. Although
the observed number density of LSBs is largely uncertain due to the incom-
pleteness of the present surveys, recent estimates give values in the range
: (cid:20) n
=n
(cid:20) : (McGaugh, ), in good agreement with the numbers
LSB
tot
in the last column of Table I .
We are aware of the risks of a one-to-one identi(cid:12)cation of density peaks with
observed galaxy populations.
In addition to stellar formation mechanisms,
other features like angular momentum distribution in protogalactic clouds can
induce a large variance in the observed properties. However, as we have al-
ready stressed, there are reasons to believe that environmental e(cid:11)ects played
a lesser important role on the formation of LSBs, in comparison with normal
spirals. The dynamical friction induced by small-scale substructure can be
seen as part of the \cross-talk" among di(cid:11)erent scales (for a review see, e.g.,
Silk and Wyse, ). At variance with other cross-talk mechanisms it is com-
pletely speci(cid:12)ed once the power spectrum is given, and does not depend on
star formation. It can then be regarded as a standard ingredient of the CDM
cosmological scenario.
Acknowledgements { I would like to express my gratitude to the referee
(G. De Vaucouleur) for pointing out two references (Hubble and De
Vaucouleur ) and for other suggestions. This work was partly supported
by Danmarks Grundforskningsfond through its support for the establishment
of the Theoretical Astrophysics Center (TAC).
References
[] Antonuccio-Delogu, V. and Colafrancesco, S., , Asrophys. J. ,
[] Antonuccio-Delogu, V., Colafrancesco, S. and Del Popolo, A., ,
preprint
[] Bardeen, J. M., Bond, J. R., Kaiser, N. and Szalay, A.S., , Astrophys.
J. ,
[] Bothun, G.D., Impey, C.D., Malin, D. and Mould, J., , Astron. Journal
,
[] Bothun, Schombert, J.M., G.D., Impey, C.D. and Schneider, S.E., ,
Astrophys. J. ,
[] Disney, M.J., , Nature ,
[] Gunn, J.E. and Gott, J.R., , Astrophys. J. ,
[] Hubble, E., , Astrophys. J. ,
[ ] Kennicut, R.C., , Astrophys. J. ,
[ ] van der Hulst, J.M., Skillman, E.D., Smith, T.R., Bothun, G.D., Mc-
Gaugh, S.S. and de Blok, W.J.G., , Astron. J., , preprint
[] Lacey, C. and Silk,
[] McGaugh, S.S., , in Quantifying Galaxy morphology at High Redshift,
HST workshop, Baltimore MD, April - ( )
[] Lilje, P.B. and Lahav, O., , Astrophys. J. ,
[] Mo, H.J., McGaugh, S.S. and Bithun, G.D., , preprint
[] Silk, J. and Wyse, R.F.G., , Phys. Reports ,
[] De Vaucouleur, G., , Astron. Journal ,
|
astro-ph/9905231 | 1 | 9905 | 1999-05-18T14:09:55 | Brown Dwarfs: Toward a New Age Scale for Young Open Cluster | [
"astro-ph"
] | We review our the status of the photometric searches of brown dwarfs and vely low mass stars in several young open clusters and present some results on the spectroscopic follow-up, including new cluster ages based on the position of the lithium depletion boundary. | astro-ph | astro-ph |
BROWN DWARFS AND VERY LOW MASS STARS: TOWARDS
A NEW AGE SCALE FOR YOUNG OPEN CLUSTERS
D. BARRADO Y NAVASCU´ES
Max-Planck Institut fur Astronomie
Konisgtuhl 17, D-69117 Heidelberg, Germany
J.R. STAUFFER
Harvard-Smithsonian Center for Astrophysics
60 Garden St. Cambridge, MA02138, USA
AND
J. BOUVIER
Observatoire de Grenoble
F-38041 Grenoble, Cedex 9, France
1. Introduction
Open clusters are one of the main available laboratories in Astronomy.
Among other things, the stellar evolutionary age scale relies primarily on
them, so it is extremely important to estimate accurately the ages of dif-
ferent associations. Until now, the dating of clusters was based on either
the Upper Main Sequence (MS), using stars evolving off it or already well
beyond the turn -- off point, or isochrone fitting. These techniques are model
dependent and provide different ages for the same cluster. Very Low Mass
Stars (VLM) and Brown Dwarfs (BDs) offer an alternative: It is well known
that the lithium abundance is depleted with the age in stars (see the review
by Balachandran 1994). This destruction depends also on mass, so dM stars
deplete their lithium even before they arrive to the MS. However, the in-
ternal part of VLM/BDs is not hot enough to destroy lithium at this stage.
In the lower part of the cluster MS, the border between members having
no lithium and those showing it in their atmosphere is called the lithium
depletion boundary (LDB). As a cluster gets older, the LDB moves toward
less massive objects. Therefore, the detection of lithium in the brightest
(e.g. bluer, hotter and more massive than others) BDs of a cluster puts a
very strong limit on its age (D'Antona & Mazzitelli 1994; Mart´ın & Montes
2
14
16
18
20
22
0
D. BARRADO Y NAVASCU´ES ET AL.
12
14
16
18
20
22
IC2391, Simon&Patten
IC2391, Pavlovsky&Patten
IC2391, this survey
.5
1
1.5
2
2.5
3
.8
1
1.2
1.4
1.6
1.8
2
2.2
2.4
2.6
2.8
Figure 1. a Color-Magnitude Diagram for the area around IC 2391. b Our selection of
possible members of IC 2391.
1997; Ventura et al. 1998). In this paper, we show how this technique works
and apply it to several young open clusters.
2. Identification of candidate members: Optical and IR surveys
The initial step in order to identify new very low mass (VLM) stars and
BDs has been to obtain optical photometry on the area around the cluster.
In order to cover a significat fraction of a particular cluster in a reasonable
amount of time, we have used in most of the cases detectors with several
CCDs. Our targets have been:
i) The Pleiades.- We have used the CFHT (with the MOSAIC camera)
and 48" telescope at MHO. We surveyed 2.5 and 1 sq. deg., respectively,
discovering 17 and 6 BDs candidates in each survey (Bouvier et al. 1998;
Stauffer et al. 1998a).
ii) α Per.- As a preliminary survey, we collected RI photometry with the
48" telescope at MHO. We discovered several candidates. Some of them
were the targets of a spectroscopic campaign (next section).
iii) IC 2391.- The CTIO 4m telescope and the BTC camera allowed us to
survey 2 sq. deg., detecting ∼40 BDs candidates (Barrado y Navascu´es et
al. 1999a). See Fig. 1a.
iv) M35.- CFHT+MOSAIC: This is a very well populated open cluster.
Our study included several thousands candidates (stars) and few possible
BD (Barrado y Navascu´es et al. 1999b).
v) NGC 2516.- CTIO 4m telescope + BTC. We covered 0.6 sq. deg., analysis
in progress.
Since most of the lists of cluster candidates contain spurious members,
it is necessary to select good candidates by obtaining near IR photometry.
NEW AGE SCALE FOR YOUNG OPEN CLUSTERS
3
4000
Per32
Ic=17.6
3000
y
t
i
s
n
e
t
n
I
2000
1000
Pleiades
14
12
10
8
6400
6600
6800
7000
7200
7400
7600
50
100
Wavelength
150
Age (Myr)
200
250
Figure 2. a Spectrum of a new member of α Per. b MI versus age. The location of the
lithium depletion boundary is shown for the Pleiades and α Per.
We have collected this information using the MHO 48" telescope and the
NASA IRTF, in the case of the Pleiades. JHK photometry on the area
around IC 2391 was obtained by the 2MASS project. All this information
allowed us to construct robust lists of possible members. Figure 1b displays
a Color-Magnitude diagram of IC 2391, and a comparison with previous
surveys. A 30 Myr isochrone (solid line) and a Zero Age MS (dashed line)
are included.
3. Spectroscopy: membership and the new age
Since these objects are too faint to have been detected with photographic
plates in the past, most of them do not have measurements of their proper
motions. Therefore, the final confirmation of membership comes from the
spectroscopy. A intermediate resolution spectra (R=2000 -- 5000) are good
enough to obtain rough radial velocities (hence, establishing whether the
target is a member). On the other hand, a detection of LiI 6708 A reveals
the BD nature of the candidate and also confirms the membership (Rebolo
1991). Using the Keck II telescope and the LRIS spectrograph we have
collected spectra of VLM and BD candidates in the Pleiades and α Per
open clusters. Figure 2a shows the spectrum of PER 32, a BD candidate
discovered in the α Per region with the 48" MHO. The intense Hα line in
emission and its radial velocity indicates that it is a real member of the
cluster. However, we have not detected lithium, which indicates that this
is a VLM star close to the LDB.
Figure 2b illustrates how ages are estimated. Following Stauffer et al.
(1998b), the faintest member without lithium and the brightest member
with it, bracket the magnitude of the LDB. We used the Chabrier & Baraffe
4
D. BARRADO Y NAVASCU´ES ET AL.
IC2391
Pleiades/M35
NGC2516
35 Myr 50 Myr
70 Myr
110 Myr
?
85 Myr
125 Myr
?
Figure 3. Comparison between the standard age scale (top) and the values estimated
from the lithium depletion boundary for BD (bottom).
(1998) latest evolutionary model to covert these magnitudes, different for
the Pleiades and α Per, into their corresponding ages. If the sampling is
well done, the accuracy is better than 10%.
Figure 3 shows the old standard ages for the 5 clusters included in our
study, together the new values for the Pleiades and α Per, which are sig-
nificantly older than the values previously accepted. Our aim is to improve
these values and to get new ones for the other clusters in the near fu-
ture, in order to verify this interesting possibility. This would have a very
important effect on several fields, including the time scale of properties
of low mass stars (lithium abundance, rotation, stellar activity), pre-main
sequence stars, and internal structure and evolutionary models.
References
Balachandran, S., 1994, in "8th Cambridge Workshop on Cool Stars, Stellar Systems,
and the Sun", ed. J.-P. Caillault, APS Conf. Series, vol. 64, p. 234
Barrado y Navascu´es, D., et al. 1999a, ApJ, in preparation
Barrado y Navascu´es, D., et al. 1999b, this volumen
Bouvier, J., Stauffer, J.R., Mart´ın, E.L., Barrado y Navascu´es, D., Wallace, B., B´ejar,
V.J.S., 1998, A&A, 336, 490
Chabrier, G., Baraffe, I. 1998, priv. comm.
D'Antona, F., Mazzitelli, I., 1994, ApJS 90, 467
Mart´ın, E.L., Montes, D., 1997, A&A 318, 805
Rebolo, R, 1991, in IAU Symp. 145, "Evolution of stars: The Photospheric Abundance
Connection", eds G. Michaud & A. Tutukov, Dordrecht, Kluwer, p. 85
Stauffer, J.R., Schild, R.,Barrado y Navascu´es, D., Backman, D.E., Angelova, A., Kirck-
patrick, D., Hambly, N., Vanzi, L., 1998a, ApJ 504, 805
Stauffer, J.R., Schultz, G., Kirkpatrick, J.D., 1998b, ApJ Letter 499, L199
Ventura, P., Zeppieri, A., Mazzitelli, I., D'Antona, F., 1998, A&A 331, 1011
|
astro-ph/0407400 | 1 | 0407 | 2004-07-19T15:44:24 | Simulations of the Population of Centaurs I: The Bulk Statistics | [
"astro-ph"
] | Large-scale simulations of the Centaur population are carried out. The evolution of 23328 particles based on the orbits of 32 well-known Centaurs is followed for up to 3 Myr in the forward and backward direction under the influence of the 4 massive planets. The objects exhibit a rich variety of dynamical behaviour with half-lives ranging from 540 kyr (1996 AR20) to 32 Myr (2000 FZ53). The mean half-life of the entire sample of Centaurs is 2.7 Myr. The data are analyzed using a classification scheme based on the controlling planets at perihelion and aphelion, previously given in Horner et al (2003). Transfer probabilities are computed and show the main dynamical pathways of the Centaur population. The total number of Centaurs with diameters larger than 1 km is estimated as roughly 44300, assuming an inward flux of one new short-period comet every 200 yrs. The flux into the Centaur region from the Edgeworth-Kuiper belt is estimated to be 1 new object every 125 yrs. Finally, the flux from the Centaur region to Earth-crossing orbits is 1 new Earth-crosser every 880 yrs | astro-ph | astro-ph | Mon. Not. R. Astron. Soc. 000, 1–15 ()
Printed 14 March 2018
(MN LATEX style file v1.4)
Simulations of the Population of Centaurs I: The Bulk
Statistics
J. Horner1,2, N.W. Evans2,3 & M.E. Bailey4
1 Physikalisches Institut, Universitat Bern, Sidlerstrasse 5, CH-3012 Bern, Switzerland
2 Theoretical Physics, Department of Physics, 1 Keble Rd, Oxford OX1 3NP
3 Institute of Astronomy, Madingley Rd, Cambridge, CB3 0HA
4 Armagh Observatory, College Hill, Armagh, BT61 9DG, Northern Ireland.
4
0
0
2
l
u
J
9
1
1
v
0
0
4
7
0
4
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
ABSTRACT
Large-scale simulations of the Centaur population are carried out. The evolution of
23 328 particles based on the orbits of 32 well-known Centaurs is followed for up to
3Myr in the forward and backward direction under the influence of the 4 massive
planets. The objects exhibit a rich variety of dynamical behaviour with half-lives
ranging from 540 kyr (1996 AR20) to 32 Myr (2000 FZ53). The mean half-life of
the entire sample of Centaurs is 2.7 Myr. The data are analyzed using a classification
scheme based on the controlling planets at perihelion and aphelion, previously given in
Horner et al (2003). Transfer probabilities are computed and show the main dynamical
pathways of the Centaur population. The total number of Centaurs with diameters
larger than 1 km is estimated as ∼ 44300, assuming an inward flux of one new short-
period comet every 200 yrs. The flux into the Centaur region from the Edgeworth-
Kuiper belt is estimated to be 1 new object every 125 yrs. Finally, the flux from the
Centaur region to Earth-crossing orbits is 1 new Earth-crosser every 880 yrs.
Key words: minor planets, asteroids – planets and satellites: general – celestial
mechanics, stellar dynamics – Kuiper belt
1
INTRODUCTION
The dynamical behaviour of the Centaurs is still poorly un-
derstood. It is possible for a Centaur to work its way slowly
inwards through the outer Solar system, leading to eventual
capture by Jupiter and designation as a short-period comet.
It is also possible for Centaurs to drift outwards to join the
Edgeworth-Kuiper belt, to be ejected from the Solar system
in an encounter with one of the massive, outer planets, or
even to be captured by these planets into temporary satellite
orbits. A small number may even impact upon the planets.
Therefore, the Centaurs potentially hold the key to under-
standing the mechanisms by which the short-period comet
population is maintained, to explaining the distant, retro-
grade satellites of the massive planets, and to allowing us
a glimpse of objects newly introduced to the Solar system
from the Edgeworth-Kuiper belt.
As early as 1990, when the only known Centaur was Ch-
iron, it was realised that such objects may lie on very unsta-
ble orbits. Numerical integrations by Hahn & Bailey (1990)
found that Chiron had a half-life for ejection of around 1
Myr, but that the half-life to become a short-period comet
for the object was around only 200 000 years, implying that
Chiron could well have been a short-period comet in the
past and could possibly become one in the future. This is of
c(cid:13) RAS
particular interest given the size of Chiron (d ≃ 140 − 180
km; Groussin, Lamy & Jorda 2004 and references therein)
and other Centaurs, because objects that large entering the
inner Solar system would be both spectacular and dan-
gerous. In fact, the idea has been mooted that objects of
such size arrive in the inner Solar system with some fre-
quency, and then fragment,
leading to swarms of debris
which have the potential to encounter the Earth. The Kreutz
sun-grazer family may represent one example of such hier-
archical fragmentation, whereas other cases in which comets
of more ordinary size have undergone catastrophic fragmen-
tation include 3D/Biela, D/1994 (Shoemaker-Levy 9) and
C/1994 S4 (LINEAR). Whether such a decay mode repre-
sents a generic process in determining the number of short-
period comets can in principle be tested by examining the
differences in the size distribution of Jupiter-family comets
from those of their probable source objects, namely Cen-
taurs, Edgeworth-Kuiper belt objects, long-period comets
and so on (cf. Lowry, Fitzsimmons & Collander-Brown 2003;
Lamy et al. 2004). Clube & Napier (1984) have suggested
that the Taurid meteoroid swarm may be the relic of the
last large object to undergo such a decay.
Despite the importance of the Centaurs, there has been
little systematic study of the population using numerical
simulations. Early calculations on Chiron (Oikawa & Ev-
2
J. Horner, N.W. Evans & M.E. Bailey
Object
Perihelion
(in au)
Aphelion
(in au)
J
JS
JU
JN
JE
JT
S
SU
SN
SE
ST
U
UN
UE
UT
N
NE
NT
EK
T
E
SP
I
L
4 ∼
4 ∼
4 ∼
4 ∼
4 ∼
4 ∼
6.6 ∼
6.6 ∼
6.6 ∼
6.6 ∼
6.6 ∼
12.0 ∼
12.0 ∼
12.0 ∼
12.0 ∼
22.5 ∼
22.5 ∼
22.5 ∼
33.5 ∼
33.5 ∼
< q ∼
< q ∼
< q ∼
< q ∼
< q ∼
< q ∼
< q ∼
< q ∼
< q ∼
< q ∼
< q ∼
< q ∼
< q ∼
< q ∼
< q ∼
< q ∼
< q ∼
< q ∼
< q ∼
< q ∼
< 6.6
< 6.6
< 6.6
< 6.6
< 6.6
< 6.6
< 12.0
< 12.0
< 12.0
< 12.0
< 12.0
< 22.5
< 22.5
< 22.5
< 22.5
< 33.5
< 33.5
< 33.5
< 60.0
< 60.0
< 4
< 4
< 4
< 4
q ∼
q ∼
q ∼
q ∼
< 12.0
< 22.5
< 33.5
< 60.0
< 22.5
< 33.5
< 60.0
< 6.6
Q ∼
< Q ∼
6.6 ∼
< Q ∼
12.0 ∼
< Q ∼
22.5 ∼
< Q ∼
33.5 ∼
> 60.0
Q ∼
< 12.0
Q ∼
< Q ∼
12.0 ∼
< Q ∼
22.5 ∼
< Q ∼
33.5 ∼
> 60.0
Q ∼
< 22.5
Q ∼
< Q ∼
22.5 ∼
< Q ∼
33.5 ∼
> 60.0
Q ∼
< 33.5
Q ∼
< Q ∼
33.5 ∼
> 60.0
Q ∼
< 60.0
Q ∼
> 60.0
Q ∼
< 33.5
< 60.0
< 60.0
< 4
Q ∼
< Q ∼
4 ∼
< Q ∼
35 ∼
Q ∼
> 1000
< 35
< 1000
Table 1. The classification scheme introduced by Horner et al.
(2003). In the upper table, the first letter designates the planet
controlling the perihelion, the second letter the planet controlling
the aphelion or the region in which the aphelion lies, with the
final two classes EK and T being beyond all the giant planets.
(J = Jupiter, S = Saturn, U = Uranus, N = Neptune, EK =
Edgeworth-Kuiper belt, T = trans-Neptunian belt). The lower
table refers to cometary bodies (E = Encke, SP = short-period,
I = intermediate and L = long-period).
erhart 1979, Hahn & Bailey 1990) and Pholus (Asher &
Steel 1993) identified the chaotic nature of these two ob-
jects, though only small numbers of clones and modest in-
tegration times (< 1 Myr) were used. Work by Dones et
al. (1996) looked at the behaviour of four Centaurs (Chi-
ron, Pholus, Nessus and 1994 TA) and two Jupiter-family
comets (29P/Schwassmann-Wachmann 1 and 39P/Oterma).
They found that the number of surviving objects decays ex-
ponentially during the early part of the integrations, whilst
the decay becomes flatter after a number of life-times. They
also noted that Centaur half-lives inferred from numeri-
cal integrations are smaller than those deduced from ap-
proximations like Opik's (1976) theory and diffusion equa-
tions (e.g., van Woerkom 1948). Levison & Duncan (1997)
ran orbital integrations of 2200 test particles evolving from
the Edgeworth-Kuiper belt to short-period comets, passing
through the Centaur region in the process. The study of the
integrations was mainly focused on the behaviour of the ob-
jects both in the Edgeworth-Kuiper belt and the cometary
region, rather than in the Centaur region itself. Hence, the
dynamics of the Centaur population remains largely unex-
plored.
with q ∼
trol of Saturn. Objects with perihelion distance q ∼
In an earlier paper, Horner et al. (2003) introduced a
new method of classifying objects in the Solar system. This
was based on the idea that the dynamical evolution is largely
determined by the planets that control the perihelion and
the aphelion of the object. This classification is particularly
useful for the Centaurs, since it breaks down the trans-
Jovian region into 20 categories given in the upper panel
of Table 1. Objects are labelled according to the control-
ling giant planet, so, for example, a JS object has perihelion
under the control of Jupiter and aphelion under the con-
> 33.5 au
are designated as either members of the Edgeworth-Kuiper
belt (EK) or trans-Neptunian disk (T). In addition, objects
< 4 au are designated as comets and are subdivided
into Encke types (E), short-period (S), intermediate (I) and
long-period (L), as summarised in the lower panel of Table 1.
The aim of this paper is to provide results from a de-
tailed set of simulations, exploiting the new classification
scheme. The orbits of 32 of the best known Centaurs were
used to create an ensemble of 23 328 clones. The clones were
integrated in the presence of the four massive outer plan-
ets in both forward and backward directions for a period of
three million years, giving a vast data set with which to ex-
amine the dynamics of their orbits. §2 describes the details
of the numerical simulations, whilst §3 provides half-lives for
individual Centaurs. §4 gives transition probabilities which
allow the main dynamical pathways through this region of
the Solar system to be identified. The simulations are used
to estimate the total population of Centaurs in §5, together
with the typical fluxes inwards from the Edgeworth-Kuiper
belt. Finally, §6 considers possible correlations between the
dynamics and the colours of the Centaurs.
2 INTEGRATIONS
In order to study the bulk statistics of Centaurs, 32 objects
were selected from the list of Centaurs given on the Mi-
nor Planet Center's website⋆. The objects were restricted
to those with an observed arc of at least 30 days and an
aphelion distance of less than 40 au. This ensures that only
Centaurs with moderately well-determined orbits were in-
cluded in our sample. The list of objects is given in Table
2. Over time, as the Centaurs are observed over longer arcs,
the accuracy with which the orbits are known increases, and
the orbits given on the Minor Planet website change accord-
ingly. The orbits used in these integrations therefore repre-
sent the best available information as of June 2002. Table 2
also gives each Centaur's absolute magnitude H, which is
defined as the apparent magnitude the object would have,
if it were placed at both 1 au from the Earth and 1 au from
the Sun and was observed at zero phase angle. This is calcu-
lated in ignorance of any out-gassing that might occur. We
can estimate the maximum and minimum diameter, assum-
ing values of the albedo between 0.15 and 0.02. This gives
a crude reckoning of the size, though photometric work is
required to obtain any more detail. Of the objects studied
⋆ http://cfa-www.harvard.edu/iau/lists/Centaurs.html
c(cid:13) RAS, MNRAS 000, 1–15
Simulations of Centaurs I: Statistics
3
Object
a
e
i
ω
Ω
H
Dmin Dmax
Class
2000 GM137
7.853
0.118
15.9
123.5
89.7
14.3
4.7
1998 SG35
8.420
0.307
15.6
337.5
173.2
11.3
2001 BL41
10.071
0.267
11.5
319.6
280.1
11.7
2001 PT13
10.624
0.197
20.4
86.6
205.3
2000 EC98
10.651
0.471
1999 UG5
12.778
0.415
Chiron
13.601
0.379
1996 AR20
15.197
0.627
4.4
5.6
6.9
6.2
9.0
9.5
163.5
173.2
289.4
87.4
10.1
339.1
209.4
6.5
170
107.9
330.1
14.0
5
Chariklo
15.775
0.171
23.4
241.4
300.4
6.4
180
2001 XZ255
16.039
0.043
2.6
294.2
77.8
11.1
21
2000 QC243
16.560
0.203
20.7
150.0
337.9
7.6
100
1994 TA
16.849
0.301
5.4
154.9
137.7
11.5
2001 SQ73
17.485
0.177
17.4
304.2
16.3
9.6
2000 CO104
17.497
0.256
1999 XX143
17.886
0.458
4.0
6.8
339.2
346.8
10.0
214.9
103.8
Asbolus
17.938
0.619
17.6
290.3
6.1
2002 GO9
19.418
0.277
12.8
92.0
117.4
8.5
9.0
8.5
1998 QM107
20.042
0.136
9.4
154.9
127.2
10.4
Pholus
20.265
0.573
24.7
354.6
119.3
7.0
140
2002 CA249
20.713
0.385
6.4
182.4
313.6
12.0
1999 HD12
21.322
0.583
10.1
288.8
177.7
12.8
2002 DH5
22.433
0.384
22.5
323.7
157.0
10.4
2002 GZ32
23.081
0.216
15.0
154.4
107.2
1995 SN55
23.564
0.663
5.0
49.3
144.6
6.9
6.0
2000 FZ53
23.765
0.479
34.9
290.8
202.4
11.4
Nessus
24.404
0.517
15.7
170.1
31.4
Hylonome
24.909
0.243
4.2
5.5
178.2
2002 GB10
25.139
0.396
13.3
238.9
315.5
2001 KF77
25.992
0.240
4.4
266.4
14.6
1998 TF35
26.429
0.383
12.6
301.8
52.0
2002 FY36
28.969
0.114
2000 QB243
28.953
0.381
5.4
6.5
194.1
332.8
339.4
331.1
9.6
8.0
7.8
9.4
9.3
8.4
8.2
14
9.4
28
140
220
18
41
86
95
45
47
72
79
19
16
54
43
33
17
41
34
68
54
68
28
13
52
43
150
120
90
470
15
490
57
280
47
110
94
190
150
190
78
370
27
26
78
390
590
49
110
240
260
120
130
200
220
S
JS
SU
SU
JU
SU
SU
JN
U
U
U
SU
U
U
SN
SN
UN
UN
SN
UN
SE
UN
UN
SE
UE
SE
UN
UE
UN
UE
N
UE
Table 2. The names of the objects simulated, arranged in order of increasing semi-major axis, together with their orbital elements as of
June 2002. Here, a is the semi-major axis measured in astronomical units (au), e is the eccentricity, i is the inclination (in degrees), ω is
the argument of perihelion (in degrees), Ω is the longitude of the ascending node of the orbit (in degrees) and H is the absolute visual
magnitude. Dmin and Dmax are the values for the diameter (in km) assuming albedos of 0.15 and 0.02 respectively, which are typical
upper and lower limits from albedo measurements of comets and Centaurs performed to date. 'Class' is the classification of the object,
using the scheme given in Horner et al. (2003). [The data are compiled from the Minor Planet Center.]
c(cid:13) RAS, MNRAS 000, 1–15
4
J. Horner, N.W. Evans & M.E. Bailey
in these integrations, the one with the brightest absolute
magnitude is 1995 SN55, with a value of H = 6.0, which
corresponds to a diameter of between 220 and 590 km. The
object with the faintest absolute magnitude is 2000 GM137,
with H = 14.3 giving a diameter of between 5 and 13 km.
This is similar to the size determined for some cometary
nuclei (e.g. Sanzovo et al. 2001; Lamy et al. 2004). Hence,
the Centaurs come in a wide range of sizes, from very large
(1995 SN55 and Chiron) to those comparable in size with
normal comets (2000 GM137).
The orbital elements of each object were used to cre-
ate a swarm of 729 clones, distributed through a small cube
of a-e-i (semimajor axis, eccentricity and inclination) space,
centred on the original orbit. The clones of the objects were
created by incrementally increasing (and decreasing) the
semi-major axis of the object by 0.005 au, the eccentric-
ity by 0.005, and the inclination by 0.01◦. These increments
are sufficiently small that the clones can be considered as
initially essentially indentical to one another, yet they are
large enough to ensure that the subsequent chaotic dynam-
ical evolution following close planetary encounters rapidly
disperse their orbits through phass space. So, nine values
were used for each of these elements, with the central (fifth)
value of the nine having the original orbital elements. This
gives 729 clones of each of the 32 Centaurs, giving a grand
total of 23 328 objects.
The use of multiple clones of an individual object in the
study of its behaviour over time is desirable for a number
of reasons. First, the observations contain some uncertainty,
which means that the orbit itself is not known beyond a
certain degree of precision. This alone would be enough to
promote the use of a cluster of orbits with slightly differ-
ent parameters. In addition, the chaotic nature of the orbits
implies that an infinitesimally small change in the initial pa-
rameters may lead to a major difference in the final outcome
of the simulation. This means that, beyond a certain time
in the future, an object could be anywhere within the Solar
system, or even beyond, as a result of a tiny change in the
initial elements. These two facts taken together suggest that
the best means to examine the future or past behaviour of
an object is to integrate a large number of clones, and to
examine the statistical properties of the dataset (Hahn &
Bailey 1990, Dones et al. 1996). The number of clones used
in such a simulation is chosen to maximise the size of the
dataset available for analysis, without requiring an excessive
amount of time for the simulations to run. The simulations
described here took about three months to run on a desktop
workstation.
The clusters of Centaurs were then integrated for 3 Myr
in both the forward and backward directions. The gravi-
tational influence of the four Jovian planets (Jupiter, Sat-
urn, Uranus and Neptune) was included in the integrations,
which were all carried out using the hybrid integrator within
the Mercury (Chambers 1999) software package. This is a
symplectic integrator, which makes use of a turnover func-
tion to switch to an accurate Bulirsch-Stoer algorithm for
close encounters. The terrestrial planets (Mercury, Venus,
Earth and Mars) were all omitted from the integrations, and
their masses added to that of the Sun. The only slight detri-
ment that this causes is loss of accuracy when objects are
captured into orbits crossing those of the terrestrial planets.
Even in this case, however, the effects of Jupiter (and the
other giant planets, if the object's aphelion lies sufficiently
far from the Sun) are generally much greater than those of
the terrestrial planets.
After some trials, a time step of 120 days was used for
the integrations, since this was found to give a good com-
promise between speed and accuracy. In order to determine
the most efficient timestep, an object was placed on a typ-
ical short period cometary orbit with perihelion near the
Earth and aphelion near Jupiter. A number of clones were
created, and the ensemble was integrated for 105 yrs with
time steps of 30, 60, 120, 240 and 360 days. The result-
ing orbital elements were then compared, and it was found
that the results for timesteps of 30, 60, 120 and 240 days
gave consistent results, while 360 days was too long a time
step. After a number of such trials, a time step of 120 days
was used for the integrations, since this was found to give
a good compromise between speed and accuracy, the mid-
range value of 120 days being chosen so as to err on the side
of accuracy where possible.
An ejection distance of 1000 au from the Sun was
used, following Levison & Duncan (1994). Any object which
reached this distance was removed from the integration †.
Also removed were those objects which impacted upon the
surface of the Sun (q < 0.005 au), or on any one of the giant
planets (the separation is less than the physical radius of
the planet). On completion of the integration, datafiles were
created for each clone which gave the values of the orbital
elements at 100 year intervals for the clone's entire lifetime
within the integration. It was on these files that the analysis
was carried out. Each one of these files was approximately
5 Mbytes in size (the exact size varied, since the file termi-
nated with the ejection of the clone from the simulation).
Hence, the 23 328 clones in total occupied ≈ 120 Gbytes of
disk space for their orbital elemental evolution alone, prior
to any analysis – a daunting dataset by any standards!
3 HALF-LIVES
It is straightforward to calculate the value of the half-life
for each Centaur. As illustrated in Figure 1, the number of
clones remaining within the simulation decays in a roughly
exponential manner as a function of time. The four objects
whose decay is shown are 1996 AR20 (the object with the
shortest lifetime of those studied), Pholus (an object with
a moderately short lifetime), Nessus (a relatively long lived
object) and 2000 FZ53 (the object with the longest lifetime).
It has often been noted in long Solar system integrations
that the number of clones remaining within a simulation
decays exponentially with time (e.g., Dones et al. 1996, Hol-
man 1997, Evans & Tabachnik 1999). The trait is less ob-
vious with the longer lived objects. For example, Dones et
al. (1996) found that the number of surviving clones in their
integrations decayed exponentially at early times, while at
later times (generally greater than twice the half-life) the
decay was slower. This is because those objects surviving
† The long-period comets (the L class in Table 1) are defined
to have aphelion Q in excess of 1000 au. Our choice of ejection
distance means that the statistical properties of the L class cannot
be reliably computed from our simulations.
c(cid:13) RAS, MNRAS 000, 1–15
Simulations of Centaurs I: Statistics
5
600
400
200
0
0
7
6
5
4
3
0
1996 AR20
Pholus
Nessus
2000 FZ53
Time (years)
Time (years)
Figure 1. The number of surviving clones (lower left panel) and its natural logarithm (lower right panel) as a function of time for 1996
AR20, Pholus, Nessus and 2000 FZ53. The upper panels show the same quantities but now for the entire ensemble of 23 328 clones. In
all cases, the graphs are deduced from the simulations in the forward direction.
the longest are the ones transferred to the most stable areas
of the Solar system.
tions with large numbers of clones are the only reasonably
reliable method for half-life estimation.
The number of clones N which remain after time t is
therefore given by
N = N0e−λt,
λ =
0.693
T 1
2
,
(1)
where N0 is the initial number of clones, λ is the decay
is the half-life for the object. To calculate
constant, and T 1
the half-life, the simulation data were analysed with the help
of least-square fitting routines from Press et al. (1992). The
software provides the value of the χ2 function
2
χ2(λ) =
N
X
i=1
h ln Ni − ln N0 + λti
σi
2
i
(2)
where Ni is the number of clones remaining at time ti and
the σi are the individual standard deviations on the data
points. As the σi are unknown, we proceed by first assigning
uniform errors, fitting for the model parameters by minimiz-
ing the χ2 and then rescaling the errors using eqn. [15.1.6]
of Press et al. (1992). Of course, this well-known procedure
precludes an independent estimate of the goodness of fit.
The overall dataset of 23 328 objects has an ensemble
half-life of 2.76 Myr in the forward direction and 2.73 Myr
in the backward direction. This gives us an estimate of the
mean lifetime of a typical Centaur. Note that this lifetime
adds weight to the argument that the population of the Cen-
taur region is in a steady state (a reservoir of objects con-
stantly being drained by Jupiter, and refilled from a long-
lived source like the Edgeworth-Kuiper belt).
2
Dones et al. (1996) calculated the half-lives of Chiron,
Pholus, Nessus and 1994 TA using ∼ 100 clones. Their re-
sults for Chiron and Nessus are in excellent agreement with
ours, but they found T 1
= 2.1 and 2.4 Myr for Pholus and
1994 TA – somewhat larger than our results. The most likely
cause of the discrepancy is in the different algorithms used
to populate the clones. Dones et al. carried out a compari-
son with approximations like Opik's (1976) theory and dif-
fusion equations (e.g., van Woerkom 1948) and concluded
that both methods significantly overestimate the lifetimes
by factors between 2 and 5. It seems that numerical simula-
c(cid:13) RAS, MNRAS 000, 1–15
The half-lives of individual Centaurs are given in Ta-
ble 3. The value of the Poisson uncertainty σ is calculated
as σ = T1/2/√N0, where N0 is the initial number of clones.
This uncertainty is added in quadrature to the uncertainty
in the fitted half-life, as judged from the χ2 surface in the
space of model parameters and as returned by our fitting
software. Note that our ignorance of individual error bars
on our simulation datapoints may lead to an underestimate
of the latter quantity – it is generally smaller than σ by
an order of magnitude – as our algorithm is tantamount to
assuming a good fit to the exponetial decay law. The ob-
ject with the shortest half-life is 1996 AR20, a JN object,
which has a half-life of approximately 540 kyr in the for-
ward integration and 590 kyr in the backward integration.
The object with the longest half-life is 2000 FZ53, a UE
object, with half-lives of 26.8 Myr (forwards) and 32.3 Myr
(backwards).
On comparison with the orbital elements (Table 2), a
correlation can be seen between the position of a Centaur's
orbit within the Solar system and its half-life. The farther
from the Edgeworth-Kuiper Belt, the shorter the half-life.
This is not unexpected – Jupiter is significantly more mas-
sive than Saturn, and Saturn in turn is more massive than ei-
ther Uranus or Neptune. So, the farther from the Edgeworth-
Kuiper Belt, the more massive are the planets with which
the Centaur interacts and the more frequently do such en-
counters occur. This effect is evident when the orbital el-
ements of the objects are plotted against the logarithm of
the half-life, as in Figure 2. There is only a rough correlation
with semi-major axis, but the data indicate a lower bound
to the half-life as a function of perihelion q and an upper
bound as a function of aphelion Q. Specifically, we find that
< T1/2 ∼
(3)
< 0.064 exp(0.275 Q).
0.392 exp(0.135 q) ∼
where q and Q are in au and T1/2 is in Myr. This holds for
all the Centaurs in our sample, but it is conceivably possi-
ble that low eccentricity orbits between the planets are ex-
tremely long-lived (e.g., Holman 1997; Evans & Tabachnik
1999). Additionally, in the plot of perihelion distance versus
half-life, there seem to be three rough groupings of objects.
The first are those in a band along the line from q = 6,
σ
Object
2001 SQ73
2001 SQ73
6
J. Horner, N.W. Evans & M.E. Bailey
Object
C
D
1996 AR20
1996 AR20
2000 EC98
2000 EC98
1998 SG35
1998 SG35
2000 GM137
2000 GM137
1995 SN55
1995 SN55
1999 UG5
1999 UG5
Asbolus
Asbolus
2001 PT13
2001 PT13
2001 BL41
2001 BL41
Chiron
Chiron
JN F
JN B
JU F
JU B
JS
JS
S
S
SE
SE
F
B
F
B
F
B
SU F
SU B
SN F
SN B
SU F
SU B
SU F
SU B
SU F
SU B
1999 XX143
1999 XX143
SN F
SN B
1999 HD12
1999 HD12
Pholus
Pholus
1994 TA
1994 TA
2000 CO104
2000 CO104
2001 XZ255
2001 XZ255
SE
SE
F
B
SN F
SN B
SU F
SU B
U
U
U
U
F
B
F
B
T 1
2
0.54
0.59
0.61
0.63
0.67
0.65
0.72
0.68
0.70
0.80
0.74
0.85
0.86
0.75
0.94
0.87
0.95
0.95
1.03
1.07
1.06
1.38
1.22
1.13
1.28
1.39
1.78
1.52
1.89
2.24
2.94
2.43
0.02
0.02
0.02
0.02
0.03
0.02
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.04
0.03
0.04
0.04
0.04
0.04
0.04
0.05
0.05
0.04
0.05
0.05
0.07
0.06
0.07
0.08
0.11
0.09
C
U
U
D
F
B
2002 GO9
2002 GO9
UN F
UN B
2000 QC243
2000 QC243
U
U
F
B
2002 CA249
2002 CA249
UN F
UN B
1998 QM107 UN F
1998 QM107 UN B
Nessus
Nessus
Hylonome
Hylonome
2001 KF77
2001 KF77
2002 DH5
2002 DH5
2002 GZ32
2002 GZ32
Chariklo
Chariklo
1998 TF35
1998 TF35
2002 GB10
2002 GB10
2002 FY36
2002 FY36
2000 QB243
2000 QB243
2000 FZ53
2000 FZ53
SE
SE
F
B
UN F
UN B
UN F
UN B
UN F
UN B
UN F
UN B
U
U
UE
UE
UE
UE
N
N
UE
UE
UE
UE
F
B
F
B
F
B
F
B
F
B
F
B
T 1
2
2.86
2.73
2.93
3.67
3.18
3.44
4.06
2.54
4.87
5.65
4.91
6.40
6.37
7.30
8.89
10.1
9.08
12.8
11.3
7.78
10.3
9.38
11.5
10.8
11.1
13.1
12.5
13.5
13.0
17.8
26.8
32.3
σ
0.11
0.10
0.11
0.14
0.12
0.13
0.15
0.09
0.18
0.21
0.18
0.24
0.24
0.27
0.33
0.4
0.34
0.5
0.4
0.28
0.4
0.35
0.4
0.4
0.4
0.5
0.5
0.5
0.5
0.7
1.0
1.2
Table 3. Half-Lives T 1
(in Myr) of the simulated objects, together with an error estimate σ (in Myr). The column labelled C gives the
2
class of the Centaur, while the column labelled D gives the direction of integration (F for forward, B for backward).
2
2
2
= 5.75 to q = 26, log T 1
log T 1
= 7.2, which accounts for
the bulk of the objects. A second group comprises six ob-
jects, which lie roughly on a parallel track at values of log T 1
greater by 0.7. Finally, 2000 FZ53 sits alone far above either
of these groups. The objects which are in the first group cur-
rently lie away from the positions of any major resonances
and so tend to have short lifetimes. The objects in the sec-
ond group tend to lie nearer to stable mean motion or sec-
ular resonances. In fact, the orbit of 2002 GB10 lies within
0.009 au of the 3:4 resonance of Uranus, while the orbit of
Chariklo lies within 0.09 au of the 4:3 resonance of Uranus.
Clones of 2000 FZ53 quite frequently display resonant be-
haviour during the course of the simulations, although it
does not currently lie near any major mean motion reso-
nances. A possible cause of 2000 FZ53's exceptionally long
half-life is its abnormally large inclination – higher than that
of any other Centaur studied by over 10 degrees. Any cor-
relations of half-life with inclination and ecccentricity are
less clear-cut than those with position. However, there is a
lack of long-lived objects at large e. This is a consequence
of the fact that Centaurs with large e must cross the orbits
of several of the outer planets, and so inevitably are more
unstable than bodies whose close approaches are restricted
to just one or two planets.
As the equations of motion are time-reversible, it might
naively be expected that the forward and backwards half-
lives should be the same. In fact, it is often found that inward
evolution of the orbits of minor bodies is more likely than
outward evolution (e.g., Oikawa & Everhart 1979). This is
c(cid:13) RAS, MNRAS 000, 1–15
Simulations of Centaurs I: Statistics
7
8
7.5
7
6.5
6
5.5
0
8
7.5
7
6.5
6
5.5
0
8
7.5
7
6.5
6
10
20
Semi-major axis (au)
30
5.5
0
10
30
Aphelion distance (au)
20
40
8
7.5
7
6.5
6
5.5
0
20
10
Inclination (deg)
30
40
10
20
30
Perihelion distance (au)
Figure 2. The relationship of the logarithm of the half-lives of the Centaurs with their semi-major axis, perihelion and aphelion distances,
and their inclination. Points in red show the results of the forward direction, and those in green show the results of the backward direction
integrations.
because the Edgeworth-Kuiper belt provides a source while
Jupiter provides a sink, so that clones ejected by Jupiter
do not have an opportunity to return. Of all the objects,
19 have shorter half-lives in the forward direction, and 13
have shorter half-lives in the backward direction. The least
discrepancy occurs for 2001 BL41, with the two half-lives
agreeing to within 0.25%. The greatest discrepancy occurs
for 2002 GZ32, where the backward half-life is 46% longer
than the forward half-life. These differences can be visualised
as the effect of the first encounters the clones have with the
major planets. If an encounter with one of the planets occurs
early enough in the simulation, the entire ensemble of clones
will be perturbed, moving the objects onto slightly differ-
ent orbits, with slightly different half-lives. For the entire
dataset, the forward and backward half-lives only diverge
by a matter of 30000 years out of 2.76 Myr – a discrepancy
of only just over 1%.
From the dataset, it is also possible to calculate the
half-lives of the starting class of the objects. The results of
this calculation are given in Table 4. The number of clones
in any particular class is not necessarily an exact multi-
ple of 729. This is because a number of the objects have
outlying clones which actually fall into a different class at
the start of the integration as compared to the seed. The
results of this analysis again show the dependence of half-
life on perihelion position – objects in the Jupiter classes
have half-lives shorter than those in the Saturn classes, and
these in turn are more short-lived than the objects in the
c(cid:13) RAS, MNRAS 000, 1–15
Uranus classes. There are also hints in the table that more
eccentric objects under the control of any particular planet
may be more long-lived than their less eccentric counter-
parts (compare, for example, the half-lives of U, UN and
UE objects). Orbital periods of the more eccentric objects
are greater than those on near-circular orbits with similar
perihelion distances, and hence encounters with the giant
planets happen less frequently.
Over the course of the integrations, the clones of each
object are repeatedly transferred between classification bins.
This allows us to evaluate the amount of time that is spent
in each of these classes over the simulation, together with
the number of times the object is transferred into that class.
From this, we can calculate the mean time that an object
spends in any particular class before being transferred into
another. The results of such calculations are presented in
Table 5. In this table, the value of the mean lifetime for
the L or long-period comet class has been ignored, since
objects which enter this classification are then removed from
the simulation as they pass 1000 au from the Sun. This
means that the value of mean lifetime for objects of class L is
unrealistically small. It is also noteworthy that the EK and T
classes have particularly short mean lifetimes. A stable orbit
in these regions requires decoupling from Neptune, and there
are no non-gravitational forces within the integrations which
could allow this to happen. Hence, the very small number
of objects which attain these two classes only do so at the
8
J. Horner, N.W. Evans & M.E. Bailey
C
Direction
N
JS
JU
JN
S
SU
SN
SE
U
UN
UE
N
Forward
Backward
Forward
Backward
Forward
Backward
Forward
Backward
Forward
Backward
Forward
Backward
Forward
Backward
Forward
Backward
Forward
Backward
Forward
Backward
Forward
Backward
729
729
729
729
891
891
729
729
3519
3519
2205
2205
1917
1917
3780
3780
5094
5094
3006
3006
729
729
T 1
2
0.67
0.65
0.61
0.64
0.59
0.63
0.72
0.68
1.03
1.02
1.11
1.22
1.75
1.77
3.72
3.36
5.84
6.19
12.5
14.2
12.5
13.5
σ
0.03
0.02
0.02
0.02
0.02
0.02
0.03
0.03
0.02
0.02
0.02
0.03
0.04
0.04
0.06
0.06
0.08
0.09
0.23
0.26
0.46
0.50
Table 4. The half-lives of the individual classification bins across
which the objects fall at the start of the integrations. N is the
number of clones in that particular class at the start of the inte-
is the half-life (in Myr), and σ is again the uncer-
grations, T 1
tainty on the half-life (in Myr).
2
extremes of a series of perturbations and are immediately
perturbed back into classes under the control of Neptune.
4 TRANSFER PROBABILITIES
It is also straightforward to calculate the probability of an
object being transferred from one class to another. This can
be visualised by constructing a 24 × 24 grid with the initial
class on the vertical axis and the final class on the horizon-
tal (these classes are just initial and final with respect to a
single transfer, not for the entire integration). This is done
by recording every transfer which occurs within the integra-
tions, and hence calculating the fraction of objects which, for
example, are transferred from class J to class JS. The results
are shown in Table 6. The numbers have been normalised
so that the sum along any row is unity. For any class, the
probabilities give the relative likelihood of leaving from that
class to the target classes given on the horizontal axis. As
an example, let us take a typical result from one of these
tables, namely that the value of the probability of transfer
from class J to class JS is ∼ 0.49 (Table 6). This means that,
for an object in the J class, there is a ∼ 49% chance of the
object being transferred directly into the JS class the next
time the classification changes. We can see that for such an
Class
Forward lifetime Backward lifetime
EC
MC
E
SP
I
J
JS
JU
JN
JE
JT
S
SU
SN
SE
ST
U
UN
UE
UT
N
NE
NT
EK
T
1960
950
4020
1680
570
290
890
1320
1110
1470
830
1400
3670
2710
3060
2960
6650
4860
4710
1560
2600
4640
4100
330
100
1990
890
3480
1630
630
300
890
1310
1120
1470
860
1350
3630
2710
3010
3090
6150
5040
4870
1480
2500
4350
4440
320
100
Table 5. The mean time (in yrs) spent in each classification
bin, before a Centaur clone is transferred to another bin. The
shortness of this mean time is understandable, as objects close
to the boundary are often tranferred to and fro. Note that EC
stands for Earth-crossing objects and MC for Mars-crossing.
object, the two most likely transfers are to the JS class or
to a short-period (SP) comet, and between them, these two
possibilities make up the great bulk of transfers for all J
objects.
Table 6 shows a number of interesting features. When-
ever an object is controlled by two planets (one at perihelion
and one at aphelion), the classes to which the object is most
likely to move correspond to transitions at the perihelion
and aphelion of the planet. For instance, an SU object is con-
trolled at perihelion by Saturn and at aphelion by Uranus. It
is most likely to be transferred to one of the classes JU or U
by an encounter at aphelion. These cases corresponds to an
encounter with Uranus either increasing the eccentricity of
the orbit, and hence pushing the perihelion down to Jupiter's
c(cid:13) RAS, MNRAS 000, 1–15
control, or decreasing the eccentricity, pulling the perihelion
away from Saturn's control. For encounters at perihelion,
the most likely classes are S or SN corresponding either to a
circularisation of the orbit at Saturn, or to a pumping of the
eccentricity of the orbit, as the aphelion moves from Uranus'
to Neptune's control. These most popular transfers can be
traced diagonally down the tables, around the empty diag-
onal corresponding to the same initial as final class. These
four parallel lines of high probabilities give the appearance
of two sets of "tram lines" running through the tables. Af-
ter these possibilities, other transfers are also viable, albeit
with lower probabilities - for example, an SU object can suf-
fer a perihelion-aphelion interchange at Saturn, moving to
the JS class. However, the fact that the four classes most
likely to be reached in a transfer lie along the "tram lines"
vindicates the classification scheme, which is based on the
idea of transfers by interaction primarily at perihelion and
aphelion.
Also of interest is an effect which can be seen on com-
paring Table 6 with the equivalent results for each individual
Centaur (given in Appendix A of Horner 2003). From any
class, the probability of transfer to another class is roughly
constant, regardless of the direction of integration or the
object in question. The main discrepancies lie in very low
probability transfers, where the uncertainty is large because
of the small numbers involved. This means that for a newly
discovered object, it is possible to give the probabilities of
its transfer to any new class, as long as the initial class can
be computed. It also permits insight into the main dynami-
cal pathways followed by a Centaur. For example, using the
values in Table 6 and assuming an initial population of 1000
short-period comets or SP objects, it can be seen that 27 of
these objects become E types, all of which would return to
the SP bin, 41 enter the I class, three of which would on av-
erage then return, 304 travel to the J class, 147 making the
return trip. For the other bins, we find 84 returning objects
from JS, 8 from the JU class and one from JN. Therefore,
270 of 1000 objects that leave the short-period class return
immediately the next time that their classification changes.
Given that the total time spent in each class and the
number of times that class becomes occupied are calculable
from the simulation data, it is also possible to compute the
probability per unit time of a transfer. We already know
the probabilities that an object in a particular class will be
transferred to any other. Dividing the mean time spent in
any class by this probability, we obtain the mean transfer
time from one class to any target class. The inverse of this
is the probability per unit time of the transfer. The values
obtained in this type of analysis are given in Horner (2003)
for the individual Centaurs, while the probabilities for the
entire dataset are given in Table 7. The results are given as
probabilities per Myrs, so that a value of 0.1 in a particular
box means that an object making the relevant transfer would
have a mean transfer time of about 10 Myrs. This means that
the population in any class N evolves according to
dN
dt
= X
i
PiNi − X
j
PjNj
(4)
where Pj and Nj are the probabilities per unit time and the
populations in the bins along the row, while the Pi and Ni
are the corresponding quantities along the column. In other
words, the ingress to a particular class is governed by the
c(cid:13) RAS, MNRAS 000, 1–15
Simulations of Centaurs I: Statistics
9
numbers in the column, and the egress by the row. Math-
ematically speaking, this gives us coupled sets of exactly
solvable linear first-order differential equations that govern
the evolution of Centaur clones. We will return to this in a
later publication, but these ideas are already prefigured in
Bailey et al. (1992).
5 CENTAUR FLUXES AND POPULATION
During the simulations, we also record the numbers of clones
that become Earth-crossing objects, Mars-crossing objects
and short-period comets. This gives us a means to reckon the
total population of Centaurs. Fernandez (1985) suggested a
flux of 10−2 new short-period comets per year, with a mean
lifetime of ∼ 6 kyr. More recently, Levison & Duncan (1994)
find that the mean fade time for a short-period comet is ∼ 12
kyr. Under the assumption that the current population of
short-period comets is in steady state, the work of Levison
& Duncan implies a flux of ∼ 0.5 × 10−2 new short-period
comets per year. This is equivalent to one new short-period
comet being captured, on average, every 200 years. If we
assume that the entirety of this flux comes from the Centaur
region, then this allows us to estimate the total population
of the Centaur region using the simulations.
From the simulation data, a total of 7 900 out of 23 328
clones (∼ 34% of the initial population) become short-period
comets at some point during the forward integrations, and
8 068 (again ∼ 34%) become short-period comets during the
backward integrations. This is a flux of one new short-period
comet every 380 years. If we assume all short-period comets
are captured from the Centaur region, an estimate of the
> 4
total number of Centaurs (with perihelion distance q ∼
< 60 au) is ∼ 44 300. This rep-
au and aphelion distance Q ∼
resents the population of objects bright enough to be vis-
ible as short-period comets, were they to be captured into
such an orbit. An effective nuclear diameter d greater than
1 km seems a reasonable limit to place for this value, though
there are an increasing number of comets with d ∼ 0.5 km
(e.g., Lamy et al. 2004). This calculation also only takes
into account objects becoming short-period comets for the
first time. Given that the mean fade time is ∼ 12 kyr, it
is reasonable to assume that objects captured for the first
time are significantly brighter as short-period comets than
those which have experienced a number of prolonged stays
in the region. Therefore, the objects, which in the simula-
tions display a number of prolonged periods as short-period
comets, would actually exhaust all their volatiles early on
and should not contribute to the new short-period comet
flux in later passages through the inner Solar system. Note
that the usage of the flux of short-period comets to normalise
the source populations has also been exploited recently by
Emel'yanenko, Asher & Bailey (2004) to estimate the to-
tal population of trans-Neptunian objects on highly eccen-
tric orbits, which reside at still greater heliocentric distances
than the Centaurs.
Given our estimate of the total population of Centaurs
and the knowledge that the half-life is ∼ 2.75 Myr, we can
estimate the influx of new Centaurs from the Edgeworth-
Kuiper belt. Neglecting those few objects that could be
captured onto Centaur-like orbits from high eccentricity or-
bits from the Oort cloud, we can see that ∼ 22150 objects
10
J. Horner, N.W. Evans & M.E. Bailey
Pre
Post
E
SP
I
L
J
JS
JU
JN
JE
JT
S
SU
SN
SE
ST
U
UN
UE
UT
N
NE
NT
E
.
.
.
.
.
.
.
SP
.
.
.
.
.
.
.
.
.
.
I
.
.
.
.
.
L
.
.
.
.
.
.
.
.
.
.
.
.
J
.
.
.
.
.
JS
.
.
.
.
.
.
.
JU
.
.
.
.
.
.
.
.
.
JN
.
.
.
.
.
.
.
.
JE
.
.
.
.
.
.
.
.
.
JT
.
.
.
.
.
.
.
S
.
.
.
.
.
SU
.
.
.
.
.
.
.
.
SN
.
.
.
.
.
.
.
.
SE
.
.
.
.
.
.
.
.
.
ST
.
.
.
.
.
.
U
.
.
.
UN
.
.
.
.
.
.
.
UE
.
.
.
.
.
.
.
UT
.
.
.
.
.
N
.
.
.
NE
.
.
.
.
.
NT
.
.
.
Pre
Post
E
SP
I
L
J
JS
JU
JN
JE
JT
S
SU
SN
SE
ST
U
UN
UE
UT
N
NE
NT
E
.
.
.
.
.
.
.
SP
.
.
.
.
.
.
.
.
.
.
I
.
.
.
.
.
.
L
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
J
.
.
.
.
JS
.
.
.
.
.
.
.
.
JU
.
.
.
.
.
.
.
.
.
JN
.
.
.
.
.
.
.
.
JE
.
.
.
.
.
.
.
.
.
JT
.
.
.
.
.
.
.
S
.
.
.
.
.
SU
.
.
.
.
.
.
.
.
SN
.
.
.
.
.
.
.
.
SE
.
.
.
.
.
.
.
.
.
ST
.
.
.
.
.
.
U
.
.
.
UN
.
.
.
.
.
.
.
UE
.
.
.
.
.
.
.
UT
.
.
.
.
.
N
.
.
.
NE
.
.
.
.
.
NT
.
.
.
Table 6. Transfer probabilities for the entire simulation in the forward (upper table) and backward (lower table) direction. Notice that
the probabilities are nearly the same regardless of the direction of integration. An empty entry means that the transfer probability is
< 10−3. The leading diagonal is empty by definition.
must be replaced every 2.75 Myr. This is equivalent to 1 ob-
ject transferred to the Centaur region from the Edgeworth-
Kuiper belt every ∼ 125 yrs. This calculation also ignores
the small flux of objects from Centaur-type orbits to the
Edgeworth-Kuiper belt. Our simulations allow neither for
non-gravitational effects, such as collisions, nor for the grav-
itational perturbations between Edgeworth-Kuiper belt ob-
jects. So, it is impossible to determine the flow of objects
from orbits which encounter Neptune to those which are sta-
ble beyond. To drive the inward flux of Centaurs, the effects
c(cid:13) RAS, MNRAS 000, 1–15
Simulations of Centaurs I: Statistics
11
Pre
Post
E
SP
I
L
J
JS
JU
JN
JE
JT
S
SU
SN
SE
ST
U
UN
UE
UT
N
NE
NT
E
.
.
.
.
.
.
.
.
SP
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
I
.
.
.
.
.
.
.
.
.
.
.
.
.
.
L
.
.
.
.
.
.
.
.
.
.
.
.
.
J
.
.
.
.
.
.
.
.
.
.
.
.
.
JS
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
JU
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
JN
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
JE
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
JT
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
S
.
.
.
.
.
.
.
.
.
.
.
.
SU
.
.
.
.
.
.
.
.
.
.
.
SN
.
.
.
.
.
.
.
.
.
.
.
.
SE
.
.
.
.
.
.
.
.
.
.
.
.
.
.
ST
.
.
.
.
.
.
.
.
.
.
.
.
.
.
U
.
.
.
.
.
UN
.
.
.
.
.
.
.
.
UE
.
.
.
.
.
.
.
.
.
.
.
UT
.
.
.
.
.
.
.
.
.
.
.
.
N
.
.
.
.
.
.
.
.
NE
.
.
.
.
.
.
.
.
.
NT
.
.
.
.
.
.
.
.
.
Pre
Post
E
SP
I
L
J
JS
JU
JN
JE
JT
S
SU
SN
SE
ST
U
UN
UE
UT
N
NE
NT
E
.
.
.
.
.
.
.
.
.
.
.
SP
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
I
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
L
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
J
.
.
.
.
.
.
.
.
.
.
.
.
JS
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
JU
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
JN
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
JE
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
JT
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
S
.
.
.
.
.
.
.
.
.
.
.
.
.
SU
.
.
.
.
.
.
.
.
.
.
.
.
.
SN
.
.
.
.
.
.
.
.
.
.
.
.
SE
.
.
.
.
.
.
.
.
.
.
.
.
.
.
ST
.
.
.
.
.
.
.
.
.
.
.
.
.
.
U
.
.
.
.
.
UN
.
.
.
.
.
.
.
.
.
UE
.
.
.
.
.
.
.
.
.
.
.
.
UT
.
.
.
.
.
.
.
.
.
.
.
N
.
.
.
.
.
.
.
NE
.
.
.
.
.
.
.
.
.
.
.
NT
.
.
.
.
.
.
.
.
.
.
.
.
.
Table 7. Transfer probabilities per Myr for the entire simulation in the forward (upper table) and backward (lower table) direction. An
empty entry means that the transfer probability per Myr is < 10−2. The leading diagonal is empty by definition.
of collisions and of perturbations between the Edgeworth-
Kuiper belt objects must be considered. Durda & Stern
(2000) suggest that collisions of objects greater than 4 m
in diameter onto comet-sized bodies within the Edgeworth-
Kuiper belt occur every few days. In addition, they reckon
that the time scale for the disruption of 1 km objects is ∼ 1
Gyr. These two facts in concert imply a high rate of collision
within the Edgeworth-Kuiper belt, sufficiently high that an
inward flux of 1 new Centaur every ∼ 125 yrs seems reason-
able.
c(cid:13) RAS, MNRAS 000, 1–15
12
J. Horner, N.W. Evans & M.E. Bailey
From the forward integrations, we can deduce that 1799
clones (7.7% of the sample) became Earth-crossing and 3799
clones (16%) became Mars-crossing. Very similar numbers
are yielded by the backward integrations. Therefore, we ex-
pect typically one Centaur to become Earth-crossing for
the first time approximately every ∼ 880 yrs, and one new
Mars-crosser every ∼ 420 yrs. Most of the known popula-
tion of Near-Earth objects (NEOs) is asteroidal in nature,
as the Main Belt provides the great majority of NEOs. Mor-
bidelli et al. (2002) state that only 6% of the NEOs are ulti-
mately of Edgeworth-Kuiper Belt origin. NEOs originating
in the Main Belt are expected to survive for far longer times
within the inner Solar system than the cometary bodies,
due to the fact that they lie on orbits significantly decou-
pled from direct perturbations by Jupiter. An approximate
flux of one new Earth-crosser of Centaur (and hence, orig-
inally Edgeworth-Kuiper belt) origin per millennium seems
to be in reasonable agreement with such work.
Though the number of NEOs of Edgeworth-Kuiper belt
origin is small, they are exceptionally important in judging
potential hazards. The size distribution of NEOs of aster-
oidal origin is heavily weighted towards small particles, con-
sistent with the idea that they are collision fragments. The
largest NEO is (1036) Ganymed with an absolute magnitude
H = 9.45 and a diameter d ≈ 50 km. Very few NEOs are
larger than 10 km across. However, for the Centaurs, the
upper end of the size distribution is well-populated, with 16
objects of the 32 listed in Table 2 intrinsically brighter than
Ganymed. The passage of a large Centaur like Chiron or
Pholus into the inner Solar system would provide a very sig-
nificant environmental disturbance (Hahn & Bailey 1990),
as its fragmentation and possible decay could overwhelm
the local space environment with debris and dust.
Another interesting set of statistics is the total number
of times a class becomes occupied during the integrations.
In the forward direction, the Earth-crossers become occu-
pied ∼ 23 300 times over three million years. However, we
already know that only 1799 of the clones enter this class,
so it is obvious that the bulk of objects which become Earth
crossing at least once will in fact enter, leave and re-enter
the class a number of times (in this case, ∼ 13 times). Ta-
ble 8 summarises the number of visits to the short-period
(SP), Earth-crossing (EC), Mars-crossing (MC) and Encke-
type (E) classes that a typical object entering these classes
would have.
The flux of objects into the Encke class of comets can
also be calculated. This is subject to greater errors – be-
cause of the much lower numbers of objects that become
Encke-type comets than become Earth-crossers and because
a time step of 120 days is insufficient to track the orbits
of such objects with reasonable accuracy and because the
gravitational influence of the neglected terrestrial planets is
now significant. For objects in the forward integrations, 303
clones became Encke types, implying a capture rate of one
every ∼ 5200 yrs. The backward integration yields the same
capture rate.
Typically, an object will become an Earth-crosser or
Mars-crosser while classified as a short-period comet, having
been transferred to the inner Solar system by an encounter
with Jupiter. Although some objects can be perturbed onto
Earth and Mars crossing orbits by the other outer planets
(for example, the comets P/1997 T3 Lagerkvist-Carsenty
Class D
N
EC
MC
SP
E
F
B
F
B
F
B
F
B
12.9
13.7
13.7
14.6
28.0
28.4
16.5
18.3
Table 8. Table showing the mean number of times an object
which enters a cometary class at least once will go on to enter and
re-enter that class through its lifetime. D gives the direction of the
integration (F = Forward, B = Backward) while N gives the mean
number of visits to that class. Here, again EC is Earth-crossing,
MC is Mars-crossing, SP is short-period and E is Encke-type.
and P/1998 U3 Jager, discussed in Horner et al. 2003), the
vast majority of such objects are under the control of Jupiter
prior to a period of cometary behaviour. This is evident in
Table 6. In both the forward and backward integrations, the
only classes from which the likelihood of transfer to short-
period orbits is greater than 0.001 are the other cometary
classes (E, I, L), the classes in which the object is controlled
by Jupiter (J, JS, etc.), and the S class.
Finally, we can use the simulation data to estimate the
impact rate on the giant planets from Centaurs. In the for-
ward integration, we find that 144 objects hit Jupiter, 53
hit Saturn, 5 hit Uranus and a further 5 hit Neptune. In
the backward integration, these numbers are 135, 48, 5 and
1 respectively. Given that the estimated population of the
Centaur region is ∼ 44300, then we expect one impact per
10 kyr on Jupiter, one per 28 kyr on Saturn, and one per 300
kyr on Uranus and Neptune. These numbers are likely to be
underestimates of the impact rate on the planets, given that
the errors in integration are at their largest when the clone
is closest to a massive body.
Of course, all the numbers derived in this Section are de-
pendent on the assumed flux of one new short-period comet
every 200 years. If the true flux is higher or lower, then
the total populations of Centaurs, Earth-crossers and Mars-
crossers would need to be correspondingly adjusted upwards
or downwards. Although the flux of new short-period comets
is perhaps uncertain by a factor of two, it is not uncertain by
a factor of ten, and so our population estimates are surely
correct to an order of magnitude.
6 DYNAMICS AND PHOTOMETRIC
COLOURS
Since the Centaurs range over a large area of the outer Solar
system, it is possible that there could be variations in their
observable characteristics (such as colour and light-curves)
as a function of their position. Observations have been car-
ried out by a number of groups (e.g. Hainaut & Delsanti
2002; Peixinho et al. 2001; Weintraub et al. 1997; Bauer
et al. 2003). Although the number of Centaurs with well-
determined colours is restricted, the situation is rapidly im-
c(cid:13) RAS, MNRAS 000, 1–15
Simulations of Centaurs I: Statistics
13
Object
Class D
2000 EC98
JU
1998 SG35
JS
1999 UG5
SU
Asbolus
SN
2001 PT13
SU
2001 BL41
SU
Chiron
SU
1999 XX143
SN
Pholus
SN
1994 TA
SU
2002 GO9
UN
2000 QC243
U
1998 QM107
UN
Nessus
SE
Hylonome
UN
2002 GB10
UE
Chariklo
U
1998 TF35
UE
2002 GB10
UE
F
B
F
B
F
B
F
B
F
B
F
B
F
B
F
B
F
B
F
B
F
B
F
B
F
B
F
B
F
B
F
B
F
B
F
B
F
B
T 1
2
0.61
0.63
0.67
0.65
0.74
0.85
0.86
0.75
0.94
0.87
0.95
0.95
1.03
1.07
1.06
1.38
1.28
1.39
1.78
1.52
2.93
3.67
3.18
3.44
4.87
5.65
4.91
6.40
6.37
7.30
11.1
13.1
10.3
9.38
11.5
10.8
11.1
13.1
B - V
Colour
V - R
R - I
0.854 ± 0.081
0.47 ± 0.06
0.466 ± 0.05
0.50 ± 0.06
0.439 ± 0.076
0.725 ± 0.089
0.456 ± 0.050
0.546 ± 0.063
0.42 ± 0.08
0.46 ± 0.04
0.88 ± 0.18
0.60 ± 0.08
0.72 ± 0.13
0.964 ± 0.085
0.607 ± 0.060
0.625 ± 0.042
0.68 ± 0.02
0.58 ± 0.02
0.75 ± 0.04
0.47 ± 0.04
0.750 ± 0.040
0.513 ± 0.068
0.523 ± 0.045
0.53 ± 0.02
0.47 ± 0.02
0.44 ± 0.03
0.48 ± 0.02
0.53 ± 0.05
0.56 ± 0.04
0.67 ± 0.06
0.34 ± 0.03
0.679 ± 0.039
0.359 ± 0.027
0.356 ± 0.037
0.36 ± 0.10
0.32 ± 0.10
0.67 ± 0.07
0.70 ± 0.06
1.35
0.71
1.19 ± 0.10
0.78 ± 0.04
0.81
1.299 ± 0.099
0.794 ± 0.032
0.814 ± 0.056
1.261 ± 0.139
0.672 ± 0.080
0.740 ± 0.210
0.74 ± 0.06
0.66 ± 0.05
0.38 ± 0.06
0.41 ± 0.06
0.724 ± 0.062
0.448 ± 0.044
0.397 ± 0.069
0.771 ± 0.100
0.730 ± 0.060
0.474 ± 0.095
0.520 ± 0.030
0.368 ± 0.102
-
0.63 ± 0.12
0.64 ± 0.10
0.88 ± 0.07
0.77 ± 0.05
1.090 ± 0.040
0.793 ± 0.041
0.695 ± 0.066
0.74 ± 0.08
0.64 ± 0.07
0.41 ± 0.10
0.46 ± 0.18
0.643 ± 0.082
0.464 ± 0.059
0.490 ± 0.122
0.50 ± 0.07
0.52 ± 0.06
0.58 ± 0.07
0.62 ± 0.07
0.47
0.55
0.802 ± 0.049
0.479 ± 0.029
0.542 ± 0.030
0.49 ± 0.02
0.51 ± 0.02
0.65 ± 0.08
0.66 ± 0.07
1.085 ± 0.111
0.697 ± 0.064
0.651 ± 0.119
0.71 ± 0.02
0.65 ± 0.02
Table 9. List of the Centaurs with colour information. D is the direction of integration and T 1
is the half-life in Myr. The colours are
2
taken from Hainaut & Delsanti (2002), Boehnhardt et al. (2002) and Bauer et al. (2003).
c(cid:13) RAS, MNRAS 000, 1–15
14
J. Horner, N.W. Evans & M.E. Bailey
proving. Colours for 19 Centaurs are given in Table 9. They
range from some of the reddest objects in the Solar system
(e.g., Pholus) to much bluer objects (e.g., Chiron). There
appears to be no correlation between colour and half-life
within the data set, or indeed between colour and classifi-
cation (for example the reddest and the bluest objects on
the list, Pholus and Chiron, are both controlled by the same
planet at perihelion).
We expect objects closest to the Sun to have undergone
some outgassing, and this would lead to a re-surfacing of the
object, covering the older, darker and redder material with
fresh material from the interior of the object. Conversely,
those objects which have not displayed cometary behaviour
since the formation of the Solar system are expected to be
darker and redder. Other factors that could alter the colours
of the Centaurs include impacts between objects. Collisions
might activate areas of the surface of the objects, exposing
fresh material from the interior, manifesting itself as bluer
colours. It follows that photometric observations of the Cen-
taurs do have the potential to provide important clues as to
the dynamical history of the object which cannot be deter-
mined by integration alone.
A good example of this is the case of Chiron and Pholus.
These two objects have similar half-lives, and are controlled
by the same planet. We can easily calculate that ∼ 60% of
the clones of Chiron become short-period comets in the for-
ward integrations, against ∼ 30% of the clones of Pholus.
Yet, it is impossible to determine whether either object has
been a short-period comet in the past, purely by means of
the integrations. All that we can calculate are percentage
probabilities. However, it is possible that the fact that Chi-
ron is the bluest of the Centaurs listed in Table 9 is caused by
its well-known cometary activity, which may indicate that
at some time in the past Chiron was a short-period comet.
Pholus' red colour may similarly indicate it has not yet been
transferred into an orbit which would activate the surface
sufficiently for resurfacing to take place. Hence, we may be
observing the surface of an object which has not been active
since the birth of the Solar system.
Bauer et al. (2003) find that the the Centaurs dis-
play a wider colour distribution than both Jupiter-family
short-period comets and Edgeworth-Kuiper belt objects.
The members of the former population are all active, while
the members of the latter are not. An intermediate popu-
lation such as the Centaurs should at least show properties
varying between one extreme and the other, and this seems
borne out by the data in Table 9.
7 CONCLUSIONS
Detailed simulations of the evolution of the Centaur popu-
lation under the gravitational influence of the Sun and the
4 giant planets (Jupiter, Saturn, Uranus and Neptune) have
been carried out. 23 328 test particles were created by taking
the orbits of 32 well-known Centaurs and producing 9 clones
in each of semi-major axis (a), eccentricity (e) and inclina-
tion (i), giving a total of 729 clones of each object. The clones
were then integrated in both the forward and backward di-
rections for a period of 3 Myr. Any clone which reached a
heliocentric distance of 1000 au was deemed ejected from
the Solar system and removed from the integration.
The half-lives for ejection of the Centaurs were calcu-
lated from the simulation data. These ranged from ∼ 540
kyr for 1996 AR20 (in the forward direction) to ∼ 32 Myr
2000 FZ53 (in the backward direction). The half-life of the
ensemble of Centaurs was ∼ 2.7 Myr, irrespective of the di-
rection of integration. To analyze the simulation data, we
exploited a classification scheme introduced by Horner et
al. (2003), which breaks down the orbits of objects in the
Solar system into different classes according to the planet's
controlling the perihelion and aphelion. The use of the new
classification scheme allowed us to determine the dynamical
pathways through which the Centaur population evolves.
Transfer probabilities were calculated between the different
classes. These were found to be remarkably independent of
the object under study, offering the prospect of determining
possible future histories for objects from the knowledge of
their current classification.
Both observations and simulations suggest that one new
short-period comet is produced every 200 yrs. This is used
to normalise our simulation results. In this case, the total
< 60
> 4 au to Q ∼
population of the Centaur region (from q ∼
> 1
au) with nuclei large enough to provide visible comets (d ∼
km) is estimated to be ∼ 44 300. A flux of 1 new object
into the Centaur region from the Edgeworth-Kuiper belt
every 125 yrs is required to maintain the population in a
steady-state. Additionally, one fresh Earth-crossing object
is expected to arise from the Centaur region every ∼ 880
yrs. This is both of interest and concern, as large Centaurs
entering the inner Solar system are likely to fragment with
production of much dangerous dust and debris.
In a companion paper (Horner et al. 2004), individual
Centaur clones are discussed, showing examples of stable
resonant behaviour, Kozai instabilities, capture into satellite
orbits, and evolution into objects spending long time periods
in the inner Solar system, amongst other things. This gives
a feel for the rich variety of behaviour of the diverse objects
known as "Centaurs".
ACKNOWLEGMENTS
JH acknowledges financial support from the Particle Physics
and Astronomy Research Council. We thank the referee S.C.
Lowry for a careful reading of the manuscript.
REFERENCES
Asher D.J., Steel D.I. 1993, MNRAS, 263, 179
Bailey M. E., Chambers J. E., Hahn G., Scotti J., Tancredi G.
1992, Proceedings of 30th Liege International Astrophysical
Colloquium, eds Brahic A., Gerard J.C., Surdej J., Universit´e
de Li`ege, Li`ege, p. 285
Bauer J. M., Meech K. J., Fern´andez Y. R, Pittichova J., Hainaut
O. R., Boenhnhard H. 2003, Icarus, 166, 195
Boehnhardt H. et al. 2002, A&A, 395, 297
Chambers J. E. 1999, MNRAS, 304, 793
Clube S. V. M., Napier W. M. 1984, MNRAS, 211, 953
Dones L., Levison H. F., Duncan M. 1996, ASP Conf. Ser. 107:
Completing the Inventory of the Solar System, 233
Durda D. D., Stern S. A. 2000, Icarus, 145, 220
Emel'yanenko V.V., Asher D.J., Bailey M.E. 2004, MNRAS, 350,
161
Evans N. W., Tabachnik S. 1999, Nature, 399, 41
c(cid:13) RAS, MNRAS 000, 1–15
Simulations of Centaurs I: Statistics
15
Fern´andez J. A. 1985, Icarus, 64, 308
Groussin O., Lamy P., Jorda L. 2004, A&A, 413, 1163
Hahn G., Bailey M. E. 1990, Nature, 348, 132
Hainaut O. R., Delsanti A. C. 2002, A&A, 389, 641
Holman M. J. 1997, Nature, 387, 785
Horner J., 2003, "The Behaviour of Small Bodies in the Outer
Solar System", D. Phil. thesis, University of Oxford.
Horner J., Evans N. W., Bailey M. E., Asher D. J. 2003, MNRAS,
343, 1057
Horner J., Evans N. W., Bailey M. E. 2004, MNRAS, submitted.
Lamy P.L., Toth I., Fern´andez Y.R., Weaver H.A. 2004, Comets
II, M. C. Festou, U. Keller, H. A. Weaver (eds), University of
Arizona Press, Tucson, in press
Levison H. F., Duncan M. J. 1994, Icarus, 108, 18
Levison H. F., Duncan M. J. 1997, Icarus, 127, 13
Lowry S. C., Fitzsimmons A., Collander-Brown S. 2003, A&A,
397, 329
Morbidelli A., Bottke W. F., Froeschl´e C., Michel P. 2002, As-
teroids III, W. F. Bottke Jr., A. Cellino, P. Paolicchi, and
R. P. Binzel (eds), University of Arizona Press, Tucson, p.409
Oikawa S., Everhart E. 1979, AJ, 84, 134
Opik E.J. 1976 Interplanetary Encounters, Elsevier, New York
Peixinho N., Lacerda P., Ortiz J. L., Doressoundiram A., Roos-
Serote M., Guti´errez, P. J. 2001, A&A, 371, 753
Press W.H., Teukolsky S.A., Vetterling W.T., Flannery B.P. 1992,
Numerical Recipes (Cambridge University Press: Cambridge),
chap. 15
Sanzovo G. C., de Almeida A. A., Misra A., Torres R. M., Boice
D. C., Huebner, W. F. 2001, MNRAS, 326, 852
van Woerkom A.J. 1948 Bull. Astron. Inst. Netherlands, 10, 445
Weintraub D. A., Tegler S. C., Romanishin W. 1997, Icarus, 128,
456
c(cid:13) RAS, MNRAS 000, 1–15
|
0707.1612 | 1 | 0707 | 2007-07-11T13:20:38 | Accretion disc coronae in black hole binaries | [
"astro-ph"
] | Most of the luminosity of accreting black hole is emitted in the X-ray band. This radiation is believed to emerge, through inverse Compton process, from a hot (Te ~ 10^8 -10^9 K) optically thin (Thomson optical depth ~ 1) plasma probably located in the immediate vicinity of the black hole. The mechanisms at work in this so called Compton corona can be unveiled through hard X-ray observations which have revealed a rich phenomenology. Depending on luminosity different spectral states are observed suggesting that the nature and geometry of the corona depends on mass accretion rate. In many instances the spectral behaviour as a function of luminosity shows some degree of hysteresis. The mechanisms triggering the transition between spectral states is very unclear although it could be related to an evaporation/condensation equilibrium in an accretion disc corona system. From the observation of correlation between the X-ray and radio band, it appears that the Compton corona is intimately related to the formation of compact jets and probably constitutes the base of the jet. | astro-ph | astro-ph | Mem. S.A.It. Vol. 75, 382
c(cid:13) SAIt 2007
Memorie della
A
ei di
ae i b a
k h e biaie
J. Malzac1
Centre d'Etude Spatiale des Rayonnements, OMP, UPS, CNRS; 9 Avenue du Colonel
Roche, 31028 Toulouse, France e-mail: [email protected]
Abstract. Most of the luminosity of accreting black hole is emitted in the X-ray band.
This radiation is believed to emerge, through inverse Compton process, from a hot (Te ∼
108 -109 K) optically thin (τT ∼ 1) plasma probably located in the immediate vicinity of
the black hole. The mechanisms at work in this so called Compton corona can be unveiled
through hard X-ray observations which have revealed a rich phenomenology. Depending
on luminosity different spectral states are observed suggesting that the nature and geome-
try of the corona depends on mass accretion rate. In many instance the spectral behaviour
as a function of luminosity show some degree of hysteresis. The mecanisms triggering the
transition between spectral states is very unclear although it could be related to an evapo-
ration/condensation equilibrium in an accretion disc corona system. From the observation
of correlation between the X-ray and radio band , it appears that the Compton corona is
intimately related to the formation of compact jets and probably constitutes the base of the
jet.
Key words. Radiation mechanisms: non-thermal -- Black hole physics -- Accretion, accre-
tion disks -- Magnetic fields -- Stars: coronae -- X-rays: binaries
1. Spectral states and the structure of
the accretion flow
Accreting black holes are observed in two main
spectral states (see e.g. Zdziarski & Girelinski
2004). At luminosities exceeding a few per-
cent of Eddington (LEdd), the spectrum is dom-
inated by a thermal component peaking at a
few keV which is believed to be the signa-
ture of a geometrically thin optically thick disc
(Shakura & Sunyaev 1974). At higher ener-
gies the spectrum is non-thermal and usually
present a weak steep power-law component
(photon index Γ ∼ 2.3 − 3) extending at least to
MeV energies, without any hint for a high en-
ergy cut-off. This soft power law is generally
Send offprint requests to: J. Malzac
interpreted as inverse Compton up-scaterring
of soft photons (UV, soft X) by a non-thermal
distribution of electrons. Since in this state the
source is bright in soft X-rays and soft in hard
X-rays it is called the High Soft State (hereafter
HSS).
At lower luminosities (L< 0.01 LEdd) , the
appearance of the accretion flow is very differ-
ent: the spectrum can be modelled as a hard
power-law Γ ∼ 1.5 − 1.9 with a cut-off at
∼ 100 keV. The νFnu spectrum then peaks
around a hundred keV. Since the soft X-ray lu-
minosity is faint and the spectrum is hard, this
state is called the Low Hard State (hereafter
LHS). LHS spectra are generally very well fit-
ted by Comptonisation (i.e. multiple Compton
up-scatering) of soft photons by a Maxwellian
7
0
0
2
l
u
J
1
1
]
h
p
-
o
r
t
s
a
[
1
v
2
1
6
1
.
7
0
7
0
:
v
i
X
r
a
Malzac: Accretion discs coronae
383
distribution of electrons in a hot (kTe ∼ 100
keV) plasma of Thomson optical depth τT of
order unity. As it will be discussed below the
LHS is asssociated with the presence of a com-
pact radio jet that seem to be absent in the
HSS. In addition to the dominant comptoni-
sation spectrum there are other less prominent
spectral features: a weak soft component asso-
ciated to the thermal emission of a geometri-
cally thin optically thick disk is occasionally
detected below 1 keV as observed for instance
in Cygnus X-1 (Balucinska-Church et al. 1995)
or in XTE J1118+480 (McClintock et al. 2001,
Chaty et al. 2003). Finally in both states, re-
flection features are generally detected in the
form of a Fe Kα line peaking at 6.4 kev and a
Compton reflection bump peaking at ∼ 30 keV.
These components are believed to be produced
when the hard X-ray emission of the corona in-
teracts and is reflected by the cold thick accre-
tion disc. They often appears to be broadened
through special and general relativistic effects,
in which case their origin must be very close to
the black hole. These reflection features appear
to be weaker in the LHS than in the HSS.
To sumarize, at high luminosities the accre-
tion flow is in the HSS caracterized by a strong
thermal disc and reflection component and
a weak non-thermal (or hybrid thermal/non-
thermal) comptonising corona. At lower lumi-
nosity the disc blackbody and reflection fea-
tures are weaker, while the corona is dominant
and emits through thermal Comptonisation.
Beside the LHS and HSS, there are several
other spectral states that often appear, but not
always, when the source is about to switch
from one of the two main states to the other.
Those states are more complex and difficult
to define. We refer the reader to McClintock
& Remillard (2006) and Belloni et al. (2005)
for two different spectral classifications based
on X-ray temporal as well as spectral criteria
and radio emission (Fender 2006). In general
their spectral properties are intermediate be-
tween those of the LHS and HSS.
The different spectral states are usually un-
derstood in terms of changes in the geometry
of the accretion flow. The standard picture is
that in the HSS there is a standard geometri-
cally thin disc extending down to the last sta-
ble orbit and responsible for the dominant ther-
mal emission. This disc is the source of soft
seed photons for Comptonisation in small ac-
tive coronal regions located above an below
the disc. Through magnetic buoyancy the mag-
netic field lines rise above the accretion disc,
transporting a significant fraction of the accre-
tion power into the corona where it is then
dissipated in through magnetic reconnection
(Galeev et al. 1977). In the corona particles
are accelerated. A population of high energy
electrons is formed which then cool down by
up scattering the soft photons coming from
the disc. This produces the high energy non-
thermal emission which in turn illuminates the
disc forming strong reflection features (see e.g.
Zdziarski & Gierli´nski 2004)
In the LHS, the standard geometrically thin
disc does not extend to the last stable orbit,
instead, the weakness of the thermal features
suggest that it is truncated at distances rang-
ing from a few hundreds to a few thousands
gravitational radii from the black hole (typi-
cally 1000 -- 10000 km). In these inner parts the
accretion flow takes the form of a hot geomet-
rically thick, optically thin disc. A solution of
such hot accretion flow was first discovered
by Shapiro, Lightman and Eardley (1976). In
these hot acccretion flow the gravitational en-
ergy is converted in the process of viscous dis-
sipation into the thermal energy of ions. The
main coupling between the electrons and the
ions is Coulomb collision which is rather weak
in the hot thin plasma. Since radiative cooling
of the ions is much longer than that of the elec-
trons, the ions temperature is much higher than
the electron temperature. This two temperature
plasma solution is thermally unstable (Pringle
1976) but can be stabilized if advection of the
hot gas into the black hole, or alternatively into
an outflow, dominates the energy transfert for
ions (Ichimaru 1977; Rees et al. 1982; Narayan
& Yi 1994, Abramowicz et al. 1995; Blandford
& Begelman 1999; Yuan 2001, 2003). In these
different models of advection dominated ac-
cretion flows (ADAF) most of the power is
not radiated, they are therefore radiatively in-
efficents. The electron have a thermal distri-
bution and cool down by Comptonisation of
the soft photons coming from the external geo-
384
Malzac: Accretion discs coronae
metrically thin disc, and IR-optical photons in-
ternally generated through self-absorbed syn-
chrotorn radiation. The balance between heat-
ing and cooling determines the electron tem-
perature which are found to be of order of 109
K as required to fit the spectrum. The weak
reflection features of the LHS are produced
through illumination of the cold outer disc by
the central source. By extension, the hot accre-
tion flow of the LHS is frequently designed as
"corona" despite the lack of a direct physical
analogy with the rarefied gaseous envelope of
the sun and other stars.
2. An alternative model for the LHS:
accretion disc corona
The presence of the hot accretion disc is not the
only possibility to explain the LHS. It was sug-
gested long ago that instead, a real accretion
disc corona system similar to that of the HSS
may reproduce the spectra as well (Bisnovatyi-
Kogan & Blinikov 1976; Liang & Price 1977)
. This would imply a cold geometrically thin
disc extending down very close to the black
hole in the LHS. Unlike in the HSS, this disc
does not produce a strong thermal compoonent
in the X-ray pectrum because it is too cold. It is
cold because most of the accretion power is not
dissipated in the disc, rather it is transported
away to power a strong corona and the compact
jet. This, at least in principle, could be achieved
either through transport via buyonancy of the
magnetic field (Miller & Stone 2000; Merloni
& Fabian 2002) or through the torque exerted
on the accretion disc by a strong magnetic field
threading the disc and driving the jet (Ferreira
1997).
Due to the complexity of the accretion disc
corona physics it is not possible to obtain sim-
ple analytical solutions giving the main prop-
erties of the corona unless some parametriza-
tion of the energy transfert between the disc
and the corona is used (see e.g. Svensson &
Zdziarski 1994). However, from an observer
perspective it does not really matter because
the appearence of the accretion disc corona
does not depend on the details of the energy
transport and dissipation mechanisms. Indeed,
Haardt & Maraschi (1991) realised the exis-
tence of a strong radiative feedback between
the cold disc and the hot corona. A fixed frac-
tion of the power dissipated in the corona in
the form of hard X-ray photons, impinge on
the cold disc where it is absorbed, heating the
cold disc, and finally reemited in the form of
thermal soft photons. A fraction of those soft
photon re-enters the corona providing the ma-
jor cooling effect to the coronna trough in-
verse Compton. As a consequence the cooling
rate of the corona scales like the heating rate.
The coronal temperature is determined only by
the geometry of the accretion disc corona (that
control the fraction of coronal power that re-
turns to the corona in the form of soft photons)
. For instance an extended corona sandwiching
the cold accretion disc, would intercept more
cooling photons from the accretion disc than
a patchy corona made of a few compact ac-
tive regions covering a small fraction of the
disc, and therefore would have a lower tem-
perature and hence a harder comptonised X-
ray spectra. During the nineties several group
(Haardt & Maraschi 1993; Stern et al. 1995 ;
Poutanen et al. 1996) performed detailed com-
putations of the resulting equilibrium spectra
for various geometries. They concluded that
the corona has to be patchy in order to produce
spectra that hare hard enough (Haardt et al.
1994; Stern et al. 1995) even when the effects
of disc ionisation are accounted for (Malzac
et al 2005). Yet, those accretion disc corona
models had an important problem: the produc-
tion of an unobserved strong thermal compo-
nent due to reprocessing of the radiation illu-
minating the disc. In the case of an active re-
gion emitting isotropically, about half of the lu-
minosity intercepts the disc and the thermal re-
processing component is comparable in lumi-
nosity to the primary emission, while in LHS
spectra the thermal comonent from the disc is
barely detectable. Morevover, the amplitude of
the reflection features observed in the LHS are
lower than what is expected from an istotropic
corona by at least a factor of 3, in some cases
they are so weak that they are not even de-
tected. This led to consider models where the
coronal emission is not isotropic. Beloborodov
(1999) suggested that the corona is unlikely to
stay at rest with respect to the accretion disc.
Malzac: Accretion discs coronae
385
Fig. 1. The black curve show the typical path followed by a black hole X-ray transient in the hardness vs
intensity diagram during an outburst. The various sketches illustrate the standard scenario for the evolution
of the geometry of the corona-disc-jet system along this path (see text).
Due to the anisotropy of the dissipation process
or simply to radiation pressure from the disc,
the hot plasma is likely to be moving at midly
relativistic velocities. Then, due to Doppler ef-
fects, the X-ray emission is strongly beamed in
the direction of the plasma velocity. In the case
of a velocity directed away from the accretion
disc, (outflowing corona) , the reprocessing
features (both reflection and thermalised radia-
tion) are strongly suppressed. Moreover, due to
the reduced feedback from the disc, the corona
is hotter, and harder spectra can be produced.
Malzac, Beloborodov & Poutanen (2001) per-
formed detailled non-linear Monte-Carlo sim-
ulation of this dynamic accretion disc corona
equilibrium, and compared the results with the
observations. They found that compact actives
regions of aspect ratio of order unity, outflow-
ing with a velocity of 30 percent of the speed
of light could reproduce the LHS spectrum of
Cygnus X-1. Moreover since the velocity of
the coronal plasma controls both the strength
of the reflection features and the feedback of
soft cooling photons from the disc, it predicts
a correlation between the slope of the the hard
X-ray spectrum and the amplitude of the reflec-
tion component. Such a correlation is indeed
observed in several sources (Zdziarski et al.
1999, 2003) and is well matched by this model.
Recently the accretion disc corona mod-
els for the LHS obtained more observationnal
support, with the discovery of relativistically
broadened iron line in the LHS of GX339-
4 (Miller et al., 2006). Such relativistically
386
Malzac: Accretion discs coronae
broadened lines require that disc illumination
is taking place very close to the black hole.
This observation suggests that, at least in some
cases, a thin disc is present at or close to the
last stable orbit in the LHS.
3. The jet coronna connection
Recent multi-wavelength observations of ac-
creting black holes in the LHS have shown the
presence of an ubiquitous flat-spectrum radio
emission (see e.g Fender 2006), that may ex-
tend up to infrared and optical wavelengths.
The properties the radio emission indicate it is
likely produced by synchrotron emission from
relativistic electrons in compact, self-absorbed
jets (Blandford & Konigl, 1979; Hjellming
& Johnston 1988). This idea was confirmed
by the discovery of a continuous and steady
milliarcsecond compact jet around Cygnus X-
1 (Sirling 2001). Moreover, in LHS sources
a tight correlation has been found between
the hard X-ray and radio luminosities, hold-
ing over more than three decades in luminos-
ity (Corbel et al. 2003; Gallo, Fender & Pooley
2003). In contrast, during HSS episodes the
sources appear to be radio weak (Tananbaum
et al. 1972; Fender et al. 1999; Corbel 2000),
suggesting that the Comptonising medium of
the low/hard state is closely linked to the con-
tinuous ejection of matter in the form of a small
scale jet.
When the importance of the connection be-
tween radio and X-ray emission was realised.
It was proposed that the hard X-ray emission
could be in fact synchrotron emission in the
jet, rather than comptonisation in a hot accre-
tion flow/corona (Markoff, Falke, Fender 2001;
Markoff et al. 2003). This model is able to ex-
plain quantitatively the correlation between the
X-ray and radio emission, it is also able to re-
produce at least roughly the shape of the LHS
X-ray spectrum of several sources. However,
it seems that synchrotron emission alone is
not enough to reproduce the details of the X-
ray spectra. In the most recent version of this
model a thermal Comptonisation component
was added which appears to provide a domi-
nant contribution to the hard X-ray spectrum
(Markoff et al., 2005). This component is sup-
posedly formed in the base of the jet which
forms a hot plasma that present strong simi-
larities with an accretion disc corona. In the
context of accretion disc coronae/hot disc mod-
els the correlation between X-ray and radio
emission, tells us that the corona and the com-
pact jet of the LHS are intimately connected.
A strong corona may be necessary to launch a
jet and/or could be the physical location where
the jet is accelerated or launched (Merloni &
Fabian 2002). Koerding et al. (2006) show that
the observed X-radio correlation can be repro-
duced provided that the hot flow/corona is ra-
diatively inefficient (i.e. luminosity scales like
the square of mass accretion rate) and that a
constant fraction of the accretion power goes
into the jet (i.e. jet power scales like mass ac-
cretion rate). However there is presently no de-
tailed model to explain how the physical con-
nection between jet and corona works. A pos-
sible basic explanation was proposed by Meier
(2001) for ADAF like accretion flows and later
extended to the case of accretion disc coronae
by Merloni and Fabian (2002). It goes as fol-
low, since models and simulations of jet pro-
duction (Blandford & Znajek 1977; Blandford
& Payne 1982, Ferreira 1997) indicate that jets
are driven by the poloidal component of the
magnetic field. If we assume that the magnetic
field is generated by dynamo processes in the
disc/corona, the strengh of the poloidal compo-
nent is limited by the scale height of the flow
(Livio, Ogilvie & Pringle 1999; Meier 2001;
Merloni & Fabian 2002). Therefore geometri-
cally thick accretion flow should be naturally
more efficient at launching jets.
Despite this strong link beween the corona
and outflow in the LHS, there are indications
that in Intermediate States (IS) the jet is con-
nected to the accretion disc rather than the
corona. For instance Malzac et al. (2006) re-
port the results of an observation of Cygnus X-
1 during a mini state transition. The dtat indi-
cate that the jet power is anti-correlated with
the disc luminosity and unrelated to the coro-
nal power. This is in sharp contrast with previ-
ous results obtained for the LHS, and suggests
a different mode of coupling between the jet,
the cold disc, and the corona in Intermediate
States.
Malzac: Accretion discs coronae
387
4. Evolution of the geometry during
outburst: the hysteresis problem
Transient black hole binaries are a class of X-
ray binaries that are detected only occasionally
when they show some period of intense activity
(outbursts) lasting for a few month (Tanaka &
Levin 1995). These outbursts have recurrence
times ranging from years to dozens of years.
Between two episodes of activity the source is
in an extremely faint quiescent state. During an
outburst, the source luminosity varies by many
orders of magnitude. The study of spectral evo-
lution of the sources during the outbursts is
perfectly suited to understand how the struc-
ture of the corona depends on the mass accre-
tion rate. The spectral evolution during an out-
burst is conveniently described using hardness
intensity diagrams. Those diagrams show the
evolution of the source luminosity as a func-
tion of spectral hardness (defined as the ra-
tio of the observed count rates in a high to
a lower energy band). Typically, during out-
bursts the sources follow a Q-shaped path in
the hardness intensity diagram (see Figure 1).
Initially in quiescence with a very low lumi-
nosity, they are in the LHS with a high hard-
ness. In the standard scenario1, the cold disc is
truncated very far away from the black hole,
and the emission is dominated by the hot ac-
cretion flow. Then, as luminosity increases by
several orders of magnitude the hardness re-
mains constant. The structure of the accretion
flow is stable, the cold disc inner radius prob-
ably moving slowly inward. During this rise
in X-ray luminosity a compact jet is present
and its radio emission correlates with the X-ray
emission. At a luminosity that is always above
a few percent of Eddington, the inner radius
of the cold accretion disk decreases quickly.
This reduction of the inner disc radius is as-
sociated with either the cold disk penetrating
inside the hot inner flow, or the later collaps-
ing into an optically thick accretion disk with
small active regions of hot plasma on top of it
(Zdziarski et al. 2002). In both cases the en-
1 We note that a different geometrical interpreta-
tion of the outburts cycle in terms of a jet dominated
accretion flow was recently proposed by Ferreira et
al. (2006).
hanced soft photon flux from the disk tends
to cool down the hot phase, leading to softer
spectra. The source therefore move from the
right toward the left hand side at almost con-
stant luminosity. During this hard to soft state
transition the sources usually show a strong
coronal activity (both the disc and corona are
strong), this is also during this transition that
optically thin radio flares associated with rel-
ativistic and sometimes superluminal ejections
are observed. This region of the hardness in-
tensity diagram is often called the Very High
State (VHS). Once a source has reached the
upper left hand corner of the hardness intensity
diagram, it never gets back to the LHS follow-
ing the same path. Instead when the luminosity
decreases, it stays in the HSS and when lumi-
nosity decreases, goes down vertically and the
strength of the non-thermal corona decreases,
leaving a bare multiblackbody disc spectrum.
This goes on until the luminosity reaches 0.02
percent of Eddington. Then it moves horizon-
tally to the right, back to the LHS, and then
down vertically back to quiescence.
It is to be noted that the luminosity of state
transition from soft to hard seems to be rela-
tively fixed around 0.02 percent of Eddington
(Maccaronne et al. 2003). On the the other
hand the hard to soft transition can vary, even
for the same source, by several order of mag-
nitude and it is always higher than the soft
to hard transition luminosity. The question of
what controls the different transitions luminos-
ity is still open. Indeed, such a complex be-
haviour is not expected a priori, since in all ac-
cretion models, there is only one relevant ex-
ternally imposed parameter: the mass acretion
rate. It was not anticipated that for the same
luminosity the accretion flow could be in dif-
ferent states. It was recently suggeted that this
complex behaviour could be explained by the
existence of a second parameter that could be
the magnetic flux advected with the accreting
material Spruit & Uzdensky (2005). But the
most detailed explanations so far are based on
the idea that this second parameter is nothing
else than the history of the system. In other
word the accretion flow would present an hys-
teresis behaviour because going from LHS to
HSS is not the same thing as going from HSS
388
Malzac: Accretion discs coronae
to LHS. Within this line of reasoning, the most
promising idea so-far was proposed by Meyer-
Hofmeister et al. (2005) who suggested that
the hysteresis could be linked to a conden-
sation/evaporation equilibrium in the acretion
disc corona system. They assume an advective
corona on top of a thin optically thick accre-
tion disc and allowing for mass exchange be-
tween the corona and the disc. Depending on
the temperature and density in the corona and
the disc, the material in the corona may con-
densate into the disc and the disc may evapo-
rate. Meyer-Hofmeister et al. (2005) have com-
puted the resulting equilibrium between an ac-
cretion disc and a corona as a function of the
distance to the black hole and mass accre-
tion rate (see also R´oza´nska & Czerny, 2000;
Mayer & Pringle, 2006). They have found that
at large distance the conditions are such that an
equilibrium solution can always be found. At
closer distances from the black hole, disc evap-
oration becomes more and more important and,
under some circumstances there is a critical ra-
dius below which the disc fully evaporate and
no thin disc is possible. Below this radius we
are then left with a pure ADAF. The location
of this transisition radius depends on the mass
accretion rate. At low mass accretion rate the
transition radius is at 103 to 104 Schwarschild
radii and decreases slowly as the the accretion
rate increases until the critical mass accretion
of about a few percent of Eddington is reached.
Then, the inner radius suddenly drops to the
last stable orbit: this is the hard to soft state
transition. As long as the source is in the LHS
the flux of soft cooling photons in the corona
is weak, and, as a consequence the corona is
hot and evaporation dominates over condensa-
tion. Once the HSS is reached however, there
is a disc down to the last stable orbit emit-
ting a strong thermal soft emission that effi-
ciently cools the corona, keeping the conden-
sation process strong even when the mass ac-
cretion rate decreases. As a consequence the
transition from the soft to LHS occurs at a lu-
minosity that is about one order of magnitude
lower then the hard to soft transition.
Although this model so-far provides the
best explanation of the hysteresis behaviour, it
is not perfect. First, the presence of the jet and
possibly impacting on the disc corona equi-
librium are completely ignored. Moreover this
model does not explain why the transition LHS
to HSS can occurs at different luminosities (the
transition luminosity is quite constrained in
this model) and a second independent param-
eter may still be required.
References
Abramowicz, M. A., Chen, X.-M., Granath,
M., & Lasota, J.-P. 1996, ApJ, 471, 762
Balucinska-Church, M., Belloni, T., Church,
M. J., & Hasinger, G. 1995, A&A, 302, L5
Belloni, T., Homan, J., Casella, P., van der Klis,
M., Nespoli, E., Lewin, W. H. G., Miller, J.
M. & M´endez, M. 2005, A&A, 440, 207
Beloborodov, A. M. 1999, ApJl, 510, L123
Bisnovatyi-Kogan, G. S., & Blinnikov, S. I.
1976, Soviet Astronomy Letters, 2, 191
Blandford, R. D., & Begelman, M. C. 1999,
MNRAS, 303, L1
Blandford, R. D., & Konigl, A. 1979, ApJ, 232,
34
Blandford, R. D., & Payne, D. G. 1982,
MNRAS, 199, 883
Blandford, R. D., & Znajek, R. L. 1977,
MNRAS, 179, 433
Brocksopp, C., Fender, R. P., Larionov, V.,
Lyuty, V. M., Tarasov, A. E., Pooley, G. G.,
Paciesas, W. S., & Roche, P. 1999, MNRAS,
309, 1063
Chaty, S., Haswell, C. A., Malzac, J., Hynes,
R. I., Shrader, C. R., & Cui, W. 2003,
MNRAS, 346, 689
Churazov, E., Gilfanov, M., & Revnivtsev, M.
2001, MNRAS, 321, 759
Corbel S., Fender R.P., Tzioumis A.K., Nowak
M., McIntyre V., Durouchoux P., Sood R.,
2000, A&A, 359, 251
Corbel S., Nowak M. A., Fender R. P.,
Tzioumis A. K., Markoff S., 2003, A&A,
400, 1007
Fender R. et al., 1999, ApJ, 519, L165
Fender, R. P. 2001, MNRAS, 322, 31
Fender, R.P., 2006,
in Compact Stellar X-
ray Sources, ed. W.H.G. Lewin & M. van
der Klis (Cambridge: Cambridge University
Press), p 381 astro-ph/0303339.
Ferreira, J. 1997, A&A, 319, 340
Malzac: Accretion discs coronae
389
Ferreira, J., Petrucci, P.-O., Henri, G., Saug´e,
L., & Pelletier, G. 2006, A&A, 447, 813
Galeev, A. A., Rosner, R., & Vaiana, G. S.
1979, ApJ, 229, 318
Gallo E., Fender R. P., Pooley G. G., 2003,
MNRAS, 344, 60
Gierli´nski, M., Zdziarski, A., A., Done, C.,
Johnson, W. N., Ebisawa, K., Ueda, Y.,
Haardt, F., Phlips, B. F. 1997, MNRAS, 288,
958
Haardt, F., & Maraschi, L. 1991, ApJl, 380,
L51
Haardt, F., & Maraschi, L. 1993, ApJ, 413, 507
Haardt, F., Maraschi, L., & Ghisellini, G. 1994,
ApJl, 432, L95
Hjellming R.M., Johnston K.J., 1988, ApJ,
328, 600
Ichimaru, S. 1977, ApJ, 214, 840
Kording, E. G., Fender, R. P., & Migliari, S.
2006, MNRAS, 369, 1451
Liang, E. P. T., & Price, R. H. 1977, ApJ, 218,
247
Livio, M., Ogilvie, G. I., & Pringle, J. E. 1999,
ApJ, 512, 100
Maccarone, T. J. 2003, A&A, 409, 697
Malzac, J., Beloborodov, A. M., & Poutanen,
J. 2001, MNRAS, 326, 417
Malzac, J., Dumont, A. M., & Mouchet, M.
2005, A&A, 430, 761
Markoff, S., Falcke, H., & Fender, R. 2001,
A&A, 372, L25
Markoff, S., Nowak, M., Corbel, S., Fender, R.,
& Falcke, H. 2003, New Astronomy Review,
47, 491
Markoff, S., Nowak, M. A., & Wilms, J. 2005,
ApJ, 635, 1203
Miller, K. A., & Stone, J. M. 2000, ApJ, 534,
398
Miller, J. M., Homan, J., Steeghs, D., Rupen,
M., Hunstead, R. W., Wijnands, R., Charles,
P. A., & Fabian, A. C. 2006, ApJ, 653, 525
Narayan, R., & Yi, I. 1994, ApJl, 428, L13
Poutanen, J., & Svensson, R. 1996, ApJ, 470,
249
Pringle, J. E. 1976, MNRAS, 177, 65
Rees, M. J., Phinney, E. S., Begelman, M. C.,
& Blandford, R. D. 1982, Nature, 295, 17
R´oza ´Nska, A., & Czerny, B. 2000, A&A, 360,
1170
Shakura, N. I. , & Sunyaev, R. A., 1973, A&A,
24, 337
Shapiro, S. L., Lightman, A. P., & Eardley,
D. M., 1976, ApJ, 204, 187
Spruit, H. C., & Uzdensky, D. A. 2005, ApJ,
629, 960
Stern, B. E., Poutanen, J., Svensson, R., Sikora,
M., & Begelman, M. C. 1995, ApJl, 449,
L13
Stirling, A. M., Spencer, R. E., de la Force,
C. J., Garrett, M. A., Fender, R. P., & Ogley,
R. N. 2001, MNRAS, 327, 1273
Tanaka Y., Lewin W.H.G., 1995,
in 'X-
ray Binaries', eds. W.H.G. Lewin, J. van
Paradjis, E. van den Heuvel. Cambridge
University Press.
Svensson, R., & Zdziarski, A. A. 1994, ApJ,
436, 599
Tananbaum H., Gursky H., Kellogg E.,
Giacconi R., Jones C., 1972, ApJ, 177, L5
Yuan, F. 2001, MNRAS, 324, 119
Yuan, F. 2003, ApJl, 594, L99
Zdziarski, A. A., Lubinski, P., & Smith, D. A.
Mayer, M., & Pringle, J. E. 2006, MNRAS, in
1999, MNRAS, 303, L11
press, astro-ph/0612751
McClintock, J.E., & Remillard, R.A. 2006, in
Compact Stellar X-ray Sources, ed. W.H.G.
Lewin & M. van der Klis (Cambridge:
Cambridge University Press), p157 astro-
ph/0306213.
McClintock, J. E., et al. 2001, ApJ, 555, 477
Merloni, A., & Fabian, A. C. 2002, MNRAS,
332, 165
Meier, D. L. 2001, ApJl, 548, L9
Meyer-Hofmeister, E., Liu, B. F., & Meyer, F.
2005, A&A, 432, 181
Zdziarski, A. A., Lubi´nski, P., Gilfanov, M., &
Revnivtsev, M. 2003, MNRAS, 342, 355
Zdziarski,
A.
A.,
Gierli´nski, M.,
Mikołajewska, J., Wardzi´nski, G., Smith,
D. M., Alan Harmon, B., & Kitamoto, S.
2004, MNRAS, 351, 791
Zdziarski, A. A., & Gierli´nski, M.
2004, Progress of Theoretical Physics
Supplement, 155, 99
|
astro-ph/0310094 | 1 | 0310 | 2003-10-03T13:23:57 | An HI census of Loose Groups of Galaxies | [
"astro-ph"
] | We present results from our Parkes Multibeam HI survey of 3 loose groups of galaxies that are analogous to the Local Group. This is a survey of groups containing only spiral galaxies with mean separations of a few hundred kpc, and total areas of approximately 1 sq. Mpc; groups similar to our own Local Group. We present a census of the HI-rich objects in these groups down to an M(HI), 1-sigma sensitivity ~7x10^5 M(sun), as well as the detailed properties of these detections from follow-up Compact Array observations. We found 7 new HI-rich members in the 3 groups, all of which have stellar counterparts and are, therefore, typical dwarf galaxies. The ratio of low-mass to high-mass gas-rich galaxies in these groups is less than in the Local Group meaning that the ``missing satellite'' problem is not unique. No high-velocity cloud analogs were found in any of the groups. If HVCs in these groups are the same as in the Local Group, this implies that HVCs must be located within ~300-400 kpc of the Milky Way. | astro-ph | astro-ph |
**TITLE**
ASP Conference Series, Vol. **VOLUME***, **YEAR OF PUBLICATION**
**NAMES OF EDITORS**
An H I Census of Loose Groups of Galaxies
D.J. Pisano
NSF Distinguished International Postdoctoral Research Fellow, ATNF,
P.O. Box 76, Epping NSW 1710, Australia
David G. Barnes
School of Physics, Univ. of Melbourne, Victoria 3010, Australia
Brad K. Gibson
Swinburne University of Technology, Centre for Astrophysics and
Supercomputing, Mail # 31, , P.O. Box 218, Hawthorn, Victoria 3122,
Australia
Lister Staveley-Smith
ATNF, P.O. Box 76, Epping NSW 1710, Australia
Ken C. Freeman
RSAA, Mount Stromlo Observatory, Cotter Road, Weston, ACT 2611,
Australia
Abstract. We present results from our Parkes Multibeam H I survey of
3 loose groups of galaxies that are analogous to the Local Group. This is a
survey of groups containing only spiral galaxies with mean separations of a
few hundred kpc, and total areas of approximately 1 Mpc2; groups similar
to our own Local Group. We present a census of the H I-rich objects in
these groups down to a 1σ MH I sensitivity ∼7×105M⊙, as well as the
detailed properties of these detections from follow-up Compact Array
observations. We found 7 new H I-rich members in the 3 groups, all of
which have stellar counterparts and are, therefore, typical dwarf galaxies.
The ratio of low-mass to high-mass gas-rich galaxies in these groups is
less than in the Local Group meaning that the "missing satellite" problem
is not unique. No high-velocity cloud analogs were found in any of the
groups. If HVCs in these groups are the same as in the Local Group, this
implies that HVCs must be located within ∼300-400 kpc of the Milky
Way.
1.
Introduction
Loose groups of galaxies are collections of a few (2-3) large galaxies and tens of
smaller galaxies. They are the most diffuse components of structure in the uni-
verse, yet they are relatively understudied despite their importance. About 60%
1
2
of galaxies reside in groups (Tully 1987) including the Milky Way which is part
of the Local Group of galaxies. Loose groups are possibly still forming (Zablud-
off & Mulchaey 1998), and may even be the site of ongoing galaxy formation
as traced by the high-velocity clouds (HVCs; e.g. Blitz et al. 1999). Measuring
the distribution of galaxy masses in groups also provides a useful constraint on
models of galaxy formation. To learn more about groups of galaxies similar to
our own, to constrain the nature of HVCs, and to learn more about galaxy and
structure formation, we have commenced a survey of loose groups of galaxies.
2. Observations
Our survey examined analogs to the Local Group: spiral-rich loose groups with-
out any large elliptical or lenticular galaxies. We selected five such groups from
the LGG catalog of Garcia (1993). A sixth group was selected from the HIPASS
group catalog (Stevens, 2003, private communication) and was not previously
identified optically. Assuming H0 = 65 km/s/Mpc, the groups lie between 10.6 -
13.4 Mpc. We observed a projected area of 1 - 1.7 Mpc2 centered on each group
using the Parkes Multibeam receiver. The observations involved scanning the
instrument in a "basket-weave" pattern in right ascension and declination mul-
tiple times until an RMS sensitivity of ∼7×105M⊙ per 3.3 km/s was reached.
The total velocity coverage of the observations was either 1700 or 3400 km/s
(depending on the group) with a velocity resolution of 1.65 km/s or 3.3 km/s.
The linear resolution of the Parkes observations was ∼50 kpc.
The data were reduced and gridded into cubes using the ATNF livedata
and gridzilla packages in aips++. The final cubes were searched by eye by
three people for detections. Fake sources were inserted into these cubes to assess
the reliability and completeness of the search. If a source was found by at least 2
people, it was considered to be a detection. Based on this analysis, our detection
algorithm was essentially 100% complete down to the 10σ level.
All our detections, not just the new ones, were re-observed with the ATCA.
The ATCA observations not only served to confirm the detections, but the higher
spatial resolution, ∼4 kpc, allowed us to uniquely identify H I detections with
optical counterparts. Furthermore, as the ATCA observations had about the
same sensitivity as the Parkes cubes, we were able to search for H I clouds
which were previously unresolved in the Parkes data.
The combination of the large area observed, high velocity resolution, and
extremely sensitive observations of multiple loose groups makes this survey
uniquely tuned to get a census of the H I content of loose groups of galaxies
and to strongly constrain the origin of HVCs. We will address these topics using
the first 3 groups studied in our survey: LGG 93, LGG 180, and LGG 478.
3. The H I Content of Loose Groups of Galaxies
Cold dark matter simulations of the formation of the Local Group of galaxies
(e.g. Klypin et al. 1999, Moore et al. 1999) reveal an extremely large number
of small dark matter halos (∼300) compared to the number of known luminous
satellite galaxies (∼20). This is the so-called "missing satellite" problem. While
invoking a different type of dark matter, such as warm dark matter, or invoking
3
Figure 1. MH I vs. Vrot for Local Group galaxies (from Mateo 1998)
with measured MH I (circles) and upper limits (arrows). The stars
indicate the sources in LGG 93. The line indicates the 10σ detection
limit
feedback processes to suppress the collapse of baryons in dark halos can solve
this problem, it is worth asking if this problem exists at the same level in groups
similar to the Local Group.
Our survey detected all the optically identified galaxies in the 3 groups. In
addition, we detected 4 new group members in LGG 93, 2 new group members
in LGG 180, and 1 new group member in LGG 478 (along with 7 background
galaxies). Only 4 of these new detections were not evident in the HIPASS data
(Barnes et al. 2001), however. All of these new detections have been confirmed
with ATCA, and have MH I = 107−9M⊙. All of the detections have associated
stellar components; they are all dwarf galaxies.
Our survey found that the "missing satellite" problem is actually worse in
the groups we surveyed than in the Local Group. Figure 1 shows that if we were
observing the Local Group, we would detect a total of 14 galaxies, 9 of which
have FWHM ≤100 km s−1 and could be considered dwarf galaxies. Data for
the Local Group galaxies comes from Mateo (1998). The dwarf-to-giant ratio is
1.45 for the three groups we studied, compared with 1.8 for the Local Group.
This discrepancy will be verified when we finish analyzing the remaining three
groups in our sample, but, at the present, it appears that the Local Group is not
unique in its deficit of luminous satellite galaxies compared to cold dark matter
models of galaxy formation.
4
4. The Nature of High-Velocity Clouds
HVCs are clouds of H I seen all around the Milky Way, but which are not in
regular galactic rotation (see Wakker, these proceedings for a review). As such,
their distances and masses are unknown, making their origins uncertain. Some
HVCs are certainly associated with the Magellanic Stream which is the result of
the tidal interaction between the LMC, SMC, and Milky Way, and it is possible
that other complexes have similar origins (e.g. Lockman 2003). Other HVCs
may be ejected material from a Galactic fountain, or may be primordial gas
falling into the Local Group and onto the Milky Way as part of ongoing galaxy
formation.
While the idea of HVCs being associated with galaxy formation is an old
one, interest in the idea was revived by Blitz et al. (1999) and Braun & Burton
(1999) who suggested that HVCs and compact HVCs (CHVCs) may contain
dark matter, have masses of ∼107M⊙, and reside at distances of ∼ 1 Mpc from
the Milky Way. These HVCs may even be associated with the large population
of dark matter halos seen in the simulations discussed above. This idea has
been expanded on by de Heij, Braun, & Burton (2002b), who proposed that
CHVCs were concentrated around the Milky Way and M31 with a Gaussian
distance distribution of width 150 - 200 kpc, and MH I ≤107M⊙. If CHVCs are
associated with galaxy formation, then some of their analogs should be visible
in the groups we have observed.
Starting with this premise, we have constructed a simple model for HVCs
to predict how many should be seen in the groups we have studied. We begin
by taking the integrated fluxes and velocity widths of all Milky Way CHVCs
cataloged by Putman et al. (2002) and de Heij, Braun, & Burton (2002a). We
then randomly determine a distance assuming a Gaussian distance distribution
of a given half-width, half-maximum. Given this distance, we get an H I mass
for each cloud. Now we ask if this CHVC were in one of our groups, would we
detect it at a 10σ level. We do this for every CHVC to determine the number
of expected detections per group for a given DHW HM , and then we repeat the
process 100,000 times to get a distribution of expected detections. The results
of these trials for two groups, LGG 93 and LGG 478, and two DHW HM values
are shown in Figure 2.
As we already stated in Section 3, we have found no HVC analogs (H I
clouds without stars) in any of our groups. Therefore, in looking at Figure 2,
we want to know for a given DHW HM what the probability of zero detections is.
Since this is a function of the mass detection limit of the group, we get different
limits on DHW HM for each group. For LGG 93, we can see in Figure 2 that
DHW HM is less than 400 kpc at a 99.76% confidence level, but DHW HM < 250
kpc is only 73% likely. For LGG 478, the figure demonstrates that DHW HM <
400 kpc is 99.997% likely, and DHW HM < 250 kpc is 93.8% certain. And for LGG
180, not shown in the figure, the limits are 99.98% and 89% certain, respectively.
There a couple of notes of caution regarding this analysis. First, it is im-
portant to note that the vast majority of CHVC analogs are not detected; we
can only detect the most massive analogs. This means that we are using a small
number of objects to infer the properties of a larger population. If the popula-
tion of CHVCs in these groups is different than in the Local Group, we may not
be able to detect any analogs. We also assume that the same number of CHVCs
5
Figure 2.
The probability of a given number of detections for LGG 93
(left) and LGG 478 (right). The solid line represents the probability
distribution for DHW HM = 400 kpc and the dashed line represents
DHW HM = 250 kpc.
is present in each group. If the number of CHVCs is a function of group mass,
for example, we would expect less CHVCs in LGG 478 than in LGG 93 or LGG
180, so the constraints on DHW HM may not be as strong as suggested above.
Nevertheless by analogy to other groups, it appears that CHVCs should be clus-
tered with DHW HM ≤ 300-400 kpc of the Milky Way. At these distances, the
total H I mass in CHVCs is only ∼108M⊙, making them an important source
of fuel for star formation, but not dynamically important to the Local Group.
5. Conclusions
We have surveyed ∼1 Mpc2 around the centers of 3 loose groups of galaxies
using the Parkes Multibeam and the ATCA. The goal of these observations is
to get a census of the H I-rich galaxies in these groups and to search for analogs
to HVCs. This will permit us to constrain models for the origin of HVCs and
to begin to test models of galaxy formation. These groups were chosen to be
analogs to the Local Group, so that they only contain a few large spiral galaxies
which are separated by a few hundred kiloparsecs. Our observations have very
high velocity resolution, <3.3 km/s, in order to facilitate the detection of low-
mass objects, so that our 1σ detection limit is ∼7×105M⊙ per 3.3 km/s. Using
fake sources to determine our completeness and ATCA observations to confirm
our detections, we are roughly 100% complete and reliable at 10× the theoretical
noise limit.
We found 7 new H I-rich objects in the groups, plus 7 background galaxies,
only 4 of which were not seen in HIPASS, and all of which have stellar coun-
terparts. The ratio of low-mass to high-mass galaxies in these groups is less
than expected from simulations and also less than the ratio in the Local Group,
therefore the "missing satellite" problem is not unique to our local neighborhood.
We found no analogs to HVCs in any of the three groups. Assuming the
HVCs in these groups are the same as in the Local Group, this implies that
compact HVCs must be clustered within ∼300-400 kpc of the Milky Way, oth-
6
erwise we would have seen their analogs in our survey. This is strong evidence
against the Blitz et al. (1999) and Braun & Burton (1999) models for the origin
of HVCs, but is still consistent with the de Heij et al. (2002b) model. At these
distances, however, the total H I mass in CHVCs is only ∼108M⊙, making them
unimportant to the dynamics of the Local Group, but still a useful reservoir of
fuel for future star formation.
We have Parkes data in hand for an additional 3 groups, and will be fol-
lowing up our detections in these groups with the ATCA and the VLA in the
near future. The ensemble of the data on these 6 groups will place stronger
constraints on the nature of HVCs and provide a better test of different models
of galaxy formation.
Acknowledgments. The authors wish to thank the excellent staff at Parkes
and the ATCA for their assistance with observing. We wish to thank Martin
Zwaan for his assistance with inserting fake sources into our data cubes. D.J.P.
acknowledges generous support from NSF MPS Distinguished Research Fellow-
ship grant AST0104439.
References
Barnes, D.G., 2001, MNRAS, 322, 486
Blitz, L., Spergel, D., Teuben, P.J., Hartmann, D., Burton, W.B., 1999, ApJ,
514, 818
Braun, R., Burton, W.B., 1999, A&A, 351, 437
de Heij, V., Braun, R., Burton, W.B., 2002a, A&A, 391, 67
de Heij, V., Braun, R., Burton, W.B., 2002b, A&A, 392, 417
Garcia, A.M., 1993, A&AS, 100, 47
Klypin, A.A., Kravtsov, A.V., Valenzuela, O., Prada, F., 1999, ApJ, 522, 82
Lockman, F.J., 2003, ApJ, 591, L33
Mateo, M., 1998, ARA&A, 36, 435
Moore, B., Ghigna, S., Governato, F., Lake, G., Quinn, T., Stadel, J., 1999,
ApJ, 524, L19
Putman, M.E., et al., 2002, AJ, 123, 873
Tully, R.B., 1987, ApJ, 321, 280
Zabludoff, A.I., Mulchaey, J.S., 1998, ApJ, 498, L5
|
astro-ph/0703158 | 1 | 0703 | 2007-03-07T20:13:28 | Dust Echoes from the Ambient Medium of Gamma-Ray Bursts | [
"astro-ph"
] | Long gamma-ray bursts (GRBs) are likely associated with the collapse of massive stars, which produce dust and are born in dusty environments. Absorption and scattering of ultraviolet/X-ray photons from the prompt, optical flash and afterglow emission of the GRB produce dust echoes. We perform time-dependent calculations of these echoes, accounting for the evolution of the dust grain distribution due to selective grain destruction by the GRB radiation, and for off-axis beaming. We explore cloud configurations of differing density and size -- the echo light curve and spectrum depend on the cloud radius, with larger clouds peaking at longer wavelengths. For a region ~ 3 pc in size with hydrogen density ~ 1000 per cubic centimeter, the echo spectrum peaks at ~ 3.6 microns and ~ 8.8 eV for thermal and scattered components, respectively. Dust echoes should be detectable with the Very Large Telescope up to z ~ 0.1, IRAC onboard the Spitzer Space Telescope up to z ~ 0.2, and NICMOS onboard the Hubble Space Telescope up to z ~ 0.3. Furthermore, the shape of the echo light curve allows one to infer: the jet opening angle; the inclination of the jet axis with respect to the line of sight; the size of the dust-emitting region. For sources with symmetric, bipolar jets, dust echoes exhibit two bumps in the light curve, making them easily distinguishable from the rebrightening due to an underlying supernova. | astro-ph | astro-ph | DUST ECHOES FROM THE AMBIENT MEDIUM OF
GAMMA-RAY BURSTS
Kevin Heng1, Davide Lazzati1 & Rosalba Perna1
ABSTRACT
Long gamma-ray bursts (GRBs) are likely associated with the collapse of mas-
sive stars, which produce dust and are born in dusty environments. Absorption
and scattering of ultraviolet/X-ray photons from the prompt, optical flash and
afterglow emission of the GRB produce dust echoes. We perform time-dependent
calculations of these echoes, accounting for the evolution of the dust grain distri-
bution due to selective grain destruction by the GRB radiation, and for off-axis
beaming. We explore cloud configurations of differing density and size -- the
echo light curve and spectrum depend on the cloud radius, with larger clouds
peaking at longer wavelengths. For a region ∼ 3 pc in size with density nH ∼ 103
cm−3, the echo spectrum peaks at ∼ 3.6 µm and ∼ 8.8 eV for thermal and scat-
tered components, respectively. Dust echoes should be detectable with the Very
Large Telescope up to z ∼ 0.1, IRAC onboard the Spitzer Space Telescope up to
z ∼ 0.2, and NICMOS onboard the Hubble Space Telescope up to z ∼ 0.3. Fur-
thermore, the shape of the echo light curve allows one to infer: the jet opening
angle; the inclination of the jet axis with respect to the line of sight; the size of
the dust-emitting region. For sources with symmetric, bipolar jets, dust echoes
exhibit two bumps in the light curve, making them easily distinguishable from
the rebrightening due to an underlying supernova.
Subject headings: gamma-rays: bursts -- radiative transfer -- ISM: dust
7
0
0
2
r
a
M
7
1
v
8
5
1
3
0
7
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
1.
INTRODUCTION
Dust echoes are produced when photons emitted from a source (nova, supernova, quasar,
gamma-ray burst, etc) interact with the ambient dust grains, which either scatter or ther-
mally re-emit the photons and divert some of them towards the observer. These photons
may be observed long after the source has faded, thus providing a way of studying both the
1JILA, University of Colorado, 440 UCB, Boulder, CO 80309-0440; [email protected]
-- 2 --
nature of the source and its surroundings. A number of papers have been written on dust
echoes, including those by Dwek (1983), Chevalier (1986), Milgrom (1987), Schaefer (1987a,
b), Emmering & Chevalier (1989), Sparks (1994), Xu, Crotts & Kunkel (1994, 1995), Suger-
man (2003) and Patat (2005, hereafter P05), following pioneering work by Couderc (1939).
All of these papers deal with the dust echoes -- also called "light echoes" -- produced by
static dust distributions illuminated by a relatively weak flux of radiation. In gamma-ray
bursts (GRBs), the radiation flux is so large that dust can be destroyed out to ∼ 10 pc
by ultraviolet (UV) and X-ray heating, ion field emission (IFE; Waxman & Draine 2000,
hereafter WD00) or Coulomb explosions (Fruchter, Krolik & Rhoads 2001).
The modeling of dust echoes in GRBs is therefore complicated by the need to perform
a time-dependent treatment of the dust, where temperatures and size distributions evolve
as a function of time and distance from the source. Dust grains are selectively destroyed
according to their size and chemical composition, making it impossible to define a destruction
radius uniquely. Previous work either relied on neglecting the dust evolution or assumed a
destruction radius within which all of the dust is destroyed and beyond which it is unaffected.
These include the work of WD00, Esin & Blandford (2000, hereafter EB00), Reichart (2001,
hereafter R01), Venemans & Blain (2001) and Moran & Reichart (2005, hereafter MR05).
The motivation for considering GRB dust echoes originates from researchers wanting to
provide an alternative interpretation of the evidence for a GRB/supernova connection (e.g.,
WD00; EB00; R01) and to explain color fluctuations in afterglow light curves (MR05). With
the discovery of the association between long GRBs and the collapse of massive stars (Hjorth
et al. 2003; Stanek et al. 2003), the alternative explanation for supernova (SN) bumps is no
longer necessary. However, since massive stars have relatively short lifetimes, the majority
of them are expected to die within the dense and dusty molecular clouds in which they were
created. Dense environments, with conditions comparable to those in molecular cloud cores,
have indeed been inferred in a few cases (Lazzati & Perna 2002; Frontera et al. 2004; Frail
et al. 2006). The study of dust echoes from GRBs then becomes motivated by the fact
that they provide a three-dimensional map of the geometry of the emitting region. For an
isotropic dust distribution, this in turn translates into a geometric mapping of the source of
radiation, constituting an independent way of constraining the degree of beaming of GRB
jets.
In this paper, we present significant improvements to existing models of dust echoes by
combining time-dependent echo calculations, which are able to account for off-axis, bipolar
GRB beaming, with a code that follows the evolution of the dust grain population. The pho-
toionization and dust destruction code, written by Perna & Lazzati (2002, hereafter PL02;
see also Perna, Lazzati & Fiore 2003 and Lazzati & Perna 2003), models the destruction of
-- 3 --
dust grains due to UV and X-ray sublimation, as well as IFE. Evolution of the (graphite and
silicate) grains, hydrogen and the 12 most abundant metals are self-consistently combined.
The mono-dimensional output of the dust code is subsequently fed to a three-dimensional
echo code that computes the contribution of the dust populations to the thermal and scat-
tered echoes. In §2, we discuss the dust echo geometry and GRB parameters adopted, and
describe our use of the PL02 code. We present our results in §3 and discuss their implications
in §4.
2. METHODS
2.1. ECHO GEOMETRY
We adopt the dust echo geometry of Sugerman (2003) and consider a coordinate system
in which a thin, planar sheet of dust, of thickness dz, is located a distance z away from the
source (Fig. 1). From the point of view of the observer, an echo of projected radius ρ and
width dρ is seen. Dust grains, each located a distance r from the source, scatter the light
onto parabolic surfaces (Couderc 1939):
z =
ρ2
2ct
−
ct
2
.
(1)
Suppose an isotropic burst of energy (per unit frequency), Eν,input = Eν,input(ν, t), is
released by the source at a time t0, over some (global) time interval δt = t − t0 and at a
frequency ν, such that Lν,input = Eν,input/δt. We first calculate the integrated, monochromatic
luminosity, Lν,sc = Lν,sc(ν, t), due to scattering, at some time t after the burst. If a denotes
the grain size and n = n(a, r, t) is the dust density,
Lν,sc =Z ZV
Lν,input a2 Qsc Φ
4 r2
dn
da
δt
∆t
dV da,
(2)
where Qsc = Qsc(a, ν) is the scattering efficiency and dV = 2πρ dρ dz is the echo volume
element in the absence of beaming. The local time interval, ∆t, is the time lag between
successive echo parabolic surfaces. In practice, we follow the evolution of the dust grains
in discrete bins of ∆a (see §2.4). The Henyey & Greenstein (1941, hereafter HG41) phase
function takes the form
(3)
1 − g2
Φ =
(1 + g2 − 2gµ)3/2 ,
with g = g(a, ν) characterizing the degree of forward scattering for a given grain. The
quantity µ = z/r is the cosine of the scattering angle, θ. In almost all of the work cited, the
-- 4 --
HG41 phase function is adopted. P05 notes that more sophisticated calculations, such as
a full Mie treatment, show that the HG41 formulation tends to underestimate the forward
scattering, but the approximation is reasonably good.
For thermal re-emission, the integrated, monochromatic luminosity is
Lν,th =Z ZV
4π2 a2 Bν Qabs
dn
da
δt
∆t
dV da,
(4)
where Qabs = Qabs(a, ν) is the absorption efficiency and Bν = Bν(T (a, r, t), ν) is the Planck
or blackbody intensity. Collectively, the monochromatic light curve observed is given by
Lν = Lν,sc + Lν,th.
(5)
Though we use a planar slab for the purpose of illustration, our formalism can be applied
to any geometry. Analytical formulae for the simple cases of a slab and a spherical shell are
given in, for example, P05. Tabulated values for g, Qsc and Qabs are provided by Draine &
Lee (1984) and Laor & Draine (1993).
In our case, we consider a bipolar jet of opening half-angle θjet, positioned at a viewing
angle ψ between the jet axis and the line of sight (Fig. 2). The cosine of the occurrence
angle, θ′, is given by
cos θ′ = cos θ cos ψ + sin θ sin ψ cos φ.
(6)
The luminosity contribution from a given dust grain is only included when either the con-
dition θ′ < θjet or θ′ > π − θjet is satisfied. The echo volume element is then generalized to
dV = ρ dφ dρ dz and there is an additional integration over the azimuthal angle, φ. Again,
in practice, one sums over discrete intervals or bins of ∆ρ, ∆φ, ∆z and ∆a.
2.2. GRB LIGHT CURVE MODELING
The GRB light curve has three components: the prompt emission, an optical flash and
the afterglow (Fig. 3). We model the prompt emission as
Lν,pm(ν, t) =
1 + (ν/6 × 1019 Hz)2.5 (exp (−3 + t/1 s)
exp (0.3 − t/10 s)
L0
, t ≤ 3 s
, t ≥ 3 s.
(7)
The quantity L0 = 1.16 × 1032 erg s−1 Hz−1 is chosen such that R R Lν,pm dν dt ≈ 1053 erg.
-- 5 --
We allow for the presence of an optical flash, as observed in GRB 990123 (Akerlof et al.
1999), modeled through the function (e.g., Draine & Hao 2002)
Lν,of = 3 × 1033 erg s−1 Hz−1
where νB = 8 × 1017 Hz.
(t/30 s)3/2
1 + (t/30 s)7/2 ((ν/4 × 1014 Hz)−0.65
100−0.65 exp (ν/νB − 1)
, ν < νB
, ν ≥ νB
(8)
For the afterglow, we use (Panaitescu & Kumar 2000)
Lν,ag = 3 × 1033 erg s−1 Hz−1h(t/30 s)−0.5 + (t/30 s)1.125/6i−6 (cid:18)
ν
4 × 1014 Hz(cid:19)−1.15
.
(9)
It follows that the input luminosity (per unit frequency) from the GRB is
Lν,input = Lν,pm + Lν,of + Lν,ag.
(10)
Figure 3 shows the R band light curve according to our model for a GRB at z = 1. If the
GRB ejecta are beamed, then the afterglow eventually evolves to Lν,ag ∝ t−p, where p = 2.5
is the index of the electron energy distribution, F (ǫ) ∝ ǫ−p, at a time
tbreak(θjet) =(cid:18)
3E
4πc5mH n(cid:19)1/3
θ8/3
jet ,
(11)
where E = 1053 erg and n is the number density of the ambient gas.
In the case of an
off-axis jet, we adopt the following approximation: at tbreak = tbreak(ψ) (Rossi, Lazzati &
Rees 2002; Zhang & M´esz´aros 2002), Lν,ag transitions from ∝ t2 to ∝ t−p; we ignore the ∝ t3
and plateau behaviors just before tbreak.
2.3. DUST MODELING
As discussed in §1, the massive-star progenitors of GRBs -- probably of the Wolf-Rayet
type -- are likely to die close to the locations where they were born, in dense, molecular
cloud cores. While we compute the peaks of the dust echo emission for different cloud
configurations, we focus most of the calculations on a typical example of such a cloud;
specifically, one with radius R = 3 pc and hydrogen column density NH = 1022 cm−2,
corresponding to nH ∼ 103 cm−3 or, equivalently, a gas mass of ∼ 1000 M⊙. The GRB is
assumed to be located at the center of the cloud. Since late-type WC stars are themselves
known to be dust-making machines (e.g., Williams, van der Hucht & The 1987), a further
concentration of dust is expected in the vicinity of the star. Observations by Williams, van
-- 6 --
der Hucht & The (1987) indicate these distances to be ∼ 1014 to 1015 cm. MR05 considered
specifically the emission from this proximate dust layer, showing that it yields a peak emission
within a few minutes after the burst. We consider a uniform distribution of dust within the
molecular cloud, with a solar dust-to-gas mass ratio, f⊙ = 0.01.
A variety of grain models exist and the interested reader is referred to, for example, the
review by Draine (2004). In general, they incorporate a combination of silicates, graphite
and polycyclic aromatic hydrocarbons (PAHs). Broadly speaking, the presence of graphite
grains may account for the 2175 A bump in the extinction curve. Steep UV extinction may
be due to the presence of silicate grains. PAHs account for several unique emission features,
as well as for far ultraviolet extinction. Various types of composite grains may also provide
the bulk of the infrared and optical extinction. We specialize to Galactic-type dust, which is
believed to be a mixture of graphite and silicate grains -- with the latter having an average
composition of MgFeSiO4, also known as "olivine"; see Draine & Lee (1984) -- following
Mathis, Rumpl & Nordsieck (1977). We assume an initial, power-law distribution
dn
da
∝ a−β,
(12)
where β = 3.5. While this value of β is typical of the dust found in our Galaxy, it might
not accurately represent the grain distributions in denser environments. Nevertheless, it is
a good first approximation. As the burst radiation propagates through the medium, the
grain population evolves and β either changes or ceases to be meaningful as portions of the
distribution may be completely destroyed. Dust echoes probe the distribution of dust while
it is being modified by the burst radiation. We properly compute the temporal evolution of
the dust distribution, as described in the succeeding sub-section.
2.4. DUST EVOLUTION
As mentioned in §1, we follow the evolution of dust as a function of position within the
cloud by using the dust code of PL02, which treats metals (He, C, N, O, Ne, Mg, Si, S, Ar,
Ca, Fe, Ni), hydrogen (in all of its atomic and molecular forms) and dust evolution in a self-
consistent manner. For dust destruction, UV and X-ray sublimation and IFE are considered.
Sublimation occurs when grains absorb energy faster than they manage to radiate away and
the excess energy goes into breaking the bonds that hold the atoms to the surface of the grain.
Small and large grains are heated predominantly by UV radiation and X-rays, respectively.
IFE occurs when a grain is charged up to a potential of Vmax = 3a−5(S/1011 dyn cm−2)1/2
kV, where a = a−510−5 cm and S is the tensile strength of the grain material, and any further
ionization leads to the emission of ions in order to avoid an increase in the surface electric
-- 7 --
field. The process of Coulomb explosion, which is the fragmentation of the grain into smaller
components, competes with that of IFE. The relative importance of the two processes is a
subject of controversy. PL02 adopt the approach of Draine & Hao (2002) and assume that
Coulomb explosion does not play a major role, due to the possibility of chemical bonds
being promptly re-established in the warm grain -- annealing. Grains close to the source
are sublimated, while those far away from it are eroded by IFE.
As dust destruction fronts propagate, dust-depleted metals are recycled into gas. As
such, their abundances in the gaseous and the dust-depleted phases are book-kept separately.
Computationally, at each time step t, the code calculates dn/da and T as a function of both
r and a [Fig. 4; see also Perna, Lazzati & Fiore (2003) for further examples of dust evolution
for various initial values of β]. In this way, the dust populations are evolved both spatially
and temporally. We see from Fig. 4 that graphite grains are more resistant to the impinging
radiation; larger grains tend to survive better as well.
2.5. CLOUD OPACITY
Neither the dust code of PL02 nor our echo code deals with the multiple-scattering
effects of radiative transfer in an optically-thick cloud. There are three regimes of opacities,
namely clouds that are: optically thin at the beginning of the burst; optically thick at the
end of the burst and afterglow; initially optically thick but become optically thin after part
of the dust is destroyed by the burst and afterglow radiation. Certainly, this definition is
dependent on wavelength.
Clouds that are initially optically thin to radiation are not a concern, since multiple
scattering and absorption/re-emission are negligible. However, a cloud can be optically
thin in the infrared but thick in the UV; in principle, UV radiation can be "downgraded"
to the infrared by multiple scattering and absorption/re-emission events, thus contributing
to the infrared lightcurve. However, this effect is small for two reasons. Firstly, the dust
temperatures at long time scales are low, due to the rapid cooling of the grains. UV radiation
will be reprocessed in the far infrared, at wavelengths ∼ 10 µm, much longer than those
considered below. Secondly, although multiple scattering will introduce a time dilution effect
to the light curve (by delaying the photons and causing them to be detected at later times), we
expect it to be small. The ratio of the contributions from the multiply-scattered component
to the singly-scattered one is ∼ τ exp (−τ ); for large optical depths (τ ≫ 1), extinction
dominates over time dilution, making the multiply-scattered component negligible.
To understand how to deal with initially opaque clouds, we must examine in detail the
-- 8 --
process of dust sublimation and the relative time scales. The former (due to heating) is
driven by the burst flux. Therefore it takes place during the GRB prompt phase, which lasts
∼ 10 s. The slower IFE process, which is driven by the burst fluence, lasts for several hours
up to a maximum of a few days. The dust destruction process takes place within a thin shell
of thickness δR . 1015 cm, a tiny fraction of the cloud radius. Therefore, to a very good
approximation, the dust destruction front propagates at the speed of light. Consider now a
photon that is re-directed towards some line of sight. Due to the time delay in its propagation,
it will be lagging behind the front. As a consequence, it will propagate only through the
post-burst dust population. To compute the number of photons, a time-dependent dust
treatment is required, but only the final configuration of the cloud is neccessary to evaluate
the extinction of the thermally re-emitted and scattered light.
This allows us to develop a very simple recipe to deal with isotropic bursts. Regardless
of the initial opacity in the band considered, we can compute the echo light curve ignoring
the opacity and eventually correct a posteriori with the residual extinction (if any). A
more complex treatment is necessary for beamed GRBs. As shown in Fig. 5, a beamed
GRB cleans out only a biconical region of the cloud and, for any initially-opaque cloud, a
proper treatment of the opacity has to be included. A simple one is not possible in this
case since, as shown in the figure, different amounts of dust lie along different lines of sight.
The treatment becomes even more complicated for off-axis observers, due to the loss of the
cylindrical symmetry. For this reason we restrict our treatment of beamed GRB echoes to
initially optically-thin clouds.
3. RESULTS
3.1. LIGHT CURVES & SPECTRA
Dust echoes have a duration techo ∼ 2R/c and the energy involved is Eecho ∼ ζE. The
"echo efficiency factor", ζ, depends on the total amount of dust present in the scattering and
thermally re-emitting region, the distance of the dust from the burst, the dust composition
and the GRB spectrum. The dependence of ζ on the GRB and cloud parameters is complex.
A large value of ζ can be achieved if the dust grains are sufficiently far away to survive fast
sublimation by the burst photons, yet close enough to be heated to high temperatures and
provide a sizable optical depth. Dust grains in the path of the optical-UV flash are expected
to be sublimated out to a distance ∼ 1 pc (WD00). Using the PL02 dust code described
in §2.4, we follow the evolution of dust grains, with sizes 0.3 nm . a . 0.3 µm, for various
values of the density and radius of the absorbing cloud.
-- 9 --
Figure 6 shows how the wavelength of the peak emission, for the thermal and scattered
components, varies as a function of the cloud parameters. The general trend is for it to
increase with cloud size (Fig. 6); since the temperature of the dust declines with distance
from the radiation source, larger clouds have higher fractions of cooler dust. On the other
hand, for a fixed radius, the peak of the emission is density-independent as long as the cloud
is optically thin (for NH = 1021 and 1022 cm−3), but it shifts to longer wavelengths when the
cloud becomes optically thick (NH = 1023 cm−3). This is due to the fact that if the cloud is
optically thick, less radiation propagates to the outer layers of dust, and therefore the grains
attain a lower temperature.
For the remainder of the paper, unless otherwise noted, we will concentrate on the
(optically thin) case of a cloud with radius R = 3 pc and NH = 1022 cm−3. Figure 7 shows
echo light curves for the source/rest wavelength λ0 = 3.55 µm -- we will justify our choice
shortly -- for various values of the jet opening half-angle; the line of sight is coincident with
the axes of symmetric, bipolar jets. In all of the cases, the GRB isotropic equivalent energy
(in photons) is E = 1053 erg. For beamed emission, a trough appears in the light curve at
t ∼ R(1 − cos θjet)/c. As θjet decreases, the width of the trough increases. The figures also
show the afterglow light curves for comparison. For this particular configuration, dust echo
bumps are visible with a brightness well above the afterglow, several weeks to months after
the GRB.
For λ0 = 3.55 µm and θjet = 22.5◦, we compute light curves for different values of the
viewing angle (Fig. 8). Even though this value of θjet is hardly typical of most GRBs, we
adopt it as an illustration and as a matter of computational expediency. The main effect of
an increasing ψ is to make the trough shallower, while lowering the light curve for t . R/c.
Eventually, for ψ = 90◦, a bump emerges in place of the trough. These behaviors can be
understood in the following manner: at ψ = 0◦, the echo surface at ∼ R/c samples the
least illuminated dust; when this configuration is rotated by 90◦, it now instead samples the
most dust. Note that for Figs. 7 and 8, we apply a smoothing and interpolation procedure
in places where numerical noise is clearly present, taking care not to alter the qualitative
behavior of the light curves.
For a GRB observed from its onset until a time tobs, one can measure the echo energy
(per unit frequency),
We compute light curves for the range of rest frame wavelengths 1 nm ≤ λ0 ≤ 1 mm,
for an isotropic fireball.
Integrating these light curves till t = tobs yields the spectrum,
In the infrared, the echo is dominated
which we obtain for two values of tobs (Fig. 9).
Eν =Z tobs
0
Lν dt.
(13)
-- 10 --
by thermal re-emission from graphite grains. Since re-emission is isotropic, the spectrum
hardly changes with time. In the soft X-ray, dust grains preferentially scatter light forwards,
causing the spectrum to be highly variable, being stronger at early and weaker at late times.
By separating the dust echoes into thermal and scattered components (Fig. 9), we see that
they peak at about λpeak = 3.55 µm (infrared) and 8.80 eV (extreme ultraviolet/soft X-
ray), respectively. The former value justifies our earlier choice of λ0 = λpeak in Figs. 7
and 8. Various studies have yielded different predictions for λpeak: 2 µm (EB00), K band
(R01) and 8 µm (Venemans & Blain 2001). The fact that these values differ is hardly
surprising, as λpeak depends sensitively on the adopted dust composition, the range of grain
sizes and temperatures, as well as the cloud configuration. It is worthwhile to note that these
calculations are for essentially static grain populations. By allowing the dust distribution to
evolve, we include contributions from hotter grains which survive only for a very short time
in the interstellar medium (Fig. 4). We also apply an extinction correction, exp (−τ ), due
to the presence of a thin, post-burst dust layer, to our spectra; it is clear that this correction
is negligible at λ0 = λpeak.
A striking observational feature of dust echoes from GRBs with symmetric, bipolar jets
is the presence of two bumps, each corresponding to emission from the region illuminated
by one of the two jets. This is an important difference from SN bumps that are expected to
contribute only a single rebrightening. Furthermore, the temporal separation between the
two peaks allows one to directly infer the opening half-angle, since the first bump appears
at t1 ∼ R(1 − cos θjet)/c while the second emerges at t2 ∼ 2R/c. For a narrow jet, we have
θjet ∼ 2r t1
t2
.
(14)
If the GRB does not sit in the center of the cloud, equation (14) becomes more complicated
and we then need to use the relative brightnesses of the peaks to constrain the burst location.
3.2. OBSERVATIONAL PROSPECTS
For a region of size ∼ 3 pc and density nH ∼ 103 cm−3, the luminosity of the dust echoes
is ∼ 1027 erg s−1 Hz−1 at a wavelength of 3.55 µm, allowing them to sit above the afterglow
light curve for time scales of about a month to over a year.
In Figs. 7 and 8, we display additional axes for the M magnitude of the echoes. At λ0
= 3.55 µm, we need the source to be located at z ≈ 0.4 in order to detect its light in the
M band (5 µm). For on-axis GRB jets, the echo bumps have a magnitude of M ∼ 26 − 27,
straining the detection limits of ground-based telescopes, but well within the reach of IRAC
-- 11 --
onboard the Spitzer Space Telescope. Specifically, for a cloud of R = 3 pc and NH = 1022
cm−2, infrared echoes should be detectable with IRAC at 3.6 and 4.5 µm up to z ∼ 0.2 (Fig.
10) and with NICMOS (in the J band) onboard the Hubble Space Telescope up to z ∼ 0.3.
In the optical, the echoes are less bright, with expected magnitudes of Rb ∼ 30 and
V b ∼ 30. Figure 10 shows minimum flux thresholds for the Very Large Telescope (VLT) and
demonstrates that optical echoes can be seen up to z ∼ 0.1. Finally, in the X-ray, the echo
flux is dominated by scattering -- in the 0.1 to 1 keV band, we have FX ∼ 10−16 erg cm−2
s−1 at z ∼ 0.1. Such a low flux is barely detectable with a long Chandra exposure, making
the X-ray band an unattractive prospect in the search for dust echoes from the circumburst
medium.
Extrapolating these results to different dust geometries is not straightforward. Naively,
the dust luminosity scales as
Lecho ∼ ζ
cE
2R
,
(15)
where ζ hides the complexity of the time-dependent dust distribution. Equation (15) gives
an optimistic scaling and should be used with caution. On one hand, ζ will be low for a very
large cloud, since the grains will not reach very high temperatures. However, a very small
region will also have a low ζ due to the fact that all of the dust is sublimated in the very first
few fractions of a second by the burst radiation. Increasing the amount of dust increases ζ,
but at the expense of causing residual absorption by dust in the cloud, delaying the echoes
and thereby lowering their luminosity.
4. DISCUSSION
We have computed dust echoes due to scattering and thermal re-emission of dust by
GRB sources. Our model allows for time-dependent dust destruction by both isotropic and
beamed radiation from the GRB, as well as for any inclination of the jet axis with respect
to the line of sight to the observer. Our source model assumes that the outer edges of the
jets are sharp. Under these conditions and the assumption of symmetric, bipolar jets, a pair
of bumps is expected in the light curve, an intrinsically different behavior from the possible
presence of a single SN bump. This feature should qualitatively remain even if the jet is
structured (Rossi, Lazzati & Rees 2002; Zhang & M´esz´aros 2002). The energy input in this
case is a function of the angle from the jet axis -- namely, ǫ ∝ Θ−k, where we typically
have 1.5 < k . 2. [However, we note that the numerical simulations of Morsony, Lazzati &
Begelman (2006) allow for k ≫ 1.] Due to the energy dependence with angle from the jet
axis, dust is destroyed out to larger radii along the jet axis than in its wings. The range
-- 12 --
of grain temperatures present is therefore expected to vary with azimuthal angle. The light
travel paths of the photons differ, due to different geometries created by the destruction
fronts. Thus, one expects varying observational features depending on the structure of the
GRB jet.
Qualitatively, one can think of a structured jet as a superposition of a number of non-
structured jets with different θjet. The component with the smallest opening half-angle, θcore,
forms the core of the structured jet, produces the brightest echo and displays a trough in its
light curve at the earliest time. The net effect from summing all of the components is to form
an echo light curve with a gentle decline at tcore ∼ R(1 − cos θcore)/c, rather than a sharp
transition to a trough. As a consequence, the first echo bump will likely be undetectable,
since it lies at very early times and is outshone by the afterglow.
Observationally, we conclude that even echoes from moderate redshift events are hard to
detect with present ground-based telescopes. Infrared observations with Spitzer and Hubble
-- and in the future with the James Webb Space Telescope -- will detect echo components
in GRBs out to z ∼ 0.3. An intriguing prospect is the detection of orphan dust echoes,
which are those from a misaligned jet. They are almost as bright as echoes from an on-axis
GRB, but are not contaminated by the burst and afterglow radiation. Detecting such echoes
in the optical light curves of supernovae may serve as the observational signature of off-
axis, "phantom" GRBs (and their accompanying "orphan afterglows"), which are otherwise
exceedingly difficult to detect.
We wish to thank: Peter Ruprecht and James McKown for steadfast and unwaivering
technical support at JILA; Brian Morsony for illuminating discussions concerning structured
jets; Enrico Ramirez-Ruiz for useful conversations regarding orphan afterglows; Richard
McCray and Svetozar Zhekov for helpful comments and suggestions, following their careful
reading of the manuscript. This work was partially supported by NASA Grants NAG5-12279
(K.H.) and NAS5-26555 (K.H.), and NSF Grant AST 0507571 (K.H., D.L. and R.P.).
REFERENCES
Akerlof, C., et al. 1999, Nature, 398, 400
Chevalier, R.A. 1986, ApJ, 308, 225
Couderc, P. 1939, Ann. d'Ap., 2, 271
Dwek, E. 1983, ApJ, 274, 175
Draine, B.T., & Lee, H.M. 1984, ApJ, 285, 89
-- 13 --
Draine, B.T., & Hao, L. 2002, ApJ, 569, 780
Draine, B.T. 2004, Astrophysics of Dust, ASP Conference Series, 309, ed. A.N. Witt, G.C.
Clayton, & B.T. Draine, 691
Emmering, R.T., & Chevalier, R.A., 1989, ApJ, 338, 388
Esin, A.A., & Blandford, R. 2000, ApJ, 534, L151
Frail, D.A., et al. 2006, ApJ, 646, L99
Frontera, F, et al. 2004, ApJ, 614, 301
Fruchter, A., Krolik, J.H., & Rhoads, J.E. 2001, ApJ, 563, 597
Henyey, L.C., Greenstein, J.L. 1941, AJ, 93, 70
Hjorth, J. et al. 2003, Nature, 423, 847
Laor, A., & Draine, B.T. 1993, ApJ, 402, 441
Lazzati, D., & Perna, R. 2002, MNRAS, 330, 583
Lazzati, D., & Perna, R. 2003, MNRAS, 340, 694
Mathis, J.S., Rumpl, W., & Nordsieck, K.H. 1977, ApJ, 217, 425
Milgrom, M. 1987, A&A, 182, L21
Moran, J.A., & Reichart, D.E. 2005, ApJ, 632, 438
Morsony, B.J., Lazzati, D., & Begelman, M.C. 2006, ApJ, in press (astro-ph/0609254)
Panaitescu, A., & Kumar, P. 2000, ApJ, 543, 66
Patat, F. 2005, MNRAS, 357, 1161
Perna, R., & Lazzati, D. 2002, ApJ, 580, 261
Perna, R., Lazzati, D., & Fiore, F. 2003, ApJ, 585, 775
Reichart, D.E. 2001, ApJ, 554, 643
Rossi, E., Lazzati, D., & Rees, M.J. 2002, MNRAS, 332, 945
Schaefer, B. 1987, ApJ, 323, L47
Schaefer, B. 1987, ApJ, 323, L51
Sparks, W.B. 1994, ApJ, 433, 29
Stanek, K. et al. 2003, ApJ, 591, L17
Sugerman, B.E.K. 2003, AJ, 123, 1939
Waxman, E., & Draine, B.T. 2000, ApJ, 537, 796
-- 14 --
Williams, P. M., van der Hucht, K. A. & The, P. S. 1987, A&A, 182, 91
Venemans, B.P., & Blain, A.W. 2001, MNRAS, 325, 1477
Xu, J., Crotts, A.P.S., & Kunkel, W.E. 1994, ApJ, 435, 274
Xu, J., Crotts, A.P.S., & Kunkel, W.E. 1995, ApJ, 451, 806
Zhang, B., & M´esz´aros, P. 2002, ApJ, 571, 876
This preprint was prepared with the AAS LATEX macros v5.2.
-- 15 --
observer
z
ρ
∆ρ
r
θ
source
∆z
Fig. 1. -- Dust echo geometry, adopted from Sugerman (2003).
-- 16 --
x
dust grain
z
θ'
θ
ψ
y
bipolar jet
jet axis
Fig. 2. -- Geometry for GRB beaming. A bipolar jet of opening half-angle θjet (not displayed
in the diagram) is positioned at a viewing angle of ψ between the jet axis and the light of
sight (z-axis). While the jet axis is located in the x-z plane, the dust grain shown is generally
not found in the same plane; θ is the scattering angle. Knowledge of the occurrence angle,
θ′, allows one to determine if the light curve contribution from a given dust grain should be
included (see text). Also not shown is the angle φ, which is the azimuthal angle about the
line of sight.
-- 17 --
Fig. 3. -- R band light curve according to our GRB model for an event at z = 1. The
solid line with dots shows the total light curve, while dotted, dashed and dot-dashed lines
represent the prompt, optical flash, and afterglow emission, respectively.
-- 18 --
Fig. 4. -- Input number densities, n, and temperatures, T , for two representative grain sizes
and at different times since the GRB, generated using the dust code of Perna and Lazzati
(2002). Shown are graphite (top) and silicate (bottom) grains.
-- 19 --
Fig. 5. -- Cartoon showing the final configuration of a cloud illuminated by a beamed GRB
-- three regions can be identified in the cloud. Region (a) has all of the dust completely
destroyed by the burst, yielding no residual opacity. Region (b) is outside the beaming cone
and holds the unmodified dust populations. Region (c) is a layer of dust left on the outskirt
of the cloud; it is not a general outcome of the simulations as region (a) extends all the
way to the edge for more compact clouds. Dashed lines show several paths of re-directed
photons from the illuminated dust towards an on-axis observer. The various paths range
from almost clear (1a) to largely obscured ones (3a). Dot-dashed lines show the paths to an
observer located outside the cone -- for such paths, relatively more extinction is generally
experienced (cfr. 123a with 123b).
-- 20 --
Fig. 6. -- Thermal (top) and scattered (bottom) peak wavelengths as a function of the cloud
radius R for hydrogen column densities NH = 1022 and 1023 cm−2. Note that these are
intrinsic wavelengths for which no extinction correction has been applied to the spectra.
-- 21 --
Fig. 7. -- Dust echo light curves for on-axis GRBs. Isotropic and beamed (θjet = 10◦, 22.5◦
and 45◦) emission are both considered. The corresponding afterglows are plotted for com-
parison and 2R/c ∼ 2400 days.
-- 22 --
Fig. 8. -- Top: Dust echo light curves for jet opening half-angle θjet = 22.5◦ and various
values of the viewing angle, ψ. Bottom: Light curves with different ψ plotted against the
corresponding afterglows.
-- 23 --
Fig. 9. -- Dust echo spectrum integrated from t = 0 to tobs. As an illustration, the insert
shows the separated scattered (denoted by "S") and thermal (denoted by "T") components
for tobs = 1 month. Also shown are the spectra with the exp (−τ ) extinction correction
applied.
-- 24 --
Fig. 10. -- Time-averaged, observed echo flux, from isotropic fireballs, as a function of red-
shift, z, with the extinction correction applied. Shown are two NH = 1022 cm−2 cloud models
for comparison, with the thin and thick lines representing R = 1 and 3 pc, respectively. We
|
astro-ph/0006028 | 1 | 0006 | 2000-06-01T22:44:02 | Crust-core coupling and r-mode damping in neutron stars: a toy model | [
"astro-ph",
"gr-qc"
] | R-modes in neutron stars with crusts are damped by viscous friction at the crust-core boundary. The magnitude of this damping, evaluated by Bildsten and Ushomirsky (BU) under the assumption of a perfectly rigid crust, sets the maximum spin frequency for a neutron star spun up by accretion in a Low-Mass X-ray binary (LMXB). In this paper we explore the mechanical coupling between the core r-modes and the elastic crust, using a toy model of a constant density neutron star with a constant shear modulus crust. We find that, at spin frequencies in excess of ~50 Hz, the r-modes strongly penetrate the crust. This reduces the relative motion (slippage) between the crust and the core compared to the rigid crust limit. We therefore revise down, by as much as a factor of 10^2-10^3, the damping rate computed by BU, significantly reducing the maximal possible spin frequency of neutron star with a solid crust. The dependence of the crust-core slippage on the spin frequency is complicated, and is very sensitive to the physical thickness of the crust. If the crust is sufficiently thick, the curve of the critical spin frequency for the onset of the r-mode instability becomes multi-valued for some temperatures; this is related to the avoided crossings between the r-mode and the higher-order torsional modes in the crust. The critical frequencies are comparable to the observed spins of neutron stars in LMXBs and millisecond pulsars. | astro-ph | astro-ph |
Mon. Not. R. Astron. Soc. 000, 000 -- 000 (0000)
Printed 29 October 2018
(MN LaTEX style file v1.4)
Crust-core coupling and r-mode damping in neutron
stars: a toy model
Yuri Levin1 and Greg Ushomirsky2
1Astronomy Department, 601 Campbell Hall, University of California, Berkeley, California, 94720;
[email protected]
2Theoretical Astrophysics, M/C 130-33, Caltech, Pasadena, California, 91125;
[email protected]
June 1, 2000
ABSTRACT
R-modes in neutron stars with crusts are damped by viscous friction at the crust-core
boundary. The magnitude of this damping, evaluated by Bildsten and Ushomirsky
(BU) under the assumption of a perfectly rigid crust, sets the maximum spin frequency
for a neutron star spun up by accretion in a Low-Mass X-ray binary (LMXB). In this
paper we explore the mechanical coupling between the core r-modes and the elastic
crust, using a toy model of a constant density neutron star with a constant shear
modulus crust. We find that, at spin frequencies in excess of ≈ 50 Hz, the r-modes
strongly penetrate the crust. This reduces the relative motion (slippage) between the
crust and the core compared to the rigid crust limit. We therefore revise down, by
as much as a factor of 102
− 103, the damping rate computed by BU, significantly
reducing the maximal possible spin frequency of neutron star with a solid crust. The
dependence of the crust-core slippage on the spin frequency is complicated, and is
very sensitive to the physical thickness of the crust. If the crust is sufficiently thick,
the curve of the critical spin frequency for the onset of the r-mode instability becomes
multi-valued for some temperatures; this is related to the avoided crossings between
the r-mode and the higher-order torsional modes in the crust. The critical frequencies
are comparable to the observed spins of neutron stars in LMXBs and millisecond
pulsars.
1
INTRODUCTION AND BASIC ARGUMENT
Gravitational radiation offers an exciting possibility to set
an upper limit on the spin frequencies of accretionally
spun-up neutron stars in close binaries (Bildsten 1998).
The source for gravitational radiation may be the mass
quadrupole supported by static stresses within the neutron
star crust (Bildsten 1998, Ushomirsky, Cutler, & Bildsten
2000), or the current quadrupole from an l = m = 2 r-mode
which is driven by gravitational radiation reaction (Ander-
sson et al. 1999, Bildsten 1998; for a review of the r-mode
instability see, e.g., Friedman & Lockitch 1999; for the dis-
covery paper see Andersson 1998). In this paper we concen-
trate on the second option, namely that accretional spin-up
triggers the r-mode instability, and the resulting spin-down
torque from gravitational radiation reaction balances or ex-
ceeds the accretion torque. The instability is triggered when
the r-mode growth rate due to gravitational radiation re-
action (which is a sharply increasing function of the stellar
spin frequency) becomes equal to the r-mode viscous damp-
ing rate.
The viscous damping timescale for r-modes was first
computed by Lindblom, Owen, & Morsink (1998). They
assumed that a neutron star is a fluid ball with a poly-
tropic equation of state, and that the viscosity is set by
c(cid:13) 0000 RAS
neutron-neutron scattering in the nuclear matter. Their vis-
cous damping implied that, if the neutron star core temper-
ature T ∼> 108 K, which is the case for the rapidly accreting
neutron stars in LMXBs (Brown & Bildsten 1998; Brown
2000), then the r-mode instability would be triggered at a
stellar spin frequency just above 100 Hz. This is inconsis-
tent with observations: most neutron stars in LMXBs are
believed to be spinning with frequencies around 300 Hz (see
van der Klis 1999 for a review), and the existence of two
1.6 ms radio pulsars means that rapidly rotating neutron
stars are somehow formed in spite of the r-mode instability
(Andersson et al. 1999). The r-mode damping rate would
have to be ∼ 250 higher than that computed by Lindblom
et al. (1998) to allow for such high spins.
Lindblom et al. (1998) conjectured that, for the range of
temperatures relevant for LMXBs, mutual friction between
the neutron and proton superfluids will provide the missing
damping. However, tour-de-force calculations of Lindblom
& Mendell (1999) showed that, except for ∼ 2% of the al-
lowed parameter space, mutual friction gives only a modest
increase of the r-mode damping rate, insufficient to account
for the observed high spins of neutron stars.
Bildsten & Ushomirsky (2000) (hereafter BU) argued
that the crust plays an important role in r-mode damping.
2
Levin & Ushomirsky
In their picture, the r-mode lives in the liquid core, and most
of the dissipation occurs in a thin (a few cm) viscous bound-
ary layer between the oscillating core and the motionless (in
the rotating frame) solid crust. The damping BU computed
is a factor of ∼ 105 higher than that of Lindblom et al.
(1998); it implies that the r-modes would not become un-
stable unless the spin frequency is close to break-up, which
is significantly higher than the observed LMXB spins. An-
dersson et al. (2000) and Rieutord (2000) have revised the
BU estimate, down and up respectively, by factors of order
unity. Both of these corrections are valid; the net result is
that the BU estimate is almost unchanged. However, there is
still a substantial correction unaccounted for in the previous
work. In this paper, we show that, for a considerable frac-
tion of the parameter space, the boundary layer damping
rate is a factor of 102 − 103 smaller than the BU estimate,
and hence the critical frequencies for the onset of the r-mode
instability are correspondingly lower, ∼ 200 − 600 Hz.
BU considered only the leading-order transverse motion
pattern of the r-modes and assumed that these motions do
not penetrate into the crust. They estimated the magnitude
of the viscous coupling between these transverse motions
and the crust, and found it to be small. Therefore, their
picture is that of a liquid in a bucket -- the amplitude of
the relative motion (slippage) between the crust and the
core is just equal to the amplitude of the fluid motion at
the boundary. However, small radial motions of the r-mode,
neglected by BU, are very effective at coupling the crust
and the core. Above a small critical frequency, r-modes do
strongly penetrate into the crust, and the crust and the core
oscillate almost in unison with very similar amplitudes. The
slippage δu in this regime is only a small fraction of the total
r-mode motion, u, with typical values of δu/u ∼ 0.05 − 0.1,
which results in a correction of a factor of (δu/u)2 to the
BU viscous boundary layer damping rate.
The fact that the crust is not rigid is central to our
story. In fact, the crust is more like a gel: the shear modulus
is much smaller than the bulk modulus. The l = m = 2
torsional mode in a non-rotating crust, which has the same
angular displacement pattern as the corresponding r-mode,
has a frequency fcr ≃ 50 Hz (McDermott et al. 1988). This
frequency is a few times lower than the r-mode frequency
in rapidly rotating stars, indicating that the elastic restor-
ing force is quite weak compared to the Coriolis force which
is responsible for the r-mode oscillations. Therefore, at suf-
ficiently high spin frequencies, one would expect the crust
to oscillate more or less like a liquid, with elasticity only
slightly modifying the mode structure. Indeed, if the shear
modulus µ of the crust were exactly zero, i.e., if the crust
were completely fluid, there would be no slippage δu between
the crust and the core. Since µ is non-zero but small, one
would expect the slippage to be proportional to the ratio of
the elastic restoring force to the Coriolis force, i.e., roughly
δu/u ∼ (fcr/frmode)2. However, our numerical work shows
that the slippage is not a monotonically decreasing function
of the spin frequency; it rises sharply at the frequencies of
avoided crossings between the r-mode and the higher-order
torsional modes of the crust. Figure 1 shows the slippage as
a function of the spin frequency computed for two neutron
star models with different crust thicknesses (10% and 20%
of the stellar radius).
We begin in § 2 by outlining the equations governing the
pulsations of constant density stars with fluid cores and solid
crusts and presenting the numerical results for the depen-
dence of the slippage δu/u on the spin frequency. In § 3 we
use the computed slippages to construct the critical r-mode
stability curves; see Figure 2. The resulting curves are quite
sensitive to the thickness of the neutron star crust. In par-
ticular, for a sufficiently thick crust, the stability curve has
a peculiar multi-valued structure, which is a consequence
of the non-monotonic dependence of the relative slippage
on the spin frequency. The stability curve determines the
maximum spin of a neutron star as a function of temper-
ature; we find that it generally goes through the region of
observed spin frequencies of neutron stars in LMXBs and in
millisecond radio pulsars. Broad implications of our results
are discussed in § 4.
2 CORE-CRUST BOUNDARY SLIPPAGE
2.1 Basic formalism
Oscillation modes of a rotating neutron star with an elastic
crust were analyzed by Lee & Strohmayer (1996) (hereafter
LS). Their analysis was performed for the case of slow rota-
tion, which was treated as a perturbation to other restoring
forces, and the oblateness of the equilibrium configuration
was neglected. We broadly utilize their general formalism,
however we derive equations specialized to the constant-
density case. For r-modes the correction to the eigenfre-
quency due to the non-sphericity of the equilibrium con-
figuration is of the same order as the correction due to the
radial motion (Provost et al. 1981). However, we expect that
the oblateness of the star will not change the qualitative pic-
ture for the crust-core slippage: at high enough spin frequen-
cies, the core and the crust will move almost in unison. The
quantitative effects of non-sphericity are not treated in this
paper.
For our toy model, we assume that the outer part, rcr <
r < R, of the neutron star of radius R possesses a small
shear modulus µ(r), such that µ(r) = const. The neutron
star fluid is assumed to be incompressible, has a constant
density ρ, and is rotating with angular frequency Ω.
The general displacement ~u of a fluid element, in a frame
rotating with the neutron star, is given by (Zahn 1966, LS):
(ul(r)Ylm r + vl(r)∇Ylm + wl(r)r × ∇Ylm)eiωt,
(1)
~u = Xl
where ω is the mode frequency in the rotating frame, m is
the fixed azimuthal number, and ∇Ylm is the gradient of
the spherical harmonic evaluated on a unit sphere. We are
interested in the l = m r-mode. For a purely fluid star, to
O(Ω/Ωk)2, where Ωk = (GM/R3)1/2 is the keplerian an-
gular velocity, the displacement for this mode is given by
~u = (wl r × ∇Yll + vl+1∇Yl+1 l + ul+1Yl+1 lr) eiωt,
(2)
i.e., vl+1/wl and ul+1/wl are of order (Ω/Ωk)2. In the solid
crust, the displacement vector need not satisfy the ordering
assumed in Eq. (2). However, at least for modes with ω close
to 2Ω/3, the Coriolis force is the dominant restoring force,
and we expect vl+1/wl and ul+1/wl to be small; this ordering
is confirmed by our numerical results (see § 3).
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
We now define dimensionless frequencies Ω = Ω/Ωk and
ω = ω/Ωk, as well as the dependent variables similar to those
utilized by LS:
z1 =
z2 = 2
ul+1
r
dul+1
dr
vl+1
r
dvl+1
dr
wl
r
z3 =
z4 =
z5 = i
−
vl+1
r
+
ul+1
r
z6 = i(cid:16) dwl
dr
−
wl
r (cid:17) .
(3a)
(3b)
(3c)
(3d)
(3e)
(3f)
In the constant density case, z2 is not a dynamical variable,
as it can be eliminated using the continuity equation, ∇·~u =
0, or
(4)
z2 + 2z1 − (l + 1)(l + 2)z3 = 0.
1
2
However, even in the ρ = const case, the Eulerian pres-
sure perturbation is non-zero, and, to leading order, is given
by δp = δpl+1(r)Yl+1 l. We define the dimensionless pres-
sure perturbation, p = δpl+1/ρgr, where g = GM (r)/r2
is the local gravitational acceleration. Finally, we define
the dimensionless shear modulus, µ = µ/ρgr. Through-
out this paper, we use µ = 1.5 × 10−4(R/r)2, chosen so
that the the frequency of the l = m = 2 fundamental
torsional crustal mode is close to the ≈ 50 Hz value com-
puted by (McDermott et al. 1988) for a non-rotating neu-
tron star; the shear wave speed corresponding to this choice
is ct = (µ/ρ)1/2 = 1.7 × 108 cm s−1.
These variables allow us to write the equations of mo-
tion in the crust as a system of first-order differential equa-
tions:
dz1
dr
dp
dr
= −2p + ω2z1 − 2ω Ωlz3 − 2ω ΩlJ l
= (l + 1)(l + 2)z3 − 3z1
l+1z5
r
r
(5a)
(5b)
+ µ(l + 1)(l + 2)(z4 + 2z3 − 2z1)
= z4 − z1
= −2z1 + 2 [l(l + 3) + 1] z3 − 3z4
+
+
1
µ (cid:20)p +
2lJ l
l+1 ω Ω
l + 1
z5
2lω Ω
(l + 1)(l + 2)
(z1 + z3) − ω2z3(cid:21)
= z6
= −3z6 + {l(l + 1) − 2} z5
r
r
dz3
dr
dz4
dr
r
r
dz5
dr
dz6
dr
(5c)
(5d)
(5e)
(5f)
+
1
µ (cid:20)(cid:18) 2ω Ω
l + 1
− ω2(cid:19) z5 +
2J l
l+1 ω Ω
l + 1
(z1 + (l + 2)z3)(cid:21)
where J l
l+1 = (2l+3)−1/2. The derivation of these equations
is analogous to that of Eqs. (9) -- (21) of LS. In the liquid core,
the only nontrivial equations are
r
dp
dr
= (l − 1)p
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
(6a)
Crust-core coupling and r-mode damping
3
r
dz1
dr
= −(l + 4)z1
2 Ω
l + 1(cid:19) (l + 1)4(2l + 3)
8l Ω3
− (cid:18)ω −
and
z5 = −
(l + 1)
2lJ l
l+1
p
ω Ω
p,
(6b)
(7)
(see, e.g., Kokkotas & Stergioulas 1999 or LS).
Both at the crust-core boundary, and at the top of the
crust, we require the tangential traction to be zero, i.e.,
z4 = z6 = 0.
(8)
We hence neglect the viscous coupling between the trans-
verse r-mode motions and the crust. This is the coupling
considered and found negligible by BU. At the crust-core
boundary the radial displacement z1 is continuous, and so is
the radial component of the traction, ∆p − 2µdul+1/dr, or
z1(r−
p(r+
c ) = z1(r+
c )
c ) − µ[2(l + 1)(l + 2)z3(r+
c ) − 4z1] = p(r−
(9b)
(9a)
c ).
At the top of the crust, the radial component of the traction
is zero:
p(R) − z1(R) − µ[2(l + 1)(l + 2)z3(R) − 4z1(R)] = 0
(10)
Equations (8) -- (10) provide the boundary conditions for the
equations of motion. We normalize our solutions to have unit
transverse displacement just inside the liquid core,
z5(r−
c ) = 1.
(11)
With this normalization, the slippage is δu/u = [(z5(r−
z5(r+
sumption, z3 ≪ z5).
c ) −
c ))2]1/2 (where we used our as-
c ))2 + (z3(r−
c ) − z3(r+
2.2 Numerical results
The keplerian spin frequency Ωk, the shear modulus µ, and
the ratio of the crust thickness to the stellar radius rc/R are
the only parameters necessary to describe a constant density
star (we neglect the thin surface ocean and the modes that
are present in it, as they are unlikely to influence the crust-
core slippage). The crust thickness is roughly proportional
to the scale height at the crust-core boundary, and is hence
∝ R2/M . In order to survey the range of parameters, we
carry out our calculations for two models, a "thin" crust
model with rc/R = 0.9, and a "thick" crust model with
rc/R = 0.8. Because of the differences in the normal mode
spectrum between these two cases, we shall see that the r-
mode instability curves for the two models have qualitatively
different behavior in the temperature range of interest for
accreting neutron stars, T = 108 − 109 K.
The left panel of Figure 1 displays the results of our
calculations for l = m = 2 for the thin crust model, while the
right panel shows the results for the thick crust model. Let us
first concentrate on the top plots, which show the dispersion
relation, ω(Ω), for the two models. Contrary to the case of a
purely fluid star, where, in the isentropic case, there is only
one "classical" l = m r-mode, all the modes displayed in
Figure 1 have displacements of the form given by Eq. (2),
with wl ∝ rl in the core. However, only the mode with ω
closest to the "classical" r-mode value, 2Ω/3, is of interest to
4
Levin & Ushomirsky
k
/
Ω
ω
y
c
n
e
u
q
e
r
F
e
d
o
M
/
u
u
δ
m
u
m
i
n
i
M
0.4
0.3
0.2
0.1
0.0
100
10−1
10−2
0.0
k
/
Ω
ω
y
c
n
e
u
q
e
r
F
e
d
o
M
/
u
u
δ
m
u
m
i
n
i
M
0.4
0.3
0.2
0.1
0.0
100
10−1
10−2
0.0
0.2
0.1
Spin Frequency Ω/Ω
0.3
0.4
k
0.5
0.2
0.1
Spin Frequency Ω/Ω
0.3
0.4
k
0.5
Figure 1. Top left: Mode angular frequency ω (in the rotating frame) as a function of the angular spin frequency Ω for the five lowest-
frequency modes of the thin crust model. The dashed line shows the "classical" r-mode dispersion relation, ω = 2Ω/3, and the dots mark
the mode closest to this relation, which has the smallest slippage. Bottom left: absolute value of the slippage of the mode marked with
thick dots in the top panel (i.e., the minimum slippage) as a function of Ω. Right: same as top left, but for the thick crust model.
us. This mode is denoted by thick dots in Figure 1. When
ω is much different from 2Ω/3, the displacements in the
crust are much bigger than the displacements in the core,
resulting in large slippage, and, hence, large damping. In
essence, unless the mode frequency is close to the r-mode
value, the elasticity of the crust plays the dominant role in
determining the mode properties, and the mode is a crustal
mode, rather than a core r-mode.
Consider the low spin frequency (Ω ≪ (µ/ρR2)1/2 ≈
0.05Ωk) limit of our results. In this case, the mode with
the lowest frequency in Figure 1 is the "classical" l = m
r-mode, which has the transverse displacement w2 ∝ r2 in
the core and negligible displacements in the crust. This be-
havior (liquid in a bucket) is the approximation used by BU.
The modes at higher frequencies are torsional crustal oscil-
lations, which have appreciable displacements in the crust,
and negligible displacements in the core. The fundamental
(nodeless) torsional mode has ω = 0.03Ωk for both models.
The lack of sensitivity of the frequency of this mode to the
thickness of the crust was first noted by Hansen & Cioffi
(1980). The modes at higher frequencies are overtones, and
we note that the mode spectrum for the thick crust model is
denser than that of the thin crust model. Qualitatively, the
mode frequency is determined by fitting a certain number
of nodes of the eigenfunction between the boundaries of the
crust, and hence a thicker crust supports more modes.
As the spin frequency is increased beyond Ω ≈
(µ/ρR2)1/2, the behavior of the modes changes. The fre-
quency of the core r-modes rises to meet with the frequencies
of the modes which originally resided in the crust, resulting
in avoided crossings. At the first avoided crossing, the core
r-mode is coupled with the lowest-order torsional mode of
the crust; at this avoided crossing the core r-mode strongly
penetrates the crust. As the spin frequency increases beyond
the first crossing, the crust-core slippage decreases along the
branch closest to the "classical" r-mode frequency, ω = 2Ω/3
(marked by a dashed line in Figure 1). Since the boundary
layer damping rate is proportional to (δu/u)2, this branch is
most interesting for us as it corresponds to the r-mode with
the lowest damping, and, hence, the most unstable.
The relative slippage as a function of the spin frequency
is plotted in the bottom panels of Figure 1. We always plot
slippage of the r-mode from the branch with the lowest
damping, marked by thick dots in Figure 1 (i.e., the most
unstable r-mode). As the spin frequency increases beyond
the first crossing, the slippage decreases to ≈ 0.1. How-
ever, the thick crust model has another avoided crossing
at Ω ≈ 0.25Ωk, and the slippage rises at this frequency be-
cause the core r-mode couples resonantly with a higher-order
torsional crustal mode. In the thick crust model, the "val-
ley" in which the second mode dominates only extends to
Ω ≈ 0.25Ωk. After the avoided crossing, the third mode
has the smallest slip, 4 × 10−2
∼< δu/u ∼< 1. Note that the
slip value is ≈ 1 near the avoided crossing and is ∼> 0.1
for 0.25Ωk ∼< Ω ∼< 0.3Ωk, or for spin frequencies between
∼ 500 and 600 Hz. In this regime, the viscous boundary
layer damping rate approaches the high BU value, and so
the critical spin frequency for the r-mode instability is high
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
as well. As we now show, the shape of the r-mode stability
curve depends quite sensitively on the thickness of the crust.
103
Crust-core coupling and r-mode damping
5
3 CRITICAL STABILITY CURVES
The dissipation rate in the viscous boundary layer is set
by the amplitude of the relative motion between the core
and the crust. BU assumed that the crust remains static
in the rotating frame, and hence the amount of boundary
slippage δu/u = 1. However, as shown in the previous sec-
tion, the slippage δu/u is generally significantly less than 1.
Therefore, the BU estimate of the damping rate has to be
multiplied by (δu/u)2. The corrected r-mode damping rate
is [cf. Eq. (4) of BU]
1
τvbl
≃ 0.01 s−1 R2
6F 1/2
M1.4T8
ρb
ρ (cid:16) f
kHz(cid:17)1/2(cid:16) δu
u (cid:17)2
,
(12)
where f = Ω/2π is the neutron star spin frequency, R6,
M1.4 and T8 are the radius, the mass and the core temper-
ature of the neutron star in units of 10 km, 1.4M⊙, and
108 K, respectively, ρ is the density at the crust-core in-
terface, ρb = 1.5 × 1014 g cm−3 is the estimated value of
this density (Pethick et al. 1995), and F is the fitting pa-
rameter used by Cutler & Lindblom (1987) to parameterize
shear viscosity found by Flowers & Itoh (1979). The Ekman
layer correction introduced by Rieutord (2000) for the con-
stant density case is of order unity and is not included in
Eq. (12); see Lindblom et al. (2000) for a treatment appli-
cable for more realistic, non-uniform density stars. We also
neglected the contribution of the elastic energy of the crust
to the total mode energy, as well as the small deviation of
the mode frequency from ω = 2Ω/3.
The r-mode growth rate due to gravitational radiation
is given by
1
τgw
≃ 0.02 s−1M1.4R4
kHz(cid:17)6
6(cid:16) f
.
(13)
Note that this expression is different from the one used by
BU. Andersson et al. (2000) pointed out that BU incorrectly
rescaled τgw used by Lindblom et al. (1998) to the canon-
ical R = 10 km. Andersson et al. (2000) gave their own
estimate of the r-mode growth rate for a constant density
star. Eq. (13) is the correct τgw for an n = 1 polytrope, ne-
glecting the deviation of the transverse displacement from
wl ∝ rl in the crust, as well as the small deviation of the
mode frequency from ω = 2Ω/3.
While we computed the slippage using a constant den-
sity model, we utilize τvbl (Eq. [12]) and τgw (Eq. [13]) ap-
propriate for a more realistic n = 1 polytrope. Compared
to more realistic models, using τgw and τrmvbl for a con-
stant density model overestimates the gravitational radia-
tion growth rate (by putting more mass at larger radii), and
underestimates the strength of the boundary layer damping
(by virtue of a larger mode energy). Both of these effects re-
sult in lower critical frequencies for the onset of the r-mode
instability. Slippage calculations for more realistic models
are in progress.
The stability curve is obtained by equating the
gravitational-radiation-induced growth rate with the viscous
dissipation rate. Since we are primarily interested in temper-
atures below the melting temperature of the crust, ≈ 1010 K,
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
)
z
H
(
y
c
n
e
u
q
e
r
F
n
i
p
S
102
107
108
109
1010
Core Temperature (K)
Figure 2. Critical frequencies as functions of the core temper-
ature for the modes in the thin crust model (solid line) and the
thick crust models (dotted line). For comparison, we also dis-
play the critical frequencies for the n = 1 polytrope fluid model
(dashed line), and for a model with a crust and δu/u = 1 inde-
pendent of the spin (dot-dashed line).
we neglect the dissipation due to bulk viscosity. We also ig-
nore the dissipation due to the shear in the interior of the
star, as well as any potentially important dissipation mech-
anisms that may be operating in the crust, and presume
that the viscous boundary layer is the dominant energy sink.
Since we do not have an analytical expression for the slip-
page, it is more convenient to evaluate the critical tempera-
ture, T8, such that τgw = τvbl:
T8 = 0.5
F 1/2
6M 2
1.4
R2
ρb
ρ (cid:16) f
kHz(cid:17)−5.5(cid:16) δu
u (cid:17)2
.
(14)
6M 2
Figure 2 shows the critical stability curves for models with
thin (solid line) and thick (dotted line) crusts, with F , M1.4,
R6 and ρb/ρ all equal to one. In other words, the proper
1.4F −1/2ρ/ρb. For
label for the abscissa of this plot is T R2
comparison, we also plot the stability curve for a purely fluid
n = 1 polytrope (Lindblom et al. 1998, Lindblom et al. 1999,
dashed line), as well as the curve with δu/u = 1 independent
of spin frequency (dash-dotted line, cf. BU). Neglecting the
shear viscosity in the interior of the star is clearly justified.
For the thin crust model (solid line), the slippage in the
frequency range of interest does not rise or fall sharply, and
so the critical frequency is a single-valued, monotonically de-
creasing function of T . However, for the neutron star with
a thick crust, the slippage and the damping rise sharply
at the spin frequency close to the second avoided crossing.
Therefore, the critical temperature is not a monotonically
decreasing function of the spin frequency f , and the stabil-
ity curve is multi-valued. Such a non-trivial stability curve
should be interpreted in terms of the critical temperature,
6
Levin & Ushomirsky
rather than the critical frequency: for temperatures larger
than the critical one given by Eq. (14), i.e., for points in
the T -- f plane which lie to the right of the viscous bound-
ary layer part of stability curve, the r-mode is unstable and
grows. Otherwise, the r-mode is stable and decays.
4 DISCUSSION, BUT NO BETS
Observationally, the spin frequencies of neutron stars in
LMXBs have been inferred from the QPO-like flux mod-
ulation during the type I X-ray bursts (the so-called burst
QPOs, see van der Klis 1999 for a review). In two out of
the six sources displaying the burst QPOs, the frequency
is close to 300 Hz, while in the remaining four it is close to
600 Hz. In one of the latter four systems, Miller (1999) found
a subharmonic at around 300 Hz, and has argued that this
subharmonic corresponds to the true spin frequency. Same
relation is sometimes believed to hold for the other ≈ 600 Hz
bursters. In addition, ≈ 300 Hz spin is often argued on the
basis of a beat frequency interpretation of the kHz QPO
peak separation in all six of the burst QPO systems, as well
as for the neutron stars in thirteen additional LMXBs that
do not display burst QPOs. On the other hand, there is no
compelling theoretical picture for why the subharmonics of
the ≈ 600 Hz burst QPOs should be so weak or invisible, so
a bimodal spin distribution is certainly not ruled out.
The new critical stability curves computed in this pa-
per for two toy models of neutron stars cut right through
the observed range of spins both for neutron stars in LMXBs
and for millisecond pulsars, for temperatures of interest for
accreting neutron stars (between 108 K and 109 K; Brown
& Bildsten 1998, Brown 2000). An accreting neutron star
can continue its spin-up so long as it is located to the left of
the stability curve in Figure 2. However, once it crosses the
critical stability curve and is located to the right from it,
the r-mode will grow and halt the spin-up, probably forcing
the neutron star into a thermal runaway cycle (Levin 1999;
Spruit 1999). The details of what happens during this run-
away and how violent it is are a topic of current research.
For example, the displacements induced in the crust may
cause it break when the strain (which is of order the r-mode
amplitude α, as defined in Lindblom et al. 1998) exceeds a
critical value (∼< 10−2, see § 6.1 of Ushomirsky et al. 2000
and references therein for a summary of estimates of crustal
breaking strains) or melt if the heating due to the viscous
boundary layer raises its temperature above ≈ 1010 K (see
Lindblom et al. 2000 for the calculation of the r-mode ampli-
tude required to melt the crust). The evolution of the spin
frequency and the temperature depends on the saturation
of the r-mode growth, which is not currently understood.
Despite these uncertainties, it is clear that, if the r-mode in-
stability operates in accreting neutron stars, the non-trivial
shape of the stability curve and its sensitivity to the crustal
structure will be reflected in a peculiar, non-uniform spin
distribution of accreting neutron stars and millisecond pul-
sars.
Much work is necessary in order to make robust compar-
isons with observations. R-mode calculations for more realis-
tic neutron star models, taking into account the complicated
structure of the crust are in progress. We have not consid-
ered the possible superfluid nature of neutrons and protons
at the crust-core interface. In particular, we have not consid-
ered the pinning of the superfluid vortices to the crust. We
assumed that the boundary layer is laminar, which may not
be true for large r-mode amplitudes. Significant magnetic
field at the crust-core boundary may also affect the crust-
core coupling (Rezzolla et al. 2000; Ho & Lai 2000). Thus,
our estimate of the r-mode damping may change when more
complex physics is considered.
We thank Lars Bildsten for comments that helped im-
prove this manuscript, and Eliot Quataert and Re'em Sari
for numerous discussions. YL thanks the Institute for Ad-
vanced Study, where part of the research for this work was
done, for hospitality. YL is supported by the Theoretical As-
trophysics Center at UC Berkeley, and GU is supported by
NSF Grant AST-9618537 and NASA grant NGC5-7034.
References
Andersson N., 1998, ApJ, 502, 708
Andersson N., Jones D. I., Kokkotas K. D., Stergioulas N.,
2000, ApJ, 534, L75
Andersson N., Kokkotas K. D., Stergioulas N., 1999, ApJ,
516, 307
Bildsten L., 1998, ApJ, 501, L89
Bildsten L., Ushomirsky G., 2000, ApJ, 529, L33
Brown E. F., 2000, ApJ, in press. Preprint available, astro-
ph/9910215
Brown E. F., Bildsten L., 1998, ApJ, 496, 915
Cutler C., Lindblom L., 1987, ApJ, 314, 234
Flowers E., Itoh N., 1979, ApJ, 230, 847
Friedman J. L., Lockitch K. H., 1999, Prog. Theor. Phys.
Suppl. 136, 121
Hansen C. J., Cioffi D. F., 1980, ApJ, 238, 740
Ho W. C. G., Lai D., 2000, ApJ, submitted. Preprint avail-
able, astro-ph/9912296
Kokkotas K. D., Stergioulas N., 1999, A&A, 341, 110
Lee U., Strohmayer T. E., 1996, A&A, 311, 155
Levin Y., 1999, ApJ, 517, 328
Lindblom L., Mendell G., 1999, Phys.Rev.D, 61, 104003
Lindblom L., Mendell G., Owen B. J., 1999, Phys.Rev.D,
60, 064006
Lindblom L., Owen B. J., Morsink S. M., 1998,
Phys.Rev.Lett, 80, 4843
Lindblom L., Owen B. J., Ushomirsky G., 2000,
Phys.Rev.D, in preparation.
McDermott P. N., Van Horn H. M., Hansen C. J., 1988,
ApJ, 325, 725
Miller M. C., 1999, ApJ, 515, L77
Pethick C. J., Ravenhall D. G., Lorenz C. P., 1995,
Nucl. Phys. A, 584, 675
Provost J., Berthomieu G., Rocca A., 1981, A&A, 94, 126
Rezzolla L., Lamb F. K., Shapiro S. L., 2000, ApJ, 531,
L139
Rieutord M., 2000, ApJ, submitted. Preprint available,
astro-ph/0003731
Spruit H. C., 1999, A&A, 341, L1
Ushomirsky G., Cutler C., Bildsten L., 2000, MNRAS, sub-
mitted. Preprint available, astro-ph/0001136
van der Klis M., 1999, in Proceedings of the Third William
Fairbank Meeting, available from astro-ph/9812395
Zahn J. P., 1966, Ann. d'Ap. 29, 313
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
|
0812.0388 | 1 | 0812 | 2008-12-01T23:15:55 | The Calibration of Stromgren uvby-Hbeta Photometry for Late-Type Stars -- a Model Atmosphere Approach | [
"astro-ph"
] | We aim to test the power of theoretical calibrations based on a new generation of MARCS models by comparisons with observational photomteric data. We calculate synthetic uvby-Hbeta colour indices from synthetic spectra. A sample of 388 field stars as well as stars in globular clusters is used for a direct comparison of the synthetic indices versus empirical data and for scrutinizing the possibilities of theoretical calibrations for temperature, metallicity and gravity. We show that the temperature sensitivity of the synthetic (b-y) colour is very close to its empirical counterpart, whereas the temperature scale based upon Hbeta shows a slight offset. The theoretical metallicity sensitivity of the m1 index (and for G-type stars its combination with c1) is somewhat larger than the empirical one, based upon spectroscopic determinations. The gravity sensitivity of the synthetic c1 index shows a satisfactory behaviour when compared to obervations of F stars. For stars cooler than the sun a deviation is significant in the c1-(b-y) diagram. The theoretical calibrations of (b-y), (v-y) and c1 seem to work well for Pop II stars and lead to effective temperatures for globular cluster stars supporting recent claims by Korn et al. (2007) that atomic diffusion occurs in stars near the turnoff point of NGC 6397. Synthetic colours of stellar atmospheres can indeed be used, in many cases, to derive reliable fundamental stellar parameters. The deviations seen when compared to observational data could be due to incomplete linelists but are possibly also due to effects of assuming plane-parallell or spherical geometry and LTE. | astro-ph | astro-ph |
Astronomy&Astrophysicsmanuscript no. Onehag
June 3, 2018
c(cid:13) ESO 2018
The Calibration of Str omgren uvby-Hβ Photometry for Late-Type
Stars – a Model Atmosphere Approach
Anna Onehag1, Bengt Gustafsson1, Kjell Eriksson1, and Bengt Edvardsson1
Department of Physics and Astronomy, Uppsala Astonomical Observatory, Box 515, S-751 20 Uppsala, Sweden
Received / Accepted
ABSTRACT
Context. The use of model atmospheres for deriving stellar fundamental parameters, such as Teff, log g and [Fe/H], will increase as we
find and explore extreme stellar populations where empirical calibrations are not yet available. Moreover calibrations for upcoming
large satellite missions of new spectrophotometric indices, similar to the uvby−Hβ system, will be needed.
Aims. We aim to test the power of theoretical calibrations based on a new generation of MARCS models by comparisons with
observational photomteric data.
Methods. We calculate synthetic uvby−Hβ colour indices from synthetic spectra. A sample of 388 field stars as well as stars in
globular clusters is used for a direct comparison of the synthetic indices versus empirical data and for scrutinizing the possibilities of
theoretical calibrations for temperature, metallicity and gravity.
Results. We show that the temperature sensitivity of the synthetic (b − y) colour is very close to its empirical counterpart, whereas
the temperature scale based upon Hβ shows a slight offset. The theoretical metallicity sensitivity of the m1 index (and for G-type stars
its combination with c1) is somewhat larger than the empirical one, based upon spectroscopic determinations. The gravity sensitivity
of the synthetic c1 index shows a satisfactory behaviour when compared to obervations of F stars. For stars cooler than the sun a
deviation is significant in the c1–(b − y) diagram. The theoretical calibrations of (b − y), (v − y) and c1 seem to work well for Pop II
stars and lead to effective temperatures for globular cluster stars supporting recent claims by Korn et al. (2007) that atomic diffusion
occurs in stars near the turnoff point of NGC 6397.
Conclusions. Synthetic colours of stellar atmospheres can indeed be used, in many cases, to derive reliable fundamental stellar
parameters. The deviations seen when compared to observational data could be due to incomplete linelists but are possibly also due
to effects of assuming plane-parallell or spherical geometry and LTE.
Key words. stars: atmospheres – stars: synthetic spectra – stars: fundamental parameters – techniques: photometric
1. Introduction
The uvby photometric intermediate-band system of Stromgren
(1963) and the Hβ narrow-band system (Crawford 1958,
1966) were combined early and soon proved to be a most
powerful means of determining fundamental parameters of
stars. Calibrations of the systems were also developed early
by Stromgren, Crawford and collaborators. Generally, semi-
empirical methods were used with the calibrations made by
means of sets of stars with parameters determined in more fun-
damental ways. The first metallicity calibration of the m1 index
for solar-type stars by Stromgren (1964) was thus based on spec-
troscopical [Fe/H] values of Wallerstein (1962) and the first lu-
minosity calibration of the c1 index by Stromgren (see Crawford
1966) used cluster stars in the main-sequence band. This general
semiempirical approach continued with calibrations like those
of the m1 index by Nissen (1970), Gustafsson & Nissen (1972),
Nissen & Gustafsson (1978) and Nissen (1981) where very-
narrow-band spectrophotometry of groups of weak lines was
calibrated with synthetic spectra. These stars were then used as
calibration stars for uvby-Hβ photometry. Other calibrations fol-
low this semiempirical approach e.g. those of Crawford (1975),
Ardeberg & Lindgren (1981) and Olsen (1988) for G and K-
type stars, as well as by Schuster & Nissen (1989) for metal-
Send
[email protected]
requests
offprint
to:
Anna
Onehag,
e-mail:
poor stars. In the calibration by Alonso et al. (1996) the effec-
tive temperature calibration was based on stellar temperatures
from the Infrared Flux Method. The calibrations by Nordstrom
et al. (2004) and Holmberg et al. (2007) additionally utilised
Hipparcos parallaxes for the surface gravity calibration.
Another development towards calibration of the systems also
started early: the direct calculation of photometric indices by
means of model atmospheres. Such theoretical calibrations were
attempted for early-type stars with relatively line-free spectra.
For late type stars, a statistical correction for the effects of spec-
tral lines was made by Baschek (1960) in his calibration of the
Stromgren m index (a predecessor to m1). A first systematic and
detailed calculation of uvby−Hβ indices for a grid of F and G
dwarf model atmospheres was published by Bell (1970), using
scaled solar model atmospheres. Bell & Parsons (1974) calcu-
lated uvby colours for flux-constant model atmospheres of F and
G supergiants, while Gustafsson & Bell (1979) produced theo-
retical colours in a number of systems, including the uvby sys-
tem, for a grid of giant-star model atmospheres. Relya & Kurucz
(1978) calculated uvby and UBV colours from early ATLAS
models, and discussed their shortcomings for late-type stars.
uvby colours for new sets of Kurucz models were published by
Lester, Gray & Kurucz (1986). Castelli & Kurucz (2006) pub-
lished Hβ indices. Sometimes, semiempirically corrected fluxes
from model atmospheres have also been used for calibrations of
Stromgren photometry, see e.g. Lejeune et al. (1999) and Clem
et al. (2004).
2
Onehag et al.: Calibration of Stromgren Photomtery
The need for reliable calibrations of uvby−Hβ photometry
has increased in the last decade, not the least for estimating pa-
rameters of new and more "exotic" stars, such as very metal-poor
and super-metal-rich stars which are not found at great abun-
dance in the solar neigbourhood so that relatively complete sets
of calibration stars cannot easily be established. Furthermore,
the preparation for the Gaia satellite includes a careful analysis
of the power of model atmospheres to provide a detailed astro-
physical calibration of the photometric system of the satellite.
This analysis needs support by a detailed test of the problems
and possibilities to make a detailed theoretical calibration of,
e.g., the uvby−Hβ photometry.
Subsequently, we shall present the theoretical models and
colours (Sect. 2 and 3). Stellar samples for empirical compar-
isons are discussed in Section 4. Next, the discussion will be
focused on the determination of effective temperature, metallic-
ity and surface gravity of the stars, by discussing the calibration
of (b − y) and Hβ indices (Sect. 5, effective temperature), of m1
(Sect. 6, metallicity) and c1 (Sect. 7, gravity) indices, devoting
more limited interest to "secondary" effects such as the metallic-
ity sensitivity of (b − y) and c1, or the gravity sensitivity of m1.
In each section the results will be compared with empirical and
semi-empirical data and calibrations. Finally, in the last Section
some comments will be made on the success and the problems of
the theoretical calibrations, conclusions will be drawn and rec-
ommendations given.
2. Model atmospheres and calculated spectra
The theoretical tools used in modelling the stellar colours are
model stellar atmospheres and their calculated fluxes. These are
based on extensive atomic and molecular data. Here, we shall
briefly present the models and data used and refer to more com-
plete descriptions.
2.1. Modelatmospheres
The stellar atmosphere code MARCS (Gustafsson et al. 2008,
http://marcs.astro.uu.se) was used to construct a grid of 168 theo-
retical 1D, flux constant, radiative + mixing-length convection,
LTE models with fundamental atmospheric parameters as fol-
lows: Teff = 4500, 5000, 5500, 6000, 6500 & 7000 K, log g =
2.0, 3.0, 4.0, 4.5 and [Me/H] = 0.5, 0.0, –0.50, –1.0, –2.0, –
3.0, –5.0, [Me/H] denoting the logarithmic over-all metallicity
with respect to the sun. Plane parallell (ppl) stratification was
assumed for log g = 4.5, 4.0 & 3.0, whereas spherical (sph) sym-
metry was assumed for log g = 2.0. For the spherically symmet-
ric models a mass of 1 M⊙ was adopted. The local mixing-length
recipe was used to describe convective fluxes, for more details
see Gustafsson et al. (2008).
Elemental abundances were adopted from Grevesse and
Sauval (1998) except for the CNO abundances which were
adopted following Asplund et al. (2005).
2.2. Syntheticspectrumcalculations
In order to calculate synthetic spectra of sufficiently high resolu-
tion the Uppsala BSYN code was used with the MARCS models
as input. The spectra were calculated within the wavlength limits
of the Stromgren uvby filters with wavelength steps of 0.02 Å.
The mictroturbulence parameter, ξt, was set to 1.0 km s−1 and
1.7 km s−1, when calculating spectra based on ppl models and
sph models, respectively. The effects of changing ξt are studied
in Section 6.
2.3. Linelists
We collected atomic line data from the Vienna Atomic Line
Data Base, VALD (version I, Kupka et al. 1999). For hydro-
gen line data a version of the code HLINOP was used, and has
been described by Barklem & Piskunov (2003). This code has
been developed based on the original HLINOP by Peterson &
Kurucz (see http:// kurucz.harvard.edu/). The hydrogen line pro-
files are calculated including Stark broadening, self-broadening,
fine structure, radiative broadening, and Doppler broadening
(both thermal and turbulent). The Stark broadening is calculated
using the theory of Griem (1960 and subsequent papers) with
corrections based on Vidal et al. (1973). Self-broadening is in-
cluded following Barklem et al (2000) for Hα, Hβ and Hγ, while
for other lines the resonance broadening theory of Ali & Griem
(1966) is used. For molecules, data for C2 including 12C13C lines
were gathered from Querci, Querci & Kunde (1971) and Querci
(1998, private communication with B. Plez), except for the Fox-
Herzberg band in the UV for which data was taken from Kurucz
(1995). Molecular data for CH with 13CH comes from Plez et al.
(2008) and Plez (2007, priv. communication). Data for the CN
molecule with 12C15N, 13C14N, and 13C15N are also from Plez
(priv. communication). CO data with 13C16O was gathered from
Kurucz (1995) as well as NH with 15NH, OH with 18OH, and
MgH with 25MgH and 26MgH. TiO was not taken into account in
the calculations since absorption by this molecule is expected to
have small effects on the spectra within the effective-temperature
range considered.
3. Colour index calculations
3.1. Filterprofiles
To determine the theoretical m1, c1 and (b − y) indices, transmis-
sion profiles of the Stromgren uvby filters (Crawford & Barnes
(1970) see Fig. 1) were multiplied with the calculated model
stellar surface flux within the wavelength range of the filters:
mag = −2.5 log
R FλTλdλ
R Tλdλ
+ const.
where Fλ and Tλ are the flux and the relevant transmission
profile, respectively. The theoretical magnitudes were converted
into colour indices via the definitions: c1 ≡ u − 2v + b, m1 ≡
v − 2b + y.
The Hβ index is defined (Crawford 1958) as the ratio of the
flux measured through a narrow and a wide profile, respectively,
both centered around the Hβ line:
Hβ = −2.5
log
R FλTN,λdλ
R TN,λdλ
− log
R FλTW,λdλ
R TW,λdλ
+ const.
where TW,λ and TN,λ are the wide and the narrow filter profiles,
respectively. We have found notable differences between syn-
thetic Hβ indices calculated with the Kitt Peak filter set denoted
(9, 10) and those calculated with the (212, 214) set, both de-
scribed by Crawford & Mander (1966, CM66). Here, the Kitt
Peak (212, 214) filter system was chosen (Fig. 1 CM66) with a
transmission function adopted from Castelli & Kurucz (2006),
see Figure 1.
Onehag et al.: Calibration of Stromgren Photomtery
3
characteristics, such as certain non-solar abundance ratios as
well as dust emission in the IR (see Gigas 1988, Hill 1995,
Ilijic et al. 1998, Adelman & Gullliver 1990, Heiter, Weiss
& Paunzen 2002). However, these departures from standard
A0 stars, as well as standard model atmospheres, are thought
to only lead to minor modifications of its uvby−Hβ indices
(Paunzen et al. 2002). A more practical problem is that neither
of the two CM66 transformation equations from H′β to Hβ
will give a fully satisfactory fit due to the fact that the index
for Vega should be transformed intermediately between the B
star and the A, F star sets. Several tests were performed which
all pointed in the direction that Vega should be transformed to
the standard system via an equation somewhere in between the
B and the A, F transformations but with a heavier weight for
the latter. By using all listed A0 stars in Table II of CM66, a
special transformation between H′β and Hβ for A0 dwarf stars
was established, as can be seen in Figure 2. The vertical line
in the plot represents the observed Hβ value for Vega (Hauck
& Mermilliod, 1998). However, the observed H′β value of
Vega is not known. In order to determine the zero point c in
the transformation from calculated (H′′β) to "observed" H′β
values, H′′β = H′β + c, we have therefore adopted an H′β value
for Vega derived from the A0 stars line in Figure 2, read off at
the observed Hβ for the star. After correction of the model H′′β
value to H′β we have then calculated the model Hβ values by
using the empirical A,F-transformation relation as mentioned
earlier.
Fig. 2. Hβ and H′β values for A0 stars tabulated in CM66 are
plotted individually and represented by a linear regression (thick
line), as well as the transformation relations for B and A+F stars,
respectively, following CM66 (thin lines).
The transmission functions of the filters used by Olsen
(1983, 1984) and Schuster & Nissen (1988) depart from those on
which the system was originally based (Crawford 1966). Filter
profiles representative for these newer studies are given by Helt
et al. (1987) and Bessell (2005). In particular, the more recent v
band is narrower, and its effective wavelength shifted by about
25 Å towards the red. To test the effects of this, we have calcu-
lated colours with the Helt et al. (1987) profiles as an alterna-
tive to the Crawford set, and then applied the transformations
as given by Schuster & Nissen (1988) back to the standard sys-
tem to mimic the procedure of the observers. From this, we find
Fig. 1. The uvby−Hβ transmission functions of the standard sys-
tems plotted as a function of wavelength. As a comparison,
the flux (per Ångstrom unit) of a model with Teff = 6000 K,
log g = 4.0 and [Me/H] = 0.0 is plotted on an arbitrary flux scale.
3.2. Transformationtotheobservationalsystem
The frequent use of filter-profiles different from those that
originally have defined the photomteric system needs some
extra considerations. For the Hβ index this mainly comes in at
the zero-point determination. Since the observed indices are
defined on a system with zero-points set by particular standard
stars, we must apply corresponding zero-point shifts to our
theoretical system. The much observed, and frequently used,
star Vega was chosen to estimate these zero-points. Thus, a
model atmosphere for Vega was calculated with parameters Teff
= 9550 K, log g = 3.95 and [Me/H] = –0.5 (mean values of
selected measurements presented in SIMBAD), and a synthetic
spectrum was computed. Theoretical indices for the Vega model
were calculated and compared to observed values: c1 = 1.088,
m1 = 0.157 and (b − y) = 0.003 (Hauck and Mermilliod 1998).
The resulting differences between the theoretical and observed
colours were then added as constants to all calculated Stromgren
colours in our model grid.
For the Hβ index, Crawford & Mander (1966) presented several
filter systems, of which we choose the (212, 214) filters as
mentioned above. Indices calculated by using this filter set,
which are referred to as H′′β below, should be transformed
via a set of equations (CM66), in order to agree with previous
Crawford–Mander observations on their standard system. The
transformation for the (212,214) filters is described by the
following two equations (CM66, Table III):
Hβ = 0.374 + 1.305 H′β, B stars
Hβ = 0.248 + 1.368 H′β, A, F stars,
derived from a set of 45 and 35 bright stars, where H′β
and Hβ are the observed and tranformed indices, respectively.
The later equation for A and F stars was applied on all Hβ
indices in our theoretical grid.
The A0 star Vega clearly poses a number of problems for
determining the zero-point of the uvby and Hβ indices. It is
known to be rapidly rotating, but with its axis close to the line of
sight (Gulliver, Hill & Adelman 1994, Hill, Gulliver & Adelman
2004). Vega has also been regarded to show mild λ Bootis star
4
Onehag et al.: Calibration of Stromgren Photomtery
changes of the calculated values, amounting to typically 0.02
magnitudes or less in the m1 and c1 index values and only 0.002
in (b − y). In particular, the changes in the differential values
measuring the sensitivity of m1 to metallicity and of c1 to grav-
ity are small, in δm1/δ[Me/H] typically 0.003 and in δc1/δ log g
typically 0.01. Such effects do not change our conclusions in the
present paper.
3.3. Modelcolours
The model colours, with zero-point added using Vega observa-
tions, are supplemented the present paper electronically.
4. Comparison star samples
In order to test the reliability of our calculated colours, we se-
lected a sample of standard stars. These were taken from vari-
ous sources: one subset with well determined spectroscopic pa-
rameters was selected from The Bright Star Catalogue (Hoffleit
& Warren 1995), another from the uvby standard stars listed
by Crawford & Barnes (1970), a third from the list of metal-
poor stars in Schuster & Nissen (1988), a fourth among stars
that have been observed by the Hubble-STIS spectrograph and
a fifth from the stars listed in the study of Pop II stars by
Jonsell et al. (2005). Altogether 388 stars were thus selected.
The fundamental parameters were taken from the sources listed,
or from other sources given in the SIMBAD catalogue and
judged to have high quality. Complementary photometry was
also obtained from SIMBAD. Parameter determinations based
on uvby−Hβ photometry were avoided as far as possible, since
we aimed at testing calibrations based on this photometric sys-
tem relative to parameters based on more fundamental methods.
In practice, this usually means effective temperatures based on
the infrared-flux method and gravities and metallicities based on
high-resolution spectroscopy. The effective temperatures gath-
ered from the Jonsell et al (2005) sample are calculated with the
Alonso et al. (1996) calibration. These stars however, constitute
less than 5 % of our total sample. In the case of multiple sources,
i.e., stellar values listed in more than one of our selected cata-
logues, a mean value was used. The uvby indices of the standard
stars were dereddened by means of the algorithm and computer
code of Hakkila (1997). For most of the stars, in particularly the
dwarfs, E(B-V) was smaller than ∼0.01.
Altogether, the standard stars span a volume in the param-
eter space ranging from 3900 K to 7850 K in Teff, 0.20 to 4.80
in log g (cgs units) and –3.0 dex to 0.45 dex in [Fe/H]. Data for
the full standard sample is accessible electronically from A&A
as on-line material supplementing the present paper. As seen in
Figure 3 our standard sample satisfactorily covers the funda-
mental parameter space, although one would need some more
cool (Teff < 5000) and hot (Teff > 6300) dwarf stars (log g >
4.0). For this purpose two complementary homogeneous sam-
ples were tested; the stellar sample published by Casagrande
et al (2006, hereafter C06) and Valenti & Fischer (2005, here-
after VF05). uvby colour indices for both samples were taken
from Hauck & Mermilliod (1998). These complimentary stel-
lar samples cover the parameter spaces: 4700 ≤ Teff ≤ 6600,
3.1 ≤ log g ≤ 5.1 and −1.9 ≤ [Fe/H] ≤ 0.5 for the VF05 sample
and 4406 ≤ Teff ≤ 6556, log g ∼ 4.5 and −1.87 ≤ [Fe/H] ≤ 0.34
for the C06 sample. Despite that the lowest log g values in the
VF05 sample are characteristic of giants/sub-giants, a majority
(93 %) of the VF05 stars are dwarfs with log g ≥ 4.0. In the C06
sample all stars are assumed to be dwarfs with log g ∼ 4.5. A
majority (85 %) of the stars in the VF05 sample are also metal-
rich, [Fe/H] ≥ −0.2, whereas the C06 sample is evenly dis-
tributed in metallicity. Likewise, the effective temperatures of the
C06 sample (determined by the authors's new calibration of the
infrared-flux method) are evenly distributed but the VF05 sam-
ple is biased towards higher temperatures, Teff ≥ 5500 (74 %).
Among the stars in the standard sample, 18 and 72 are found
in the C06 and VF05 samples, respectively. For the stars in the
VF05 sample no significant trends of discrepacies for the val-
ues given, Teff, log g and [Fe/H], were found. In the C06 sam-
ple we find an overall difference in given effective temperatures,
growing for higher effective temperatures at low metallicities
(Teff standard − Teff Casagrande ≈ −175 K at Teff ∼ 6000 K and −2.0 ≤
[Fe/H] ≤ −1.0).
Fig. 3. Fundamental parameters, i.e. Teff, [Fe/H] and log g, for
the standard sample.
5. The effective-temperature calibration
Precise determinations of effective temperatures of stars are
critical for a variety of reasons, not only for direct applica-
tions such as comparisons of isochrone calculations to observed
colour-magnitude diagrams, but also in more indirect applica-
tions e.g. determinations of chemical abundances. Photometric
indices play a critical role when determining the effective tem-
peratures of stars. For stars located on the main sequence, model
atmosphere calibrations of these indices may be particularly im-
portant since determinations of diameters are generally few and
poor.
The Stromgren (b − y) index provides a sensitive temperature
measure for F, G and K stars. The Hβ index is another frequently
used criterion for deriving temperatures. We study both indices
here.
In Figure 4 (left) we explore the temperature sensitivity of
the theoretical (b − y) index for different metallicities. As a com-
parison we also plot (b − y) for a set of Kurucz models (Lester,
Gray & Kurucz, 1986). We note that the indices, calculated with
the different model sets, more or less show the same behaviour
although the models of Lester et al. are systematically bluer at
given Teff. In comparison with observed values for stars our syn-
thetic indices show a satisfactory temperature sensitivity over the
full temperature range (4500–7000 K), see Figure 5.
In Figure 4 (right) the Hβ index is plotted versus effective
Onehag et al.: Calibration of Stromgren Photomtery
5
temperature, together with the theoretical data of Castelli &
Kurucz (2006). For the warmer part of the temperature range
we find no differences in sensitivity, whereas for the cooler tem-
peratures (5500 K) we note that MARCS indices have a steeper
gradient, thus lower Hβ values than those of Castelli & Kurucz.
Our steeper gradient tends to agree with observed stellar colours,
Figure 6, which is probably due to improvements of the broad-
ening theory of Hβ.
5.1. Teff(b − y)
Several attempts (Alonso et al. 1996, Nordstrom et al. 2004,
Ram´ırez & Mel´endez 2005, Holmberg et al. 2007) have been
made to derive empirical calibrations for uvby−Hβ colours,
where the fundamental parameters are expressed as explicit for-
mulae in the photometric indices. We will here derive such cali-
bration equations by using theoretical colours. We make use of
the formal expressions of Alonso et al. (1996, 1999, hereafter
A96 resp. A99) based on (b − y) (their eq. 9 and 14 & 15, respec-
tively) but derive new coefficients via an iterative least squares
method and by using the model colours.
The result for dwarfs (log g ≥ 4.0) are given numerically in
Appendix A.1 and can be seen in Figure 7 where we also merge
our standard sample with the C06 sample and plot temperatures
for the individual stars derived from our theoretical calibration
equation. The standard deviations for our calculated tempera-
tures with respect to literature values can be found in Table 1
A linear regression for the calculated temperatures of the two
merged samples is also shown in Figure 7 and yields a dispersion
σ = 117 K. We note that our theoretical calibration gives higher
temperatures, as compared with literature values, by ∼100 K for
the cooler stars (Teff < 5000 K) and tends to lead to lower temper-
atures (∼100–150 K) than the values found in the literature for
hotter stars (Teff > 6500 K). Our calibration suggests lower tem-
peratures than the values listed in VF05 by 100-200 K for stars
within the range of 5200-6300 K. Below that our theoretical cali-
bration gives systematically ∼100 K higher temperatures for the
VF05 stars, in accordance with the tendencies found also for the
other comparison samples. The standard deviation of the calcu-
lated temperatures for the VF05 sample can be seen in Table 1.
It is also interesting to compare our results to previous empir-
ical calibrations. The standard deviations of the calculated tem-
peratures using four different empirical calibrations applied on
our three comparison samples, are listed in Table 1. In order to
illustrate the empirical trends the resultant linear regressions for
the merged standard and C06 samples are shown in Figure 7
(note the shifted scale by 1000 K).
The calibration results for giants (1.5 ≤ log g ≤ 3.5) is pre-
sented in Appendix A.2 and shown in Figure 8. The standard
deviations with respect to effective temperatures of the standard
stars are shown in Table 1 together with the standard deviations
for the same sample when using the A99 empirical calibration.
5.2. Teff(Hβ)
A theoretical temperature calibration for dwarfs was also derived
for the Hβ index, with a formal expression adopted from A96,
Eq. 10, see Appendix A.1. The result can be seen in Figure 9.
Standard deviations with respect to the literature values of the
standard sample are listed in Table 1, values somewhat higher
than the calibration based on (b−y). Here we are, however, using
different stars in the Teff(b − y) and Teff(Hβ) equations (fit limits
defined in A96). To make a fair comparison between the two
Table 1. Standard deviations for Teff and [Fe/H] derived using
models as compared with literature values of the different com-
parison samples.
Theoretical
Standard
C06
VF05
All
A06,99 + S&N89
Standard
C06
VF05
All
HNA07
Standard
C06
VF05
All
RM05
Standard
C06
VF05
All
Dwarfs
(b − y)
133
101
185
172
A06
110
118
135
130
101
92
101
100
155
110
128
134
Teff
Hβ
151
-
-
-
156
-
-
-
-
-
-
-
-
-
-
-
Giants
(b − y)
144
-
-
-
A99
116
-
-
-
-
-
-
-
105
-
-
-
[Fe/H]
G-stars
0.263
0.321
0.392
0.368
S&N89
0.183
0.158
0.128
0.141
0.208
0.159
0.123
0.144
0.268
0.176
0.116
0.161
F-stars
0.254
0.190
0.399
0.355
0.182
0.128
0.146
0.157
0.164
0.132
0.130
0.140
0.176
0.139
0.081
0.122
calibrations we have restricted our sets of stars to be identical
for both calibrations. Such a comparison is made in Sec. 5.3.
A linear regression for the calculated effective temperatures is
also shown in Figure 9. The dispersion around this fit is σ =
143 K. We see that our theoretical calibration suggests somewhat
higher effective temperatures, of roughly 100 K, in the cooler
part of the temperature range (Teff ∼ 5000 K), and that it possibly
implies lower effective temperatures of the order of ∼ 100 K for
the warmest stars (Teff > 6500 K). Some of these tendencies may
possibly be traced also in the (b − y) − Teff calibration (c.f. Fig.
7). As an alternative, the effective temperatures for the standard
sample are calculated with the empirical calibration equation by
A06. The standard deviations with respect to literature values
are presented in Table 1, and a linear regression to the calculated
effective temperatures is also plotted in Figure 9.
5.3. Teff(b − y)vs Teff(Hβ)
Now we compare the derived temperatures based on (b − y) and
Hβ, respectively. 119 stars matching the restrictions in the pa-
rameter space for both calibrations were selected out of our stan-
dard sample. The effective temperatures based on (b − y) and Hβ
and the result are displayed in Figure 10. The test reveals that
there is some disagreement between the two theoretical calibra-
tions. The Hβ based equation suggests higher temperatures for
the cooler stars (Teff . 5000 K) and lower temperatures for the
warmer stars (Teff < 6900 K) of some 200 K. The empirical equa-
tion of Alonso et al. (1996) shows more or less the same spread
for the standard sample but no significant departures from the
1-1 line, which is also to be expected. The failure for the theo-
retical calibration for the coolest stars is not very remarkable,
since the metal lines for those stars are strong and dominate the
Hβ line, and other parameters affect the index such as gravity.
The departure in the hotter end is only dependent on a few stan-
dard stars.
It is worth noting that the difference between the (b − y)
and Hβ calibrations responds in opposite directions to differ-
6
Onehag et al.: Calibration of Stromgren Photomtery
Fig. 4. The (b − y) colour index (left) and Hβ index (right) versus effective temperature for MARCS and Kurucz (Lester, Gray &
Kurucz (1986), LGK, Castelli & Kurucz, 2006, CK) model atmospheres with different [Me/H] and log g (indicated to the left and
right of each curve). Corresponding theoretical MARCS (b − y) and Hβ curves for [Me/H] = 0.5, –0.5, and –3.0 can be seen in
Figure 5 and Figure 6, respectivley.
ent gravities. I.e., the temperature sensitivity δ(index)/δ(Teff)
decreases with increasing gravity for (b − y) while it increases
for Teff ≤ 5000 K for Hβ (see Fig. 4). As is seen, these differ-
ences are also metallicity dependent, and tend to vanish for small
metallicities. However, for stars with known reddening and as-
suming e.g. the metallicity to be known and relatively large, this
may make it possible to obtain a temperature and rough gravity
classification from (b − y) and Hβ, only.
6. The metallicity calibration
The basic aim of m1 ≡ (v − b) − (b − y) was to measure the
total intensity of the metal lines in the v band. As was early
appreciated by Stromgren, for late F and G stars of Pop I these
lines are, however, to a large extent located on the flat part of the
Curve of Growth and are thus not very sensitive to metallicity,
but rather to microturbulence. Moreover, for the hotter stars, the
Hδ line is strongly affecting the band, and for stars later than
G5, CN lines of the (0,1) band in the Violet System are also
significant. Thus, the effects of the value of the mircroturbulence
parameter, as well as of the individual CNO abundances, e.g.
due to dredge-up of CNO processed material from the interior,
must be taken into consideration. Certainly, m1 also varies with
effective temperature and, to a less degree, with surface gravity.
The variation of the calculated m1 with (b − y) and metal-
licity for the model atmospheres is shown in Figure 11 and
compared with calculated indices by Lester et al. (1986). As is
seen, the index offers a good discrimination in metallicity ex-
cept for stars of Extreme Population II for which it only works
for the cooler end of the temperature interval. A characteristic
measure of the sensitivity of the index to overall metallicity is
(δ(m1)/δ[Me/H])(b−y), where the subscript denotes that the sen-
sitivity is measured at a constant (b − y). This quantity, as mea-
sured for models with –3.0 ≤ [Me/H] ≤ 0.5 and log g ≈ 4.5 &
4.0, is given in Table 2.
We may compare the sensitivity of the m1 index with the
empirical results. However, recent calibrations of the Stromgen
photometry, e.g. by Holmberg et al. (2007), contain complex
non-linear expressions in which all the indices are involved –
in principle a reasonable approach since e.g. also the c1 index
carries information on the metallicity. Since this metallicity de-
pendence is, however, far from independent of that carried by
the m1 index the terms in the calibration expressions involving c1
may well mask some of the dependence of m1 on metallicity. So,
as we wish to understand the way m1 changes with metallicity,
Onehag et al.: Calibration of Stromgren Photomtery
7
Fig. 5. Theoretical (b −y) values versus effective temperature for models with different [Me/H] and log g plotted together with values
for the standard stars (with [Fe/H] within ±0.1 dex of the given [Me/H]). To guide the eye the theoretical log g curves for 2.0 and
4.5 are plotted as a dashed and a dotted line, respectively.
we have instead turned back to the earlier empirical calibrations
like that of Nissen (1988), where the metallicity dependence of
m1 was still treated separately.
Thus, Nissen (1988) finds
[Fe/H] = −(10.5 + 50(Hβ − 2.626)) · δm0 + 0.12
(1)
where δm0 is calculated as mstand − m1 relative to the Hyades
standard sequence at constant Hβ and corrected for interstel-
lar extinction. From Figure 11 or Table 2 we may estimate
δm1/δ[Me/H] at constant (b − y). We can easily calculate the
corresponding empirical quantity from the coefficient in Eq. (1)
(correcting for the derivative at constant Hβ to constant (b − y)
by adding 0.03 which is easily found with sufficient accuracy
from the model colours) and then find empirical sensitivities
δm1/δ[Fe/H] of 0.093 (7000 K) and 0.15 (5750 K), with rele-
vant effective temperatures within parentheses. These values for
the sensitivities are valid for [Fe/H]≈ 0.0. The corresponding
theoretical sensitivities δm1/δ[Me/H] for dwarf stars are typi-
cally 0.10 and 0.14, respectively. The agreement between the
empirical and theoretical results is quite satisfactory. It seems
possible that a basic reason why the theoretical calibration, in
spite of this agreement, does not succeed very well for the more
metal-poor stars (see below) may rather be due to the modelling
of the fluxes of those stars than due to failures in calculating the
change of flux with metallicity for solar-type stars. It may also be
associated with the measured v band being different to that com-
puted, as a single transformation equation is unlikely to correct
solar metallicity stars and metal-poor stars by the same amount.
In order to further explore the properties of the calculated
m1 indices we have plotted individual stars with fairly well-
determined fundamental parameters, chosen from our standard
sample in the m1 − (b − y) diagram. As is seen in Figure 12,
these stars match the calculated indices relatively well, although
there seems to be a tendency of the sensitivity of the m1 index to
metallicity to be exaggerated by the model fluxes for the hotter
stars. Also, it is clear from Figure 12 that the metal-rich stars lie
somewhat low, possibly suggesting that the zero-point of the m1
index as determined from Vega may be somewhat in error.
When studying the calculated indices of Lester, Gray &
Kurucz (1986) in Figure 11 we find that they reproduce the
observed sensitivity of the m1 index more successfully than
ours. However, these authors have transformed their calculated
colours to match a set of standard stars and have thus scaled the
amplitude of the m1 index by a correction factor to fit the obser-
vations. This is the probable reason for the closer agreement of
those calculations with observations. We note that our line list
is more complete, and that our treatment of the hydrogen line
broadening (affecting Hδ and thus the v band) is more accurate.
8
Onehag et al.: Calibration of Stromgren Photomtery
Fig. 6. Theoretical Hβ values versus effective temperature for models with different [Me/H] and log g plotted together with values
for the standard stars (with [Fe/H] within ±0.1 dex of the given [Me/H]). To guide the eye the theoretical log g curves for 2.0 and
4.5 are plotted as a dashed and a dotted line, respectively.
Table 2. The metallicity sensitivity, δ(m1)/δ[Me/H])(b−y), for
models with log g = 4.5 & 4.0.
(b − y)/[Me/H]
0.5−(−0.5)
−1−(−2)
−2−(−3)
log g = 4.5
0.25
0.30
0.35
0.40
0.45
0.50
0.55
log g = 4.0
0.25
0.30
0.35
0.40
0.45
0.50
0.55
0.124
0.145
0.149
0.139
0.136
0.176
0.233
0.125
0.143
0.148
0.137
0.125
0.145
0.198
0.031
0.042
0.056
0.098
0.135
0.146
0.127
0.030
0.041
0.059
0.081
0.097
0.100
0.093
0.008
0.013
0.028
0.026
0.032
0.048
0.070
0.010
0.014
0.020
0.031
0.052
0.076
0.089
What is then the reason for our discrepancy for early F stars?
We have compared the MARCS model fluxes with observed so-
lar and stellar fluxes from ground-based and space observations
(Edvardsson 2008, Edvardsson et al. 2008) and traced probably
significant discrepancies in the region 4000 Å – 5000 Å, with
empirical fluxes of the Sun and solar-type stars being some-
what smaller than model fluxes in the b band, while the blue-
violet fluxes from HST/STIS of the more metal-poor stars are
clearly in excess of the model fluxes in the violet-blue spectral
region. These departures in both the b and v band may conspire
to cause the discrepancy in calculated metallicity sensitivity. As
discussed by Edvardsson et al. (2008), 3D model simulations
suggest that these effects may be due to thermal inhomogeneities
in the stellar atmospheres. Other systematic errors in the models,
e.g. due to errors in opacities, line data and effects of departures
from LTE are probably less significant.
As a further test of our m1 indices we will now be guided
by the separate metallicity calibrations for F and G stars, re-
spectively, by Schuster & Nissen (1989, eq. 2 & 3, hereafter
S&N89). The derived F-star equation is based on the m1 and
(b − y) index and for the G-star equation the c1 index is also in-
cluded. We derive a calibration expression based upon the form
of these equations and using our theoretical colour grid. The
results, when applying this to the standard stars and the C06
sample, can be seen in Figures 13 and 14. For F-stars we find
standard deviations of derived [Me/H] values compared with lit-
Onehag et al.: Calibration of Stromgren Photomtery
9
Fig. 11. The m1 index versus (b − y) for MARCS and Lester, Gray & Kurucz (1986, LGK) model atmospheres. Arrows show the
effect of increasing the N abundances by a factor of 3.
erature values for the standard-star sample as given in Table 1.
We note, that for the standard sample, our theoretical calibration
tends to suggest lower metallicities than the adopted values for
a majority of the stars in the sample and increasing differences
with increasing metallicity so that the differences amount to typ-
ically 0.3 dex at [Me/H] = 0.0. The overall trend might indicate
a zero-point problem; a shift of all stars by 0.130 dex gives a
lower spread of the calculated metallicities compared to adapted
values, σ+0.130 = 0.217. This is however still not satisfactory in
comparison with the empirical equations which generally show
smaller spread and imply higher metallicities (see Table 1 and
Figure 13). A linear regression for the F-star calibration values
to the adopted values for the standard stars and the C06 sample
is plotted in Figure 13. The standard deviation for this line fit is
σ = 0.172 dex. When applying our calibration to the C06 sample
we see the same indications as for the standard sample, except
for the lower metallicities, where the theoretical calibration sug-
gest somewhat lower metallicities (∼0.05 dex) than derived in
C06. However, the 13 F-stars in this sample are too few to allow
any definite conclusions. When making use of the 266 F-stars in
the VF sample and remembering that this sample is highly biased
towards metal-rich dwarf stars ([Fe/H]> −0.2 & log g > 4.0) we
see the same trend as for the standard sample, i.e. our calibration
suggests lower metallicities than those given by VF05 (see also
Table 1).
For the G-stars we find similar standard deviations for our
standard sample with respect to literature values, see Table 1.
There is an overall tendency that our theoretical calibration im-
plies lower metallicities than listed in the literature for the metal-
rich standard stars. The increasing deviation with increasing
metallicity is of the same order as for the F-star calibration, i.e
≈ 0.3 dex at [Me/H] = 0.0. By adding 0.190 dex to all stars we
would obtain a lower spread, σ+0.190 = 0.181, which is of the
same order as the deviation shown by empirical equations (see
Table 1). A linear regression to the calculated metallicities for
the standard sample and C06 is plotted as a solid line in the fig-
ure (σ = 0.200). When using the 69 G-stars in the C06 sample,
the result for the standard sample is confirmed, the theoretical
calibration suggests lower metallicities for the higher metallic-
ity range ([Fe/H] > −0.5) than listed in literature. Applying the
calibrations to the 694 G-stars in the VF05 sample, we obtain
similar results as for the C06 sample.
For comparison the metallicities of our three comparison
samples are calculated with 3 different empirical calibrations.
The resulting standard deviations with respect to literature val-
ues can be seen in Table 1 and linear regressions when applied
to our standard sample can be seen in Figure 13 & 14 for F-stars
and G-stars, respectively.
We note in passing that Clem et al. (2004) also found dis-
crepancies between their calculated m1 indices and observations.
10
Onehag et al.: Calibration of Stromgren Photomtery
Fig. 12. The m1 index versus (b − y) for MARCS models plotted together with standard stars of different gravities in different panels
(log g for the stars within ±0.1 dex of the given log g). The metallicities of the stars are indicated by different symbols.
For their coolest models they had to apply upward corrections to
their calculated m1 indices of 0.1–0.3 mag. Our m1 indices de-
part considerably less from observations but some upward cor-
rections would be needed to fit the coolest stars in Figure 12.
The strong effects of microturbulence on the strengths of the
dominating lines in the v band makes it important to investi-
gate whether this could be the reason for the mis-match of the
m1 index in the theoretical calibration. In Table 3 we examine
the effects of changing the microturbulence parameter ξt. The
changes have been chosen to be larger for the lower-gravity mod-
els to take the larger parameter values usually obtained by spec-
troscopy for giants into account. We see that the effect of in-
creasing the microturbulence is generally larger for lower log g
and lower temperatures, as expected. When comparing to Figure
12 we see that this effect would indeed improve the situation by
shifting the theoretical curves and steepening the curves in the
low Teff end. Yet, the effects are considerably smaller than those
needed to eliminate the mis-match.
Another circumstance which might have some significance
as an explanation for the problems with the m1 sensitivity is the
effect of lines from the (0,1) band of the CN violet system in the
v band. In particular for the cooler giant stars, which may be af-
fected by the first dredge-up of CNO-processed material, the CN
lines may become stronger due to this; even if the carbon abun-
dances are reduced by CNO processing, the enhanced N abun-
Table 3. The change in the uvby indices when the microturbu-
lence parameter is increased
log g = 4.0
Teff
4500
5500
7000
4500
5500
7000
log g = 3.0
Teff
4500
5500
7000
4500
5500
7000
[Me/H]
0.00
0.00
0.00
−1.00
−1.00
−1.00
[Me/H]
0.00
0.00
0.00
−1.00
−1.00
−1.00
∆(ξt = 1.0
∆c1
0.0
0.003
0.004
0.006
0.006
0.004
∆(ξt = 1.0
∆c1
−0.016
0.004
0.001
0.015
0.017
0.006
→
∆m1
0.04
0.008
0.005
0.005
0.005
0.002
→
∆m1
0.023
0.025
0.014
0.018
0.013
0.005
ξt = 1.5)
∆(b − y)
0.011
0.005
0.001
0.004
0.002
0.00
ξt = 2.0)
∆(b − y)
0.032
0.014
0.002
0.012
0.003
0.0
dance (to which C is converted) makes the CN lines stronger.
Also the 13CN lines should be significantly enhanced due to the
production of 13C and N by CNO processing. We have explored
these effects by systematically increasing the N abundances by a
factor of 3, keeping the C abundance constant. This should lead
Onehag et al.: Calibration of Stromgren Photomtery
11
Fig. 7. Comparison of calibrations of (b − y) for dwarf stars with
individual values for the standard stars and C06 plotted. The
solid line represents a fit of the effective temperatures adopted
for the stars relative to the corresponding values obtained from
the theoretical calibration. Below that (shaded area and right
y-axis) corresponding linear regressions of the adopted effec-
tive temperatures relative to empirical calibrations (Alonso et al.
1996, Holmberg et al. 2007 and Ram´ırez & Mel´endez 2005) are
shown. The dotted line is a one-to-one line.
Fig. 9. Comparison of calibrations of Teff(Hβ) for dwarf stars
with individual values for the standard stars plotted. The solid
line represents a fit of the effective temperatures adopted for the
standard stars relative to the corresponding values obtained from
the theoretical calibration. Below that (shaded area and right y-
axis) a corresponding linear regression of the adopted effective
temperautres relative to an empirical calibration (Alonso et al.
1996) is shown. Note that the star located in the shaded area be-
longs to the theoretical calibration and should therefore be read
off at the left y-axis. The dotted line is a one-to-one line.
Fig. 8. Comparison of calibrations of (b − y) for giant stars with
individual values for the standard stars plotted. The solid line
represents a fit of the effective temperatures adopted for the stan-
dard stars relative to the corresponding values obtained from the
theoretical calibration. Below that (shaded area and right y-axis)
corresponding linear regressions of the adopted effective tem-
peratures relative to empirical calibrations (Alonso et al. 1999
and Ram´ırez & Mel´endez 2005) are shown. The dotted line is a
one-to-one line.
to an overestimate of the effect, except for possibly stars high-up
on the giant branch. As indicated in Figure 11 this only leads to
some effects for the more metal-rich giant stars and is not the
explanation for the mis-match discussed here.
Fig. 10. The effective temperatures calculated by the equations
based on Hβ and (b − y). Solid line theoretical fit, dashed line
empirical fit (Alonso et al. 1996). The dotted line is the one-to-
one relation.
7. Surface-gravity calibration
The c1 ≡ (u −v) −(v −y) index is designed to measure the Balmer
discontinuity which is a temperature indicator for B- and A-
type stars and a surface-gravity indicator for the late-type stars.
Figure 15 shows its behaviour with changing parameters of the
models. Obviously, it works nicely as a gravity criterion, with
some dependence on metallicity for the most metal-rich stars,
which has been a disputed issue in earlier calibration work. We
note that for dwarfs cooler than the Sun it seems not very useful
as a gravity measure, while for bright giants, and not the least
metal-poor ones, it should work down to effective temperatures
12
Onehag et al.: Calibration of Stromgren Photomtery
where 0.22 ≤ (b − y)0 ≤ 0.47, (b − y)0 being the dereddened
(b − y), and δc0 is the difference of a dereddened c1 index and
a standard sequence with MV,ZAMS at a given Hβ. From this one
may estimate the empirical sensitivity δc1/δ log g to be approxi-
mately proportional to 2.5/ f . We thus obtain the following em-
pirical values of (δc1/δ log g; (b − y)) for [Fe/H] = 0: (0.21;0.30),
(0.16;0.40) and (0.15;0.47). From Figure 15 and Table 4 we
measure the corresponding theoretical δc1/δ log g values to be
0.19, 0.08 and 0.02, respectively. Thus, we find that the empirical
gravity sensitivity of the c1 index is well reproduced by the hotter
models while it becomes underestimated for the cooler ones. We
have also compared the calculated c1 indices with the observed
ones for our samples of standard stars and find a good agreement
for the hotter stars (cf. Fig. 16) while for the cooler stars there
is a severe mismatch. This is in itself not very remarkable – the
total line blocking in the ultraviolet spectra of the cooler stars is
considerably greater than 70%, and it is to be expected that this
will not be very accurately described by the model spectra. The
fluxes in the u and v bands are highly sensitive to other param-
eters, such as microturbulence, and in particular CN abundance
for the cooler giants, and these are known to vary systematically
with gravity. In fact, as is indicated in Figure 15, the CN line
strengthening which is expected for the red giants improves the
fit to observed c1 indices for the red giant models considerably.
Clem et al. (2004) applied semiempirical corrections also to
their calculated (u − v) and (v − b) colours in order to fit obser-
vations. The resulting effect on their c1 indices is typically less
than 0.1 mag while we would need a downward correction of our
c1 indices of about 0.2 mag for the coolest models to fit (cf. Fig.
16) the metal poor stars and even more for the metal-rich ones.
Table 4. The sensitivity to surface gravity, (δc1/δ log g)(b−y), for
models with [Me/H] = 0.5 and 0.0
(b − y)/ log g
[Me/H] = 0.5
0.25
0.30
0.35
0.40
0.45
0.50
0.55
[Me/H] = 0.0
0.25
0.30
0.35
0.40
0.45
0.50
0.55
2.0–3.0
3.0–4.0
4.0–4.5
0.231
0.213
0.183
0.148
0.114
0.083
0.056
0.247
0.236
0.215
0.182
0.144
0.106
0.065
0.216
0.187
0.148
0.112
0.086
0.071
0.062
0.230
0.205
0.169
0.126
0.081
0.040
0.002
0.207
0.165
0.122
0.091
0.074
0.060
0.049
0.223
0.185
0.131
0.082
0.035
–0.006
–0.026
In a recent article, Twarog et al. (2007) discuss the metal-
licity dependence of the c1 index for disk stars ([Fe/H] ≥ –1.00)
and conclude that the metallicity sensitivity of the index gener-
ally has been underestimated. An interesting test would there-
fore be to follow their recipe, i.e. divide our model grid into
three groups (hot, warm and cold), and check whether or not our
synthetic colours show the same behaviour. For the hot mod-
els ((b − y) < 0.43) we find the same strong metallicity de-
pendence of the m1 index and the same lack of sensitivity of
the c1 index as is found by Twarog et al. For warm models
(0.43 ≤ (b − y) < 0.50) we do find, like Twarog et al., a strong
Fig. 13. The [Me/H] calibration for F stars from the standard
sample and Casagrande et al. (2006, C06). The solid line repre-
sents a linear regression of the effective temperatures adopted for
the stars relative to the corresponding values obtained from the
theoretical calibration. Below that (shaded area and right y-axis)
corresponding linear regressions of the adopted effective tem-
perautres relative to empirical calibrations (Schuster & Nissen
1989, Holmberg et al. 2007 and Ram´ırez & Mel´endez 2005) are
shown. The dotted line is a one-to-one line.
Fig. 14. The [Me/H] calibration for G stars. For plot description
see Figure 13. Note that empirical calibrations of HNA07 and
RM05 overlap and could therefore be difficult to distinguish.
around 5000 K. In Figure 15 we have also plotted the indices
calculated by Lester, Gray & Kurucz (1986). In view of the dif-
ferences in line data and hydrogen-line theory we find the agree-
ment satisfactory.
Analogously with our treatment of the metallicity depen-
dence of the m1 index, we have measured the quantity δc1/δ log g
at constant (b − y) as a measure of the gravity sensitivity of c1.
Again we have to turn back to earlier calibrations to find corre-
sponding direct empirical measures. Thus, Schuster & Nissen
(1989) have elaborated the methodology of Crawford (1975,
1979) and write
MV = MV,ZAMS − f · δc0
f ≡ 9.0 + 38.5 · ((b − y)0 − 0.22),
(2)
Onehag et al.: Calibration of Stromgren Photomtery
13
Fig. 15. The c1 index versus (b − y) for MARCS and Lester, Gray & Kurucz (1986, LGK) model atmospheres with different log g
values indicated at corresponding curves in the figure and different [Me/H] plotted in different panels. Arrows show the effect of
increasing the N abundances by a factor of 3.
c1 sensitivity to metallicity, but likewise a relatively strong de-
pendence of the m1 index in contrast to the result of Twarog
et al. For cool stars ((b − y) ≥ 0.50), we separate unevolved
dwarfs from subgiants and giants by making use of the defined
LC (≡ c1 − 2.0m1 + 3.0(b − y) − 0.15) calibration in Twarog et al.
The models follow the same trends and LC separates the more
metal-poor models ([Fe/H] = –0.5 & –1.0) but does not succed
in separating the metal-rich ones ([Fe/H] ≥ 0.0), as is also found
by Twarog et al. It should however be noted that when apply-
ing the metallicity calibrations for hot and warm stars as defined
by Twarog et al. on our standard sample, we do not reach any
higher accuracy in reproducing the literature values of [Fe/H]
than when using e.g. S&N89.
7.1. Applications
In the area of application of the uvby photometry, some particu-
lar questions have been of special interest to us: to which extent
can the photometry be used for determining gravities and tem-
peratures for metal-poor stars, and how sensitive is the synthetic
c1 index to certain elements such as nitrogen.
Following Clem et al. (2004), we shall test our model colours
versus observed colours for globular clusters. Unlike the ap-
proach of Clem et al., our main emphasis lies on exploring the
possibilities and shortcomings of our theoretical model colours
sooner than correcting them semiempirically to establish new
calibrations.
7.1.1. M92
As a check on the capabilities of the c1 index for determining
gravities, we have tested the model indices relative to the ob-
served uvby photometry for the extreme Pop II globular clus-
ter M92. Adopting a metallicity of [Fe/H]= –2.22, a distance of
8.3 kpc, and a reddening of E(B − V) = 0.023 (Grundahl et al.
2000), we have converted the observed y − (b − y) diagram of
Grundahl et al. (private communication) to the fundamental pa-
rameters along the evolutionary sequence past the turnoff point,
using Mv, (b − y), Teff (determined by MARCS models) and as-
suming a mass of 0.8M⊙ for the giants; c.f. the Teff/ log g values
given in Table 5. The bolometric corrections used were taken
from VandenBerg & Clem (2003). Next, we have calculated val-
ues for the uvby colours using MARCS models for these param-
eter values, to compare with the directly observed dereddened
indices for M92 stars, according to Grundahl et al. As is seen in
Table 5, the observed colours are well reproduced by the models.
We have also calculated the derivatives δc1/δ log g along the gi-
ant branch and find typical values ranging from –0.2 to –0.1, thus
14
Onehag et al.: Calibration of Stromgren Photomtery
Fig. 16. The c1 index versus (b−y) for MARCS models with different [Me/H] and log g plotted together with values for standard stars
and Casagrande et al. (2006, C06) stars (marked as asterisks). The [Fe/H] intervals for the standard and C06 stars are [Fe/H] ±0.1 dex
and [Fe/H] ±0.2 dex, respectively.
agreeing within 20 % between the observations and the calcula-
tions. This suggests that we can use the uvby system to estimate
gravities for metal-poor giant stars.
Table 5. Observed and calculated colours for log g values along
the sub-giant - giant branch for M92 and MARCS, respectively.
The tabulated colours are mean colour values of stars close to
the given fundamental parameters.
log g
1.58
2.05
2.51
2.99
3.50
3.72
4.00
Teff
4682
4915
5125
5306
5525
5836
6302
(b − y)0
0.584
0.522
0.473
0.437
0.400
0.355
0.299
M92
c0
0.516
0.414
0.374
0.307
0.281
0.332
0.383
m0
0.109
0.088
0.075
0.069
0.057
0.047
0.064
MARCS
c1
0.519
0.440
0.384
0.328
0.295
0.335
0.411
m1
0.117
0.088
0.067
0.052
0.047
0.045
0.055
7.1.2. NGC 6397
As a second test, effective temperatures for stars in the metal-
poor globular cluster NGC 6397 were derived. This is an ap-
plication of current interest as Korn et al. (2007) have re-
cently claimed that they have identified abundance trends be-
tween groups of cluster stars which they suggest are caused by
atomic diffusion. Tracing abundance differences between differ-
ent groups, at the turnoff, on the subgiant branch and on the
lower and higher red-giant branch (TOP, SGB, lRGB, hRGB),
requires well determined stellar parameters, and most impor-
tantly well measured effective temperature differences between
the groups. Korn et al. derive effective temperatures with a
photometric and spectroscopic approach and they find consis-
tent temperatures. One potential problem, however, is the fact
that two different photometric calibrations needed to be applied
(Alonso et al. 1996, 1999) which may introduce systematic er-
rors. For our theoretical calibration we derive effective temper-
atures for each of the 18 stars in the four different groups, by
using log g values and dereddened colour indices (b − y) and
(v − y) listed in Korn et al. and adopting their [Fe/H] of –2.0.
Mean values of the effective temperatures for the four groups are
shown in Table 6. The results indicate that our effective temper-
atures are 50 K to 100 K hotter than obtained from the empirical
Onehag et al.: Calibration of Stromgren Photomtery
15
Table 6. Derived effective temperatures from empirical calibra-
tions of photometry ((b − y) and (v − y)), and spectroscopy (Korn
et al. 2007), as well as our theoretical calibration, for groups of
stars in the metal-poor cluster NGC 6397. The temperatures are
mean values for a number of stars/models in each group.
Teff [K]
hRGB
lRGB
SGB
TOP
∆(TOP-RBG)
∆(TOP-SGB)
Phot.
(b − y)
5121
5455
5797
6229
1108
433
(v − y)
5132
5408
5824
6214
1081
390
Spec. MARCS
(b − y)
5225
5541
5839
6288
1063
449
5130
5456
5805
6254
1124
449
(v − y)
5226
5535
5853
6281
1055
428
Table 7. Calculated δcy/δ[N/Fe] for giant models as a compar-
ison to the observed cy sensitivity of NGC 6752 (Yong et al.
(2008))
Teff
4749
4829
4841
4904
4950
log g
1.95
2.10
2.15
2.33
2.42
0.0−−0.75
0.75−−1.5
0.03
0.04
0.04
0.04
0.04
0.00
0.02
0.02
0.04
0.04
calibrations. More importantly, the temperature differences be-
tween TOP, SGB and hRGB stars are close to (even somewhat
smaller than) the ones obtained from the empirical calibrations
by Korn et al. This strengthens the claim of Korn et al. that abun-
dance differences in Li, Mg and Fe, reflecting atomic diffusion,
are present at a significant level.
7.1.3. NGC 6752
In another test, the nitrogen sensitivity of of the c1 index was
examined. In a recent paper, Yong et al. (2008), measure the
nitrogen content of giants in the globular cluster NGC 6752
([Fe/H] ∼ –1.6) based on high-resolution observations of NH.
The authors find a strong correlation between nitrogen content
and the c1 index and establish the cy index (c1 − (b − y)) that
shows a close to linear correlation with respect to nitrogen con-
tent. To examine the synthetic sensitivity of this cy index, a set of
models and high resolution spectra were calculated (Teff = 4749,
4829, 4841, 4904, 4950 K, log g = 1.95, 2.10, 2.15, 2.33, 2.42,
[Me/H] = –1.50) based on a selection of stars discussed by Yong
et al. The [N/Fe] content was set to vary as 0.00, 0.75 and 1.50
for each of the models in order to cover a great part of the [N/Fe]
span in NGC 6752 found by Yong et al. As a result we find that
our synthetic cy index is sensitive to the nitrogen content of the
model, even though the sensitivity δcy/δ[N/Fe] of the models is
somewhat smaller, ∼ 0.03 − 0.04 (see Table 7), compared to the
sensitivity Yong et al. find for their giant stars, ∼0.06, as is esti-
mated from the best-fit straight line in Yong et al. Figure 7.
8. Conclusions and recommendations
We have explored the possibilities and shortcomings of syn-
thetic uvby−Hβ photometry based on new MARCS model atmo-
spheres. In general a good agreement with empirical calibrations
of this photometric system is found. However, a number of sys-
tematic deviations between theory and observations also become
apparent. The temperature sensitivity of the (b − y) colour (i.e.
δ(b − y)/δTeff) seems marginally larger for the calculated colours
than is found when using infrared-flux-method determinations
of temperatures. A similar, and even somewhat larger, difference
occurs for temperatures based on the Hβ index, when compared
with the empirical scale. The (b − y) calibration for Pop II stars
supports Korn et al. (2007), who claim the signatures of atomic
diffusion in the metal-poor globular cluster NGC 6397.
For the metallicity index m1 the theoretical sensitivity
δm1/δ[Me/H] is somewhat larger than the empirical one that
is based on spectroscopic [Fe/H]-determinations. For the grav-
ity sensitivity of the c1 Balmer discontinuity measure, we find
a reasonably good agreement with observations for stars hot-
ter than the Sun, where the Balmer discontinuity is significant.
Considerable problems remain for the cooler stars, although the
model calibration works well for Pop II giants.
One might ask whether these problems may be solved when
even more detailed atomic and molecular line data become avail-
able to feed into model atmospheres. This is possible but it is
also possible that thermal inhomogenities as generated by con-
vection and possibly also non-LTE effects, contribute to the dif-
ference between synthetic and observed values of (b − y) and m1.
Concerning the problems in reproducing the observed c1 indices
of cooler stars, these may well be due to failures in line-data,
although thermal inhomogeneities in the atmospheres may also
here turn out to be the important effect. One should also real-
ize that there are still residual problems associated with differ-
ences and uncertainties in the observed and adopted Stromgren
passbands (Manfroid & Sterken, 1987). The current study illus-
trates the problems in synthetic photometry in the visual wave-
length regions. Continued efforts along such lines must be com-
plemented with more detailed studies of the shortcoming of clas-
sical model atmospheres and the replacement of those, also in
large-scale calibration efforts, by physically more realistic mod-
els.
Acknowledgements. Anders Eriksson is thanked for a major contribution at the
startup of this project. Remo Collet is thanked for discussions on 3D model at-
mospheres and Ulrike Heiter and Andreas Korn for valuable suggestions and
comments on the manuscript. Ana Garc´ıa P´erez is thanked for the calculations
of reddening.
References
Adelman, S.J., Gulliver, A.F. 1990, ApJ 348, 712
Ali, A. W., & Griem, H. R. 1966, Phys. Rev. 144, 366
Alonso, A., Arribas, S., Martinez-Roger, C. 1996, A&A 313, 873
Alonso, A., Arribas, S., Martinez-Roger, C. 1999, A&AS 140, 261
Ardeberg, A., Lindgren H. 1981, RMxA&A 6, 173
Asplund, M., Grevess, N., Sauval, A.J. 2005, ASPC 336, 25
Barklem, P. S., & Piskunov, N. 2003, in Modelling of Stellar Atmospheres, ed.
N. Piskunov, W. W. Weiss, D. F. Gray, Proc., IAU Symp. 210
Baschek, B. 1960, Zeitschrift Ap 50, 296
Bell, R.A. 1970, MNRAS 148, 25
Bell, R.A. 1971, MNRAS 154, 343
Bell, R.A., Parsons, S.B. 1974, MNSAS 169, 71
Bessell, M.S. 2005, A&A Rev., 43, 293
Casagrande, L., Portinari, L., Flynn, C. 2006, MNRAS 373, 13
Castelli, F., Kurucz, R.L. 2006, A&A 454, 333
Christlieb, N., Gustafsson, B., Korn, A.J. et al. 2004, ApJ 603, 708
Clem, J.L., VandenBerg, D.A., Grundahl, F., Bell, R., 2004, AJ 127, 1227
Crawford, D.L. 1958, ApJ 128, 185
Crawford, D.L. 1966, IAU Symp. 24, 170
Crawford, D.L. 1975, AJ 80, 955
Crawford, D.L. 1979, AJ 84, 1858
Crawford, D.L., Mander, J. 1966, AJ 71, 114
Crawford, D.L., Barnes, J.V. 1970, AJ 75 978
Edvardsson, B. 2008, in "A stellar journey. A symposium in celebration of Bengt
Gustafsson's 65th birthday", Physica Scripta, in press
Edvardsson, B., Eriksson, K., Gustafsson, B. et al. 2008, A&A, in preparation
Gigas, D. 1988, A&A 192, 264
16
Onehag et al.: Calibration of Stromgren Photomtery
Grevesse, N., Sauval, A.J. 1998, SSRv 85, 161
Griem, H. R. 1960, ApJ 132, 883
Grundahl, F., VandenBerg, D.A., Bell, R.A., Andersen, M.I., Stetson, P.B. 2000,
A.J. 120, 1884
Gulliver, A.F., Hill, G., Adelman, S.J. 1994, ApJ 429, L81
Gustafsson, B., Bell, R.A. 1979, A&A 74, 313
Gustafsson, B., Edvardsson B., Eriksson K., Jørgensen U.G., Plez B. 2003, Conf.
Ser. Vol. 288, p.331
Gustafsson, B., Edvardsson B., Eriksson K., et al. 2008, A&A in press,
arXiv:0805.0554v1
Gustafsson, B., Nissen, P.E. 1972, A&A 19, 291
Hauck, B. & Mermilliod, M. 1998, A&AS 129, 431
Hakkila, J., Myers, J.M., Stidham, B.J., & Hartmann, D.H. 1997, AJ 114, 2043
Heiter, U., Weiss, W.W., Paunzen, E. 2002, A&A 381, 971
Helt, B.E., Franco G.A.P., Florentin Nielsen R., 1987, in: ESO Workshop on the
SN1987A, Danziger I.J.(ed.), Garching, p. 89
Hill, G.M. 1995, A&A 294, 536
Hill, G., Gulliver, A.F., Adelman, S.J. 2004, IAU Symposium 224, 35
Hoffleit, D., Warren, W.H. Jr. 1995, VizieR On-line Data Catalog: V/50,
SIMBAD
Holmberg, J., Nordstrom, B., Andersen, J. 2007, A&A 475, 519
Ilijic, S., Rosandic, M., Dominis, E., et al. 1998, CoSka 27, 467
Jonsell, K., Edvardsson, B., Gustafsson, B. et al. 2005, A&A 440, 321
Korn, A.J., Grundahl, F., Richard, O., Mashonkina, L., Barklem, P.S., Collet, R.,
Gustafsson, B., Piskunov, N. 2007, ApJ 671, 402
Kupka, F., Piskunov, N. E., Ryabchikova, T.A., et al. 1999, A&AS, 138,119
Kurucz, R.L. 1995, Kurucz CD-ROM No.15, Cambridge, Mass., Smithsonian
Astrophysical Observatory
Lester, J.B., Gray, R.O., Kurucz, R.L. 1986, ApJS 61, 509
Lejeune, T., Lastennet, E., Westera, P., Buser, R. 1999, ASP Conf Proc. 192, 207
Manfoid, J. & Sterken, C. 1987, A&AS 71, 539
Nissen, P.E. 1970, A&A
Nissen, P.E. 1981, A&A 97, 145
Nissen, P.E. 1988, A&A 199, 146
Nissen, P.E., Gustafsson, B. 1978, in Astronomical Papers Dedicated to Bengt
Stromgren, Copenhagen 1978, p. 43
Nordstrom, B., Mayor, M., Andersen, J. et al. 2004, A&A 418, 989
Olsen, E.H. 1983, A&AS 54, 55
Olsen, E.H. 1984, A&A 57, 443
Olsen, E.H. 1988, A&A 189, 173
Paunzen, E., Iliev, I.K., Kamp, I. et al. 2002, MNRAS 336, 1030
Plez, B., Masseron, T., Van Eck, S. et al. 2008, ASP, Conf. Ser., in presss
Querci, F., Querci, M., Kunde, V. 1971, A&A 31, 265
Ram´ırez, I., Mel´endez, J. 2005, ApJ 626, 465
Ram´ırez, I., Mel´endez, J. 2005, ApJ 626, 446
Relyea, L.J., Kurucz, R.L. 1978, ApJS 37, 45
Schuster, W.J., Nissen, P.E. 1988, A&AS 73, 225
Schuster, W.J., Nissen, P.E. 1989, A&A 221, 65
Stromgren, B. 1963, QJRAS 4, 8
Stromgren, B. 1964, Astrophys. Norveg. 9, 333
Stromgren, B. 1966, Ann Rev Astr Ap 4, 433
Twarog, B.A., Vargas, L.C., Anthony-Twarog, B.J. 2007, AJ 134, 1777
VandenBerg, D.A., Clem, J.L. 2003, AJ 126, 778
Valenti, J.A., Fischer, D.A. 2005, ApJS 159, 141
Vidal, C. R., Cooper, J., & Smith, E. W. 1973, ApJS 25, 37
Wallerstein, G. 1962, ApJS 6, 407
Yong, D., Grundahl, F., Johnson, J.A., Asplund, M. 2008, ApJ 684, 1159
Appendix A: Calibrations
In the following Teff and [Fe/H] calibrations we have systematically used
the form, and in most cases the index limits for the validity of the calibra-
tions, according to Alonso et al. (1996,1999) and Schuster & Nissen (1989),
respectively. The coefficients have then been calculated by a least square
method to optimize the fit to the model indices.
A.1. TheoreticalTeff calibrationfordwarfstars, (b − y) and
Hβ
Θeff ≡ 5040/Teff
Θeff(b − y)dwarf = 0.466 + 1.025(b − y) − 0.123(b − y)2
+0.212(b − y)c1 − 0.051(b − y)[Me/H]
+0.003[Me/H] − 0.005[Me/H]2
The models, upon which the calibration is based, are selected to follow
the restrictions given for the empirical equation (A96, Eq. 9. Restrictions
adopted from Fig. 11a): −3.00 ≤ [Me/H] ≤ 0.50; 0.70 ≤ (b − y) ≤ 0.20;
0.10 ≤ c1 ≤ 0.55; 4.00 ≤ log g (set as limit for dwarf stars).
Θeff(Hβ)dwarf = 28.60 − 19.79Hβ + 3.504Hβ2 + 0.422Hβ[Me/H]
−1.068[Me/H] + 0.002[Me/H]2
In analogy with the theoretical calibration based on (b − y), the limits
determining which models to use, were set to be equal to the ones for
empirical calibration (A96, Eq. 10 and its following applicable ranges):
2.44 ≤ H β ≤ 2.74 for +0.5 ≥ [Me/H] > −0.5; 2.50 ≤ H β ≤ 2.70 for
−0.5 ≥ [Me/H] > −1.5; 2.50 ≤ H β ≤ 2.63 for −1.5 ≥ [Me/H] > −2.5;
2.51 ≤ H β ≤ 2.62 for −2.5 ≥ [Me/H] > −3.5; 4.0 ≤ log g (set as a limit
for dwarf stars).
A.2. TheoreticalTeff calibrationforgiants,(b − y)
Θeff(b − y)giant = a0 + a1(b − y) + a2(b − y)2 +
a3(b − y)[Me/H] + a4[Me/H] + a5[Me/H]2
Table A.1. The derived coefficients for the theoretical calibration
in the intervalls I: 0.15 ≤ (b − y) ≤ 0.424, II: 0.424 ≤ (b − y) ≤
0.712 & −5.0 ≤ [Me/H] ≤ −0.5 and III: 0.428 ≤ (b − y) ≤ 0.794
& −0.5 < [Me/H] ≤ 0.5
a0
0.6732
a1
0.0859
a2
1.1455
a3
-1.080e-2
a4
-0.132e-2
a5
-0.082e-2
0.1983
2.0931
-0.9978
4.709e-2
-2.66e-2
-0.01e-2
I:
II:
III:
0.4522
1.1745
-0.3093
-0.1693
2.165e-2
-1.679e-2
Here we deviate from the same restriction limits for the calibrations as pre-
sented in Alonso et al (1999, Table 2 and 3). New theoretical limits were set
due to model restrictions: (b − y) ≤ 0.424 and (b − y) ≥ 0.424 (cf. Alonso
et al 1999, Eq. 14: (b − y) ≤ 0.55 and Eq. 15: (b − y) ≥ 0.50). The log g
limit was set to 1.5 ≤ log g ≤ 3.5.
A.3. TheoreticalmetallicitycalibrationforG-stars
The models, upon which the metallicity calibrations for G and F stars
are based, are selected to follow the restrictions given for the empirical
equations of S&N89, Eq. 3. and 2., respectively. The G-stars limits cover
the following intervals: −2.6 ≤ [Me/H] ≤ 0.4; 0.37 ≤ (b − y) ≤ 0.59;
0.03 ≤ c1 ≤ 0.57; 3.00 ≤ log g (set as limit for log g).
[Me/H]G = −2.796 + 39.21m1 − 88.97m2
1 − 73.43m1(b − y)
+181.4m2
1(b − y) + [27.03m1 − 1.220c1 − 41.42m2
1]c1
Onehag et al.: Calibration of Stromgren Photomtery
17
A.4. TheoreticalmetallicitycalibrationforF-stars
F-star limits are given by: −3.5 ≤ [Me/H] ≤ 0.2; 0.22 ≤ (b − y) ≤ 0.38;
0.17 ≤ c1 ≤ 0.58; 3.00 ≤ log g (set as limit for log g).
[Me/H]F = 1.850 − 34.21m1 + 105.34m1(b − y)
1(b − y) − 242.4m1(b − y)2
+179.8m2
+[2.757 − 20.38m1 + 0.2777(b − y)] log(m1 − c3)
c3 = 0.4462 − 2.233(b − y) + 2.885(b − y)2
|
astro-ph/0612223 | 1 | 0612 | 2006-12-08T15:27:58 | Deep imaging survey of the environment of Alpha Centauri - II. CCD imaging with the NTT-SUSI2 camera | [
"astro-ph"
] | Context: The nearby pair of solar-type stars Alpha Centauri is a favorable target for an imaging search for extrasolar planets. Indications exist that the gravitational mass of Alpha Cen B could be higher than its modeled mass, the difference being consistent with a substellar companion of a few tens of Jupiter masses. However, Alpha Centauri usually appears in star catalogues surrounded by a large void area, due to the strong diffused light. Aims: We searched for faint comoving companions to Alpha Cen located at angular distances of the order of a few tens of arcseconds, up to 2-3 arcmin. As a secondary objective, we built a catalogue of the detected background sources. Methods: In order to complement our adaptive optics search at small angular distances (Paper I), we used atmosphere limited CCD imaging from the NTT-SUSI2 instrument in the Bessel V, R, I, and Z bands. Results: We present the results of our search in the form of a catalogue of the detected objects inside a 5.5 arcmin box around this star. A total of 4313 sources down to mV~24 and mI~22 were detected from this wide-field survey. We extracted the infrared photometry of part of the detected sources from archive images of the 2MASS survey (JHK bands). We investigate briefly the nature of the detected sources, many of them presenting extremely red color indices (V-K > 14). Conclusions: We did not detect any companion to Alpha Centauri between 100 and 300 AU, down to a maximum mass of ~15 times Jupiter. We also mostly exclude the presence of a companion more massive than 30 MJup between 50 and 100 AU. | astro-ph | astro-ph | Astronomy&Astrophysicsmanuscript no. 6432
December 1, 2018
c(cid:13) ESO 2018
6
0
0
2
c
e
D
8
1
v
3
2
2
2
1
6
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
Deep imaging survey of the environment of α Centauri
(ResearchNote)
II. CCD imaging with the NTT-SUSI2 camera
P. Kervella1 and F. Th´evenin2
1 LESIA, UMR 8109, Observatoire de Paris-Meudon, 5, place Jules Janssen, F-92195 Meudon Cedex, France
2 Laboratoire Cassiop´ee, Observatoire de la Cote d'Azur, BP 4229, F-06304 Nice Cedex 4, France
Received ; Accepted
ABSTRACT
Context. Thanks to its proximity, α Centauri is an outstanding target for an imaging search for extrasolar planets.
Aims.We searched for faint comoving companions to α Cen located at angular distances of a few tens of arcseconds, up to 2-3 arcmin.
Methods. We obtained CCD images from the NTT-SUSI2 instrument in the Bessel V, R, I, and Z bands, and archive data from 2MASS.
Results. We present a catalogue of the detected objects inside a 5.5 arcmin box around this star. A total of 4313 sources down to mV ≈ 24 and
mI ≈ 22 were detected in the SUSI2 images. The infrared photometry of part of these sources has been extracted from the 2MASS images.
Conclusions. No comoving companion to α Centauri were detected between 100 and 300 AU, down to a maximum mass of ≈15 times Jupiter.
We also mostly exclude the presence of a companion more massive than 30 MJ between 50 and 100 AU.
Key words. Stars: individual: α Cen, planetary systems, Stars: imaging, solar neighbourhood, Astronomical data bases: miscellaneous
1. Introduction
Our closest stellar neighbour, the α Cen visual triple star (d =
1.34 pc), is an extremely attractive target for extrasolar planet
detection. The main components α Cen A (HD 128620) and
B (HD 128621) are G2V and K1V solar-like stars, while the
third member is the red dwarf Proxima (M5.5V). In all imag-
ing planet searches, the main difficulty is to retrieve the plane-
tary signal in the bright diffuse halo from the star. As discussed
in Paper I (Kervella et al. 2006), there are indications that the
gravitational mass of α Cen B could be higher than its modeled
mass. For these reasons, we engaged in a search for compan-
ions to α Cen at two different angular scales: a few arcseconds,
using adaptive optics (Paper I), and a few tens of arcseconds,
using CCD imaging (present work). To our knowledge, this is
the first wide-field search for companions to α Cen.
2. Observations
Very close to the two stars, within a radius of about 20", adap-
tive optics imaging allows to reach the best sensitivity. At dis-
tances of more than 20", the diffused light is less of a problem,
and atmosphere limited imaging is most cost-effective solution.
We thus observed α Cen using ESO's 3.6 m NTT, equipped
Send offprint requests to: P. Kervella
Correspondence to: [email protected]
with the SUperb Seeing Imager 2 (SUSI2). This instrument is
based on a mosaic of two 2k×4k thinned EEV CCDs (15 µm
pixels). In order to avoid an heavy saturation of the detector,
we positioned the α Cen pair aligned within the 8 arcsec gap be-
tween the two detectors. The Nasmyth adapter of the telescope
was rotated to align the two stars with the direction of the gap.
The pixel scale of SUSI2 is 0.0805 "/pixel, therefore providing
an excellent sampling of the point spread function. We obtained
a series of exposures through four filters, Bessel V, R, I, and Z1,
in order to be able to confirm potential companions based on
their colors. The individual exposures times were short, typi-
cally 30 to 60 s, to limit the saturation of the detector to a small
area around α Cen. The journal of the SUSI2 observations is
presented in Table 1. The total shutter open time obtained on
α Cen with SUSI2 in all bands reaches almost 4 hours.
In order to obtain infrared photometry for part of the
sources detected in the SUSI2 images, we retrieved from
the IPAC archive2 the FITS images from the 2MASS survey
(Skrutskie et al. 2006) covering the field of view of our SUSI2
images.The selected images were all obtained in early 2000.
1 The transmission curves of the SUSI2 filters are available from
http://www.ls.eso.org/lasilla/sciops/ntt/susi/docs/SUSIfilters.html
2 http://irsa.ipac.caltech.edu/
2
P. Kervella & F. Th´evenin: CCD survey of the environment of α Centauri (RN)
Table 1. Log of the SUSI2 observations of α Cen. The indi-
cated UT time corresponds to the middle of the exposures, θ is
the FWHM of the point spread function of the composite im-
ages, and AM is the airmass. The observations of 26 Feb 2004
were excluded due to poor seeing.
Date
2004-02-25
2004-02-26
2004-02-29
2004-02-29
2004-04-02
2004-04-02
2004-04-02
2006-04-11
2006-08-17
UT
6:40
7:48
6:31
7:54
2:55
4:21
5:49
8:29
1:22
Filter
Total exp. (s)
R
V
I
V
Z
V
I
I
I
1185
555
2010
1305
2360
1305
2010
1649
1649
θ (") AM
0.74
1.30
1.20
2.67
1.29
0.81
1.19
0.84
0.89
1.51
1.29
1.01
1.19
0.97
1.30
1.01
0.65
1.49
3. Data reduction
3.1. Rawdataprocessingandastrometry
We applied the standard CCD image processing steps to our
SUSI2 images, using IRAF: bias subtraction, flat-fielding, bad
pixel masking. The 2MASS images are available fully cali-
brated from the IPAC archive. The resulting SUSI2 band image
is presented in Fig. 1, where the VRI images have been mapped
to the blue, green and red colors. The lower ≈10% area of this
image is shown in grey levels as only two colors are avail-
able. The astrometric calibration of our SUSI2 images has been
obtained by matching the coordinates of the brightest sources
with their counterparts in the 2MASS images. The astromet-
ric accuracy of the 2MASS images is about 0.1" (Skrutskie et
al. 2006). Due to the presence of stronger diffused light in the
visible, we estimate our astrometric uncertainty at 0.2" over
the SUSI2 field, therefore this astrometric calibration method
is sufficient for our purpose.
3.2. Sourcesextraction
The SUSI2 sources were extracted using the SExtractor tool
(Bertin & Arnouts 1996), completed by a manual selection of
the sources close to α Cen. We applied the same hybrid selec-
tion procedure on the 2MASS images. We then obtained the
photometry of the sources using IRAF and 12 pixel apertures
for SUSI2 (1" radius). This relatively small aperture is justified
by the variable background from the diffused light of α Cen and
the relative crowding of the field. For the 2MASS images, we
used 3" apertures. The SUSI2 airmass coefficients were taken
as the average values for La Silla (ESO User's Manual 1993,
see also Burki et al. 1995): kV = 0.11, kR = 0.07, kI = 0.02,
and kZ = 0.01 mag/airmass. In all cases, the airmass correc-
tions are negligible, with values below 0.04 mag. The aperture
corrections for SUSI2 were determined on a bright, isolated
star located relatively far from α Cen to ensure a flat sky back-
ground. They were found to range between 0.20 and 0.42 mag
depending on the color and the observation epoch. The pho-
tometric zero points were taken from ESO's routine instrument
monitoring program for SUSI2, and from the image headers for
2MASS. Considering the difficulty of estimating accurately the
sky background level, we chose to add a systematic ±0.20 mag
uncertainty to the SUSI2 magnitudes. For the 2MASS images,
we applied the aperture correction recommended by IPAC for
a 3" aperture, i.e. 0.05 ± 0.02 mag in all three JHK bands.
The catalogue of the objects detected in the SUSI2 and
2MASS images is available in electronic format3. A total of
4313 sources were identified on the SUSI2 images, with 391
recovered in the 2MASS images (9.1%). Among the detected
background sources, several objects present extreme color in-
dices. We checked that their presence is expected based on
Galactic models (Robin et al. 2003).
4. Sensitivity and companion limits
To define our practical limiting magnitudes, we chose to con-
sider the median magnitude of the detected sources. This def-
inition has the advantage to give an empirical, statistically
meaningful definition of the sensitivity, that can be expressed
as a function of the distance to the two bright stars by com-
puting the median within angular distance bins. We computed
the median magnitude of the detected objects in the SUSI2
VRIZ bands for four angular distance bins: 23 −50", 50 −100",
100−150", 150−200" and 200−236". In all bands, the limiting
magnitudes (as defined above) at large angular distances are in
the 20-22 range, with the faintest objects at magnitudes of 23-
25. The limiting magnitudes decrease to 16-18 for an augular
separation of 35". For the 2MASS images, the limiting mag-
nitudes as defined above are13-14 for the JHK bands at large
angular separations (the faintest sources are at 15-16).
We carefully searched our SUSI2 images for moving
sources using blinking and image subtraction techniques, with-
out success. Using planetary models from Baraffe et al. (2003)
for the age of α Cen (5 Gyr, from Th´evenin et al. 2002, see also
Eggenberger et al. 2004), we can estimate the maximum mass
of the possible companions. The most constraining limits are
provided by our I band images. The Z filter did not provide a
significant improvement in sensitivity, due to the cutoff of the
CCD quantum efficiency. The I band provides in addition a bet-
ter defined average wavelength that makes the comparison with
model magnitudes more accurate. Fig. 2 gives the sensitivity of
our search in terms of companion mass. We can set a maximum
mass of ≈ 15 MJ for projected separations larger than 100 AU.
This limit can be extended down to 10 MJ if we consider our
faintest detected objects. Between 50 and 100 AU, the presence
of a companion with a mass larger than about 30 MJ is highly
unlikely.
5. Conclusion
From deep CCD images of α Cen, we did not identify any co-
moving companion. Within the explored area, this negative re-
sult sets an upper mass limit of 15-30 MJ to the possible com-
panions orbiting α Cen B or the pair, for separations of 50-
300 AU. When combined with existing radial velocity searches
(e.g. Endl et al. 2001) and our adaptive optics results (see
3 http://vizier.u-strasbg.fr/viz-bin/VizieR
P. Kervella & F. Th´evenin: CCD survey of the environment of α Centauri (RN)
3
Fig. 1. Tricolor filtered image from the SUSI2 images of α Cen (February-April 2004).
)
J
M
(
s
s
a
m
m
u
m
x
a
M
i
50
40
30
20
10
0
0
50
100
150
200
250
300
Projected distance to ! Cen (AU)
Fig. 2. Maximum mass of the possible companions orbiting
α Cen. The solid curve corresponds to the model masses by
Baraffe et al. (2003) for our SUSI2 limiting magnitudes in the
I band. The dashed curve corresponds to the faintest detected
objects.
Fig. 7 in Paper I), this mostly excludes the presence of a 20-
30 MJ companion within 300 AU. This non-detection should
have consequences on the modeling of the stellar interior of
α Cen in particular, but also of other stars. In order to settle
the debate on the masses of α Cen (and in particular of B),
one should now check one by one the hypotheses made for
the calibration of this system. This also implies that the new
abundances of Asplund et al. (2004) should be considered as
soon as possible. As an illustration of the importance of this
parameter, it has been debated recently a possible underesti-
mate of the Ne solar abundance that could counter balance the
effect of the decrease of the solar opacity due to the decrease of
the oxygen abundance from 3D models (Young 2005). Liefke
& Schmitt (2006) have published a new value of the chemi-
cal abundance ratio of Ne/O based on coronal studies. These
recent results will be advantageously introduced in the deter-
mination of the metallicity of the α Cen binary to better model
the evolutionnary track of both stars.
Acknowledgements. Based
Telescopes at La Silla under programs 272.C-5010 and 077.C-0587.
observations made with ESO
on
References
Asplund, M., Grevesse, N., Sauval, A. J., Allende Prieto, C., &
Kiselman, D. 2004, A&A, 417, 751
Baraffe, I., Chabrier, G., Barman, T. S., Allard, F., & Hauschildt, P. H.
2003, A&A, 402, 701
Bertin, E. & Arnouts, S. 1996, A&AS, 117, 393
Burki, G., Rufener, F., Burnet, M., et al. 1995, A&AS, 112, 383
Eggenberger, A., Charbonnel, C., Talon, S., et al. 2004, A&A, 417,
235
Endl, M., Kurster, M., Els, S., Hatzes, A. P., & Cochran, W. D. 2001,
A&A, 374, 675
Kervella, P., Th´evenin F., Coud´e du Foresto, V., & Mignard, F. 2006,
A&A, in press (Paper I)
4
P. Kervella & F. Th´evenin: CCD survey of the environment of α Centauri (RN)
Liefke, C., Schmitt,
astro-ph/0609015
J. H. M. M. 2006, A&A,
in press,
Robin, A. C., Reyl´e, C., Derri`ere, S., Picaud, S. 2003, A&A, 409, 523.
Erratum 2004, A&A, 416, 1572
Skrutskie, R. M., Cutri, R., Stiening, M. D., et al. 2006, AJ, 131, 1163
Th´evenin, F., Provost, J., Morel, P., et al. 2002, A&A, 392, L9
Young, P. R. 2005, A&A, 444, L45
List of Objects
'HD 128620' on page 1
'HD 128621' on page 1
|
astro-ph/0008503 | 1 | 0008 | 2000-08-30T23:30:09 | Warm Absorbing Gas in Cooling Flows | [
"astro-ph"
] | We summarize the discovery of oxygen absorption and warm (10^5-10^6 K) gas in cooling flows. Special attention is given to new results for M87 for which we find the strongest evidence to date for ionized oxygen absorption in these systems. We briefly discuss implications for observations of cooling flows with Chandra and XMM. | astro-ph | astro-ph |
Gas & Galaxy Evolution
ASP Conference Series, Vol. **VOLUME**, 2000
J. E. Hibbard, M. P. Rupen and J. H. van Gorkom, eds.
Warm Absorbing Gas in Cooling Flows
David A. Buote
UCO/Lick Observatory, University of California, Santa Cruz, CA 95064
Abstract. We summarize the discovery of oxygen absorption and warm
(105
−106 K) gas in cooling flows. Special attention is given to new results
for M87 for which we find the strongest evidence to date for ionized
oxygen absorption in these systems. We briefly discuss implications for
observations of cooling flows with Chandra and XMM.
1.
Introduction
The inhomogeneous cooling flow scenario is often invoked to interpret the X-
ray observations of massive elliptical galaxies, groups, and clusters. The key
prediction of this scenario is the existence of large quantities of gas that have
cooled out of the hot phase and dropped out of the flow. The only evidence
for large amounts of mass drop-out arises from the excess soft X-ray absorption
from cold gas found for many cooling flows especially from spectral analysis of
Einstein and ASCA data. This interpretation is highly controversial because for
systems with low Galactic columns no excess absorption from cold gas is ever
found with the ROSAT PSPC which should be more sensitive because of its
softer bandpass, 0.1-2.4 keV. Furthermore, the large intrinsic columns of cold
H indicated by the Einstein and ASCA data are in embarrassing disagreement
with H i and CO observations. We have re-examined the ROSAT PSPC data
of cooling flows to search for evidence of intrinsic soft X-ray absorption and in
particular have allowed for the possibility that the absorber is not cold.
2. M87
M87 is arguably the best target for study of its hot plasma with ROSAT because
it is the brightest nearby galaxy, group, or cluster that also possesses an ambient
gas temperature (∼ 2 keV) that lies within the bandpass of the PSPC. From
analysis of spatially resolved, deprojected ROSAT PSPC spectra we find the
strongest evidence to date for intrinsic oxygen absorption and multiphase gas
in the hot ISM/IGM/ICM of a galaxy, group, or cluster. When attempting to
describe the 0.2-2.2 keV ROSAT emission of M87 by a single-phase hot plasma
modified by standard Galactic absorption the best-fitting model displays striking
residuals in the spectrum (Figure 1): (1) excess emission above the model for
0.2-0.4 keV and (2) excess absorption below the model for 0.5-0.8 keV. These
features are apparent out to the largest radii investigated (∼ 100 kpc) and cannot
be attributed to errors in the calibration or the background subtraction.
1
2
Buote
Excess Emission
Excess Absorption
Figure 1.
(Left) ROSAT PSPC spectrum of the the central arcminute
of M87. The fitted model has hydrogen column density fixed to Galactic and
Fe abundance fixed to 0.78Z⊙. (Right) Radial profiles of hydrogen column
density of the standard foreground absorber for different values of the mini-
mum energy of the bandpass for single-phase models. Only data with energies
E > Emin are included in the fits. The solid lines (magenta) indicate the range
of Galactic column density. Both figures are taken from Buote (2000c).
The principal result is that the 0.5-0.8 keV absorption is consistent with
that of a collisionally ionized plasma with a temperature of 105−6 K, where the
lack of evidence of absorption below 0.5 keV strongly excludes the possibility
of absorption from cold material as has been assumed in essentially all previ-
ous studies of absorbing material in cooling flows. In fact, the excess emission
observed between 0.2-0.4 keV, which is also manifested as sub-Galactic column
densities in models with standard (cold) absorbers, has a temperature that is
consistent with the emission from gas responsible for the 0.5-0.8 keV absorption
and could not be explained if dust were responsible for both the absorption and
soft emission.
Only multiphase models can provide good fits over the entire PSPC band-
pass while also yielding temperatures and Fe abundances that are consistent with
results from ASCA and SAX at large radii. Both cooling flow and two-phase
models indicate that the fraction of warm gas with respect to the total emission
measure differs qualitatively for radii interior and exterior to ∼ 5′ (∼ 26 kpc).
For r > 5′ the data are consistent with a constant (or slowly varying) fraction
of warm gas as a function of radius. But for r < 5′ the warm gas fraction varies
from ∼ 20% within the central bin (1′) to essentially zero within the next few
bins (out to 5′).
This behavior is to be contrasted with the absorption optical depth profiles
which are approximately constant with radius. It is puzzling why the oxygen
absorption optical depth does not dip between R ∼ 1′
− 5′ as would be expected
if the absorption and excess soft emission arise from the same material. Perhaps
this is a result of the simplifications we have employed (e.g., only assume a single
absorption edge), and with rigorous consideration of the radiative transfer and
the absorption from several ionization states of oxygen (and carbon and nitrogen)
a self-consistent description of the multiphase medium will be obtained.
Warm Absorbing Gas in Cooling Flows
3
The oxygen absorption and soft emission from warm (105−6 K) gas in M87
we have detected using the ROSAT PSPC (0.2-2.2 keV) is also able to satisfy the
detections of excess emission with data at lower energies (0.07-0.25 keV) from
EUVE (Lieu et al. 1996) and the detection of excess absorption at higher energies
(0.5-10 keV) from Einstein and ASCA. Previous studies using only EUVE data
could not decide between a non-thermal and thermal origin for the excess soft
emission. Similarly, the previous detections of absorption with Einstein and
ASCA could not constrain the temperature of the absorber and always assumed
a cold absorber with solar abundances. Even previous studies with ROSAT that
neglected data below ∼ 0.5 keV also could not constrain the temperature of the
absorber (Sanders et al. 2000) because of the strong dependence of the inferred
absorption on the lower energy limit of the bandpass (Figure 1).
Hence, the ROSAT detection of intrinsic absorption that is localized in
energy (0.5-0.8 keV) is the key piece of evidence for establishing the presence of
warm (105−6 K) gas distributed throughout (at least) the central 100 kpc. This
evidence for a multiphase ISM in M87 essentially confirms the original detection
within the central ∼ 2′ using the Einstein FPCS. However, instead of intrinsic
oxygen absorption Canizares et al.
(1982) inferred a super-solar O/Fe ratio
which does not agree with subsequent analyses of M87 using other instruments.
The anomalous O/Fe ratio is probably attributed to either calibration error in
the FPCS or to an underestimate of the continuum due to the absorption edges.
The total mass of the warm gas implied by the oxygen absorption is con-
sistent with the amount of matter deposited by an inhomogeneous cooling flow.
On the other hand, the mass deposition profile and the profile of warm emission
fraction of the two-phase models indicate that the emission of the warm compo-
nent is suppressed over r ∼ 1′
− 5′ where the radio emission from the AGN jet
clearly distorts the X-ray isophotes (Owen et al. 2000). This coincidence sug-
gests that the AGN has influenced the hot ISM in these central regions and may
have suppressed the cooling emission of the warm component (see Binney 1999).
Within the central arcminute the gas has apparently readjusted and is cooling
while at large radii, r > 5′, the cooling flow was not disturbed significantly by
the AGN. A hybrid model of a standard cooling flow with AGN feedback seems
promising for M87.
3. Other Cooling Flows
The highly significant detection of intrinsic oxygen absorption in the ROSAT
PSPC data of M87 confirms the picture of a multiphase warm+hot medium in
cooling flows deduced from the lower S/N PSPC data of galaxies, groups, and the
cluster A1795 in Buote (2000a,b). Unlike M87 the absorption and soft emission
features associated with the warm gas are not obvious upon visual examination
of the lower S/N PSPC spectra of these systems (e.g., see Figure 5 of Buote
2000b), but five galaxies and groups (NGC 507, 1399, 4472, 4649, and 5044)
and the cluster A1795 clearly show the same type of sensitivity of NH to the
lower energy limit of the bandpass (Figure 1). Most of these lower S/N systems
also have sub-Galactic values of NH when fitted to the entire bandpass consistent
with excess soft emission like M87. Hence, warm absorbing gas appears to be
common in cooling flows ranging from galactic to cluster scales.
4
Buote
4. Observing Oxygen Absorption with Chandra and XMM
With the launches of Chandra and XMM it is an advantageous time to study
oxygen absorption and warm ionized gas in cooling flows. The XMM (EPIC)
and Chandra (ACIS-S) CCDs both have substantially better energy resolution
than the PSPC and, unlike ASCA, both also have bandpasses that extend down
to 0.1-0.2 keV important for measuring the emission from the warm gas.
At the moderate spectral resolution of the Chandra and XMM CCDs the
absorption signature of the warm gas is expected to be a relatively broad trough
over energies ∼ 0.5 − 0.8 keV: thus future Chandra and XMM observations will
not see a single sharp feature. This broad feature is the result of the cumulative
absorption of edges of primarily oxygen but with significant contributions from
C and N (see Krolik & Kallman 1984 for a model of a warm absorber in coronal
equilibrium). Perhaps the simplest means to initially quantify the warm absorp-
tion would be to show the sensitivity of NH to Emin for a standard absorber
model (Figure 1).
The grating spectrometers of XMM and Chandra have even better energy
resolution (but smaller effective area) and, in principle, might detect individual
absorption edges. These gratings are strictly valid only for point sources and
thus will only be able to obtain spectra for the very central regions of cooling
flows (<∼ 10′′). Since it is very likely that the warm absorbing gas emits over
a range of temperatures and does not obey strict coronal equilibrium, the high
resolution spectra offered by the gratings may detect departures from simple
isothermal coronal absorption models and thus provide vital insights into the
thermodynamic state of the warm gas.
References
Binney, J. J. 1999, in The Radio Galaxy M87, ed. H.-J. Roser & K. Meisenheimer,
(Springer-Verlag), 136
Buote, D. A. 2000a, ApJ, 532, L113
Buote, D. A. 2000b, ApJ, in press (astro-ph/0001330)
Buote, D. A. 2000c, ApJ, in press (astro-ph/0008408)
Canizares, C. R., Clark, G. W., Jernigan, J. G., Markert, T. H. 1982, ApJ, 262, 33
Krolik, J. H., & Kallman, T. R., 1984, ApJ, 286, 366
Lieu, R., Mittaz, J. P. D., Bowyer, S., Lockman, F. J., Hwang, C.-Y., & Schmitt, J. H.
M., 1996, ApJ, 458, L5
Owen, F. N., Eilek, J. A., & Kassim, N. E. 2000, ApJ, in press (astro-ph/0006150)
Sanders, J. S., Fabian, A. C., & Allen, S. W. 2000, MNRAS, in press (astro-ph/0006394)
|
astro-ph/0601602 | 1 | 0601 | 2006-01-26T14:37:08 | Production of antimatter in the galaxy | [
"astro-ph"
] | The astronomical dark matter could be made of weakly interacting massive species whose mutual annihilations should produce antimatter particles and distortions in the corresponding energy spectra. The propagation of cosmic rays inside the Milky Way plays a crucial role and is briefly presented. The uncertainties in its description lead to considerable variations in the predicted primary fluxes. This point is illustrated with antiprotons. Finally, the various forthcoming projects are rapidly reviewed with their potential reach. | astro-ph | astro-ph | LAPTH -- Conf -- 1132/05
Production of antimatter in the galaxy
P Salati
LAPTH, 9 Chemin de Bellevue, BP110, 74941 Annecy -- le -- Vieux Cedex, FR
E-mail: [email protected]
Abstract. The astronomical dark matter could be made of weakly interacting massive
species whose mutual annihilations should produce antimatter particles and distortions in the
corresponding energy spectra. The propagation of cosmic rays inside the Milky Way plays a
crucial role and is briefly presented. The uncertainties in its description lead to considerable
variations in the predicted primary fluxes. This point is illustrated with antiprotons. Finally,
the various forthcoming projects are rapidly reviewed with their potential reach.
1. Cosmic ray propagation throughout the Milky Way
Supersymmetric neutralinos or Kaluza -- Klein particles could explain the dark matter that is
indirectly observed around galaxies, inside their clusters and on cosmological scales. These
species are hunted for by several experimental collaborations. As they annihilate inside the
Milky Way halo, they produce high -- energy cosmic rays and in particular antimatter particles
such as antiprotons, antideuterons and positrons. These are rare elements which are already
manufactured inside the galactic disk through the spallations of primary cosmic rays on the
interstellar gas. That conventional process provides the natural background inside which the
DM induced antimatter signature may be burried and whose spectral distortions could reveal
an exotic signal. It is therefore crucial to derive as precisely as possible the energy spectra of
the various antimatter species irrespective of the production mechanisms and to evaluate how
well they are known. To reach that goal, a precise understanding of cosmic ray propagation
is mandatory. We briefly sketch here how that propagation is understood and how well it is
modeled.
6
0
0
2
n
a
J
6
2
1
v
2
0
6
1
0
6
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
Interstellar nuclei are accelerated by the passage of supernovae induced shock waves. The
sources for primary cosmic ray nuclei are therefore located inside the disk of the Milky Way.
Primaries propagate then inside the galactic magnetic fields and undergo collisions on its
irregularities -- the Alfv´en waves. This motion is described in terms of a diffusion process
whose coefficient K(E) = K0 β Rδ increases with the rigidity R of the cosmic ray particle as
a power law. Because the scattering centers move with a velocity Va ∼ 20 to 100 km s−1, a
second order Fermi mechanism yields some diffusive reacceleration whose coefficient KEE may
be expressed as
KEE =
2
9
V 2
a
E2β4
K(E)
.
(1)
Ionization, adiabatic and Coulomb energy losses must also be taken into account through the
energy loss rate bloss(E). Finally, galactic convection could wipe cosmic rays away from the disk
with a velocity VC ∼ 5 to 15 km s−1. The number of particles par unit of volume and space
ψ = dn/dE may be derived form the master equation
VC ∂zΨ − K ∆Ψ − ∂Enbloss(E) Ψ + KEE(E) ∂E Ψo = q ,
(2)
that applies for any cosmic ray species -- primary or secondary -- as long as the various production
processes are appropriately described in the rate q. That general scheme may be implemented
either numerically [1, 2] or semi -- analytically through the two -- zone model discussed in [3].
The latter method allows an easier derivation of the theoretical uncertainties associated to the
various parameters at stake -- namely K0, δ, Va, VC and the thickness L of the confinements layers
that extend above and beneath the Milky Way disk. The space of these propagation parameters
has been extensively scanned [3] in order to select the allowed regions where the predictions on
B/C -- a typical CR secondary to primary ratio -- match the observations. Tens of thousands
of propagation models have survived that crucial test. The diffusion parameters are indeed still
largely undetermined contrary to what is generally believed in our community. Recently, the
same conclusion has been independently reached [4] with the help of a fully numerical code [1]
in which the convective wind VC increases linearly with vertical heigh z. However, the B/C
ratio could not be accounted for when galactic convection and diffusive reacceleration were both
implemented at the same time.
2. The various antimatter backgrounds and signals
2.1. Antiprotons
Secondary antiprotons are produced through the spallations of cosmic ray protons on the
interstellar material
The corresponding source rate may be expressed as
p (CR) + H (ISM) → ¯p + X .
qsec
¯p (r, E¯p) = Z +∞
E 0
p
dσpH→¯p
dE¯p
{Ep → E¯p} nH vp ψp(r, Ep) dEp ,
(3)
(4)
and can easily be modified in order to incorporate heavier nuclei. The top -- of -- atmosphere
antiproton flux has been derived [7] for each of the numerous propagation models that are
compatible with the B/C measurements [3]. In spite of the above -- mentioned large uncertainties
on the propagation parameters, the predictions all lie within the two solid black lines of Figures 1
and 2 and are in good agreement with the observations -- a conclusion that has also been reached
in [4] where a slightly larger uncertainty band is nevertheless obtained. The estimate of the
energy spectrum of secondary antiprotons is therefore quite robust and does not depend much
on the numerous and poorly known parameters that account for cosmic ray galactic propagation.
Actually since boron and secondary antiprotons both originate from the same production site --
Milky Way disk -- it is not surprising if the former species constrains the latter.
On the contrary, primary antiprotons are produced by the mutual annihilations of DM
particles
χ + χ → q + ¯q , W + + W − , Z 0 + Z 0 , H + H → ¯p + X ,
(5)
that take place all over the Milky Way DM halo, in a region that encompasses the above --
mentioned confinement layers whose thickness L is still unknown. As a result, the production
of primary antiprotons whose rate may be expressed as
qDM
¯p
(r, E¯p) =
1
2
hσannvi
dN¯p
mχ )2
dE¯p ( ρχ(r)
,
(6)
Figure 1. When the diffusion parameters are varied over the entire domain that is compatible
with the B/C ratio, antiproton primary fluxes span two orders of magnitude whilst the secondary
component lies within a much narrower band. The case of a resonant LZP [5] has been featured
here [6] with MLZP = 40 -- blue dashed -- and 50 GeV -- solid magenta. A canonical isothermal
DM distribution has been assumed.
In the case of minimal diffusion, the LZP signal is well
below the background.
suffers from large uncertainties. Predictions span a range of ∼ two orders of magnitude at low
energy [8, 6]. This crucial point is illustrated in Figure 1 where the case of a resonant LZP
is featured for the maximal, median and minimal configurations of table 1. Another source of
Table 1. Astrophysical parameters of the cosmic ray galactic propagation models giving the
maximal, median and minimal primary antiproton fluxes compatible wih B/C analysis [8].
case
δ
K0
[kpc2/Myr]
L
Vc
[kpc]
[km/sec]
max
0.46
med 0.70
min
0.85
0.0765
0.0112
0.0016
15
4
1
5
12
13.5
uncertainty is related to the DM distribution itself ρχ(r) whose radial profile is generically given
by
ρχ(r) = ρ DM ⊙ (cid:26) r⊙
r (cid:27)γ (cid:26) 1 + (r⊙/a)α
1 + (r/a)α (cid:27)(β−γ)/α
.
(7)
The distance of the Solar System from the galactic center is r⊙ = 8 kpc. The local -- Solar
System -- DM density has been set equal to ρ DM ⊙ = 0.3 GeV cm−3. In the case of the pseudo-
isothermal profile, the typical length scale a is the radius of the central core. The profile indices
α, β and γ of the dark matter distributions which have been considered in Figure 2 are indicated
in table 2. Notice finally that the halo could be clumpy so that the DM species annihilations
Table 2. Dark matter distribution profiles in the Milky Way.
Halo model
Cored isothermal [9]
Navarro, Frenk & White [10]
Moore [11]
α
2
1
1.5
β
2
3
3
γ
0
1
1.3
a [kpc]
4
25
30
could be enhanced by a boost factor that is still very uncertain.
Figure 2. The effect of the DM halo profile is presented [6] in this figure where the mass of
the LZP [5] has been set equal to MLZP = 60 GeV with a Kaluza -- Klein scale MKK of 3 TeV.
The more divergent and concentrated the LZP distribution at the center of the Milky Way,
the larger the antiproton yield. That effect is particularly acute in this plot where maximum
diffusion parameters have been assumed.
Constraining the MSSM parameter space or deriving informations on Kaluza -- Klein particles
from antiproton measurements may turn out to be tricky. However in the case of light neutralinos
whose relic density accounts for the WMAP measurement, stringent limits may be extracted. In
particular should L exceed 3 kpc, a neutralino lighter than 40 GeV would be basically excluded
-- see [12] for details.
2.2. Antideuterons
Secondary antideuterons are manufactured by CR nuclei spallations in which an antiproton --
antineutron pair is produced and then merges. This fusion is a very rare event because two
antinucleons must be produced at the same time and the difference between their momenta
must not exceed Pcoal. The typical value that has been derived in [13] and used in [14] for that
coalescence momentum is Pcoal = 58 MeV. Because antideuterons are easily destroyed when
they interact -- their binding energy is only 2.2 MeV -- they should not a priori undergo inelastic
and non -- annihilating scattering. In the case of antiprotons, that reaction is very efficient and
flattens the spectrum at low energy -- hence the difficulty to disentangle a primary signal from
the secondary background. But in the case of antideuterons, that process has been assumed
until recently to be negligible. If so, the secondary antideuteron spectrum exhibits a deficiency
at low energy as advocated in [13] and primary antideuterons should be hunted for below a
kinetic energy of ∼ 3 GeV/n.
This point has been extensively reanalyzed recently in [15].
If actually the ∆ -- resonance
excitation
¯D + p → ¯D + ∆
(8)
is forbidden on the basis of isospin conservation, the production of pions is nevertheless possible
through
¯D + p → ¯D + p + n π .
(9)
The cross section for inelastic and non -- annihilating ¯D interactions is therefore small but non --
vanishing. This may allow a certain flattening of the spectrum in the low energy region and could
weaken the relevance of antideuterons as a clear signature for DM species. An additional source
of spectral flattening investigated in [15] is the production of antideuterons through the collisions
of already propagating antiprotons on interstellar H and He. The threshold for manufacturing
a single antineutron is just 7 mp -- instead of 17 mp for a pair -- and low energy antideuterons
should be more abundant. The final conclusion reached in [15] -- with a more correct value of
Pcoal = 79 MeV -- strengthens the case for antideuterons. Below ∼ 1 − 2 GeV/n, the flux of
TOA secondary antideuterons does not exceed ∼ 10−7 ¯D m−2 s−1 sr−1 (GeV/n)−1, a typical
value expected for a primary component in the case of annihilating supersymmetric neutralinos.
As a matter of fact, a detailed analysis of the detection prospects of supersymmetric dark
matter in the mSUGRA framework has been performed in [16] for models whose thermal relic
density is in the range favored by current precision cosmological measurements. Direct detection
and searches for antideuterons were found to be the most promising signatures.
2.3. Positrons
Energy losses play a crucial role in the propagation of positrons. Because galactic convection
and diffusive reacceleration can be disregarded, the master equation (2) simplifies into
− ~∇ ·nK (~x, E) ~∇Ψo − ∂Enbloss(E) Ψo = q (~x, E) .
(10)
Above a few GeV, positron energy losses are dominated by synchrotron radiation in the galactic
magnetic fields and by inverse Compton scattering on stellar light and on CMB photons. The
energy loss rate increases with the positron energy ǫ = E/1 GeV as bloss(E) = ǫ2/τE where
the relevant timescale is typically τE = 1016 s. The diffusion coefficient K is assumed to
be homogeneous and equal to the galactic averaged value that is used for antiprotons and
antideuterons -- see section 1. But if those antinuclei can originate from remote regions [17, 18]
and probe a significant portion of the Milky Way, the high energy positrons that are detected
at the Earth cannot come from far away and the unknown solar neighborhood value should be
preferred. Future high precision measurements on the 10Be to 9Be ratio will be very valuable
since the former nucleus is unstable and also originates from the local bubble [19]. Equation (10)
may be solved either numerically [1] or with the help of the smart trick proposed in [20]
which consists in translating the positron energy E into pseudo -- time. Measurements of the
positron flux show evidence for a spectral excess [21, 22] at energies ∼ 10 GeV. That feature is
hard to explain only with secondary positrons whose flux is nevertheless subject to theoretical
uncertainties -- especially below ∼ 5 GeV -- as shown in [4]. Numerous analysis have been
devoted to the explanation of the putative HEAT excess in terms of annihilating supersymmetric
neutralinos or Kaluza -- Klein species. The latter are more promising because they directly yield
electron -- positron pairs with a branching ratio ∼ 20%. Large values for the boost factor or for
the annihilation cross section are in any case necessary.
The PAMELA [23] and AMS -- 02 [24, 25] missions will collect positrons up to hundreds of GeV
and will bring valuable informations on the HEAT excess and on the possible presence in the
Milky Way halo of DM particles. The reach of these experiments have been studied in [26] where
three years of observation and a boost factor of 5 have been assumed. In the case of a bino -- like
neutralino which mostly produces b¯b pairs, PAMELA will be sensitive to configurations down
to an annihilation cross section of order a few times 10−26 cm3 s−1 whereas AMS -- 02 will lower
that value by a factor of 4. If the neutralino is now a higgsino, PAMELA will be sensitive to
masses up to 380 GeV and AMS -- 02 will improve that limit up to 650 GeV. If now the galactic
DM is made of Kaluza -- Klein species, that reach will become 550 GeV in the case of PAMELA
and 1 TeV for AMS -- 02. Remember finally that -- as in the case of antiprotons -- these results
are affected by large uncertainties related to the propagation model.
3. Forthcoming experiments
The forthcoming satellite missions PAMELA and AMS -- 02 have just been mentioned. They are
compared in table 3 with the BESS -- polar experiment [27]. Long duration flights around the
poles where the geomagnetic shield is no longer active allow for large values of the acceptance.
Below 10 GeV, the effective sensitivity of BESS -- polar for low energy antiprotons will be actually
3 times larger than for PAMELA. In the same spirit, the CREAM collaboration [28] will measure
the B/C ratio between 1 and 103 TeV. It aims at the determination of the index δ that drives
the energy dependence of the diffusion coefficient K. Such a measurement is mandatory in order
to reduce the large uncertainties that plague the theoretical predictions on cosmic rays [29] and
in particular those on the DM induced primaries. The search for antideuterons is also extremely
active. A new upper limit of 1.9 × 10−4 m−2 s−1 sr−1 (GeV/n)−1 has been derived [30] for the
differential TOA flux of cosmic ray antideuterons between 0.17 and 1.15 GeV/n. A pioneering
concept [31, 32] for detecting low energy antideuterons is based on their capture by atoms whose
electrons are rapidly expelled. The result is an antideuteron that orbits alone around the nucleus.
As it cascades towards the ground state, several well defined X -- ray photons are emitted and
yield a unique signature as explained by C. Hailey in this conference.
Acknowledgments
Pierre Salati would like to thank the organizers for their generous hospitality as well as for the
inspiring and exciting atmosphere of this very nice meeting.
References
[1] Strong A W and Moskalenko I V 1998 Astrophys. J. 509 212 -- 228 (Preprint astro-ph/9807150)
[2] Moskalenko I V, Strong A W and Mashnik S G 2005 AIP Conf. Proc. 769 1612 -- 1617 (Preprint
astro-ph/0411400)
Table 3. Comparison between PAMELA, BESS -- polar and AMS -- 02 as presented in [27].
Project
PAMELA
BESS -- polar
AMS -- 02
Satellite
3 years
300 -- 600 km
70.4◦
0.0021 m2 sr
740
Flight vehicle
Flight duration
Altitude
Orbit
Acceptance
MDR (GV)
Particle identification TOF/TRD/CAL TOF/ACC
Number of helium
Launch
long -- duration balloon ISS
10 -- 20 days
37 km (5 g cm−2)
≥ 70◦S Lat.
0.3 m2 sr
150
3 -- 5 years
320 -- 390 km
51.7◦
0.3 m2 sr
∼ 1000
TOF/TRD/RICH/CAL
2 × 109
2007
4 × 107
December 2005
(1 − 2) × 107
December 2004
[3] Maurin D, Donato F, Taillet R and Salati P 2001 Astrophys. J. 555 585
[4] Lionetto A M, Morselli A and Zdravkovic V 2005 JCAP 0509 010 (Preprint astro-ph/0502406)
[5] Agashe K and Servant G 2005 JCAP 02 002
[6] Barrau A, Salati P, Servant G, Donato F, Grain J, Maurin D and Taillet R 2005 Phys. Rev. D72 063507
(Preprint astro-ph/0506389)
[7] Donato F, Maurin D, Salati P, Barrau A, Boudoul G and Taillet R 2001 Astrophys. J. 563 172
[8] Donato F, Fornengo N, Maurin D, Salati P and Taillet R 2004 Phys. Rev. D 69 063501
[9] Bahcall J N and Soneira R M 1980 Astrophys. J. Suppl. 44 73
[10] Navarro J F, Frenk C S and White S D M 1996 Astrophys. J. 462 563
[11] B Moore et al 1999 Mon. Not. Roy. Astron. Soc. 310 1147
[12] Bottino A, Donato F, Fornengo N and Salati P 2005 Phys. Rev. D72 083518 (Preprint hep-ph/0507086)
[13] Donato F, Fornengo N and Salati P 2000 Phys. Rev. D62 043003 (Preprint hep-ph/9904481)
[14] Barrau A, Boudoul G, Donato F, Maurin D, Salati P, Stefanon I and Taillet R 2003 Astron. Astrophys. 398
403 -- 410 (Preprint astro-ph/0207395)
[15] Duperray R, Baret B, Maurin D, Boudoul G, Barrau A, Derome L, Protasov K and Buenerd M 2005 Phys.
Rev. D71 083013 (Preprint astro-ph/0503544)
[16] Edsjo J, Schelke M and Ullio P 2004 JCAP 0409 004 (Preprint astro-ph/0405414)
[17] Taillet R and Maurin D 2003 Astron. Astroph. 402 971
[18] Maurin D and Taillet R 2003 Astron. Astroph. 404 949
[19] Donato F, Maurin D and Taillet R 2002 Astron. Astroph. 381 539 -- 559
[20] Baltz E A and Edsjo J 1999 Phys. Rev. D59 023511 (Preprint astro-ph/9808243)
[21] S W Barwick et al [HEAT Collaboration] 1997 Astrophys. J. 482 L191
[22] M A DuVernois et al [HEAT Collaboration] 2001 Astrophys. J. 559 296
[23] Pearce M (PAMELA) Prepared for International Workshop on Frontier Science: Physics and Astrophysics
in Space, Frascati and Rome, Italy, 14-19 Jun 2004
[24] Demirkoz B M (AMS) Prepared for International Workshop on Frontier Science: Physics and Astrophysics
in Space, Frascati and Rome, Italy, 14-19 Jun 2004
[25] Jacholkowska A (AMS) Prepared for International Workshop on Frontier Science: Physics and Astrophysics
in Space, Frascati and Rome, Italy, 14-19 Jun 2004
[26] Hooper D and Silk J 2005 Phys. Rev. D71 083503 (Preprint hep-ph/0409104)
[27] Yoshida T et al. (BESS) Prepared for 28th International Cosmic Ray Conferences (ICRC 2003), Tsukuba,
Japan, 31 Jul - 7 Aug 2003
[28] Seo E S et al. Prepared for 28th International Cosmic Ray Conferences (ICRC 2003), Tsukuba, Japan, 31
Jul - 7 Aug 2003
[29] Castellina A and Donato F 2005 Astropart. Phys. 24 146 -- 159 (Preprint astro-ph/0504149)
[30] Fuke H et al. 2005 Phys. Rev. Lett. 95 081101 (Preprint astro-ph/0504361)
[31] Mori K, Hailey C J, Baltz E A, Craig W W, Kamionkowski M, Serber W T and Ullio P 2002 Astrophys. J.
566 604 -- 616 (Preprint astro-ph/0109463)
[32] Hailey C J et al. 2005 Submitted to JCAP (Preprint astro-ph/0509587)
|
astro-ph/0401268 | 2 | 0401 | 2004-01-19T14:29:35 | First results with the IOTA3 imaging interferometer: The spectroscopic binaries lambda Vir and WR 140 | [
"astro-ph"
] | We report the first spatially-resolved observations of the spectroscopic binaries lambda Vir and WR 140, which includes the debut of aperture-synthesis imaging with the upgraded three-telescope IOTA interferometer. Using IONIC-3, a new integrated optics beam combiner capable of precise closure phase measurement, short observations were sufficient to extract the angular separation and orientation of each binary system and the component brightness ratio. Most notably, the underlying binary in the prototypical colliding-wind source WR 140 (WC7 + O4/5) was found to have a separation of ~13 milli-arcseconds with a position angle consistent with the images of the 2001 dust shell ejection only if the Wolf-Rayet star is fainter than the O star at 1.65 microns. We also highlight lambda Vir whose peculiar stellar properties of the Am star components will permit direct testing of current theories of tidal evolution when the full orbit is determined. | astro-ph | astro-ph |
First results with the IOTA3 imaging interferometer: The
spectroscopic binaries λ Vir and WR 140
J. D. Monnier1, W. A. Traub2, F. P. Schloerb3, R. Millan-Gabet4, J.-P. Berger5,
E. Pedretti2, N. P. Carleton2, S. Kraus3, M. G. Lacasse2, M. Brewer3, S. Ragland2,
A. Ahearn2, C. Coldwell2, P. Haguenauer6, P. Kern5, P. Labeye7, L. Lagny5, F. Malbet5,
D. Malin3, P. Maymounkov2, S. Morel8, C. Papaliolios2, K. Perraut5, M. Pearlman2,
I. L. Porro9, I. Schanen10, K. Souccar3, G. Torres2, and G. Wallace3
ABSTRACT
We report the first spatially-resolved observations of the spectroscopic bina-
ries λ Vir and WR 140, which includes the debut of aperture-synthesis imaging
with the upgraded three-telescope IOTA interferometer. Using IONIC-3, a new
integrated optics beam combiner capable of precise closure phase measurement,
short observations were sufficient to extract the angular separation and orienta-
tion of each binary system and the component brightness ratio. Most notably,
the underlying binary in the prototypical colliding-wind source WR 140 (WC7
+ O4/5) was found to have a separation of ∼13 milli-arcseconds with a position
angle consistent with the images of the 2001 dust shell ejection only if the Wolf-
Rayet star is fainter than the O star at 1.65µm. We also highlight λ Vir whose
peculiar stellar properties of the Am star components will permit direct testing
of current theories of tidal evolution when the full orbit is determined.
[email protected]: University of Michigan (Astronomy), 500 Church St, Ann Arbor, MI 48109-1090
2Harvard-Smithsonian Center for Astrophysics, 60 Garden St, Cambridge, MA, 02138, USA
3University of Massachusetts, Amherst
4Michelson Science Center, California Institute of Technology, Pasadena, CA
5Laboratoire d'Astrophysique de Grenoble, 414 Rue de la Piscine 38400 Saint Martin d'Heres, France
6Alcatel Space Industries, Cannes, France
7LETI-CEA, Grenoble, France
8European Southern Observatory, Germany
9Massachusetts Institute of Technology, Cambridge, MA
10IMEP-INPG, Grenoble, France
-- 2 --
Subject headings: instrumentation: interferometers -- techniques: interferometric
-- binaries: spectroscopic -- stars: individual (WR 140, λ Vir),
1.
Introduction
Only two optical (visible and infrared) interferometers have published results combin-
ing three or more separated telescopes, the Cambridge Optical Aperture Synthesis Telescope
(Baldwin et al. 1996) and the Navy Prototype Optical Interferometer (Benson et al. 1997).
Combining more than two telescopes is crucial in order to measure the closure phase, a phase
quantity that is uncorrupted by atmospheric turbulence and necessary for image reconstruc-
tion in "very long baseline interferometry" (Jennison 1958; Monnier 2000); indeed, closure
phase lies at the heart of so-called "self-calibration" techniques widely used in the radio (e.g.,
Readhead et al. 1980; Cornwell & Wilkinson 1981).
Here we present the first scientific results of the upgraded 3-telescope IOTA interfer-
ometer (Traub et al. 2003), with the new integrated-optics combiner IONIC-3 (Berger et al.
2003) for precise measurements of visibilities and closure phases. IOTA has much greater
near-infrared sensitivity (demonstrated H mag 7) than the other imaging arrays, allowing
new observational programs including imaging of young stellar objects, mass determinations
of low-mass binaries, and mapping photospheric surface structures of red giants. After de-
scribing our novel instrumentation, we present the first resolved observations for the A-star
binary λ Vir and colliding wind system WR 140 and discuss the significance of our findings.
Here we alert the astronomical community to the need for radial velocity follow-up of these
specific sources as well as to the new research opportunities available with the IOTA3 array.
2. Observations and Analysis
All the data presented were obtained using the IOTA (Infrared-Optical Telescope Array)
interferometer (Traub et al. 2003), which is operated by a consortium of institutions, most
notably the Smithsonian Astrophysical Observatory and the University of Massachusetts at
Amherst. The 0.45-m telescopes are movable among 17 stations along 2 orthogonal linear
arms (telescopes A & C can move along the 35-m northeastern arm, while Telescope B
moves along the 15-m southeastern arm), allowing an aperture 35m×15m to be synthesized
(corresponding to an angular resolution of ∼5×12 milliarcseconds at 1.65µm). Capping a
long development (e.g., Schloerb 1990), the third telescope ("C") was fully incorporated on
-- 3 --
UT 2002 February 23 and now allows visibilities on three baselines and one closure phase
to be measured simultaneously. The initial observations used a prototype integrated optical
(IO) combiner (e.g., Kern et al. 1997; Berger et al. 2001) which operated only with polarized
light. The upgraded combiner IONIC-3 corrected this deficiency in 2002 November, and all
observations presented here were obtained in unpolarized light.
We only just introduce IONIC-3 (Berger et al. 2003) here; an engineering paper with
detailed description of the IO component and its performance is in preparation (Berger et
al. 2004). Light from each telescope is focused into a dedicated single-mode fiber, and the
three fibers are aligned using a silicon v-groove array mated to the planar waveguides on the
IO device. The optical circuit acts to split the light from each telescope before recombining
each telescope pair (AB,BC,AC) at 3 IO couplers. This "pair-wise" combination scheme
leads to 6-interferometric channels (two for each baseline), where the interference fringes are
temporally-modulated by scanning piezo mirrors placed in two of the telescope beams; a
sensitive HgCdTe (Rockwell PICNIC) array (Pedretti et al. 2004a, submitted) detects the
signals, and data are recorded by a vme-based control system linked to a Sun workstation.
We followed established observing procedures, as outlined in previous IOTA publica-
tions (Coud´e du Foresto et al. 1997; Millan-Gabet et al. 2001): observations of target and
nearby calibrator stars were interspersed to calibrate slowly varying system visibilities and
instrumental closure phases. The science targets presented here were observed using a stan-
dard H band filter (λ0 = 1.65µm, ∆λ = 0.30µm); Table 1 contains other observing details.
Figure 1 shows the (u,v)-coverage obtained during each epoch for each science target.
Reduction of the V 2 data was carried out using an Interactive Data Language (IDL)
implementation of the method outlined by Coud´e du Foresto et al. (1997).
In short, we
measured the power spectrum of each interferogram (proportional to the target V 2), after
correcting for intensity fluctuations and subtracting out bias terms from read noise, residual
intensity fluctuations, and photon noise (e.g., Perrin 2003). To assure good closure phase
measurement, we required that interferograms be detected on ≥ 2 baselines, a condition
nearly always maintained by a realtime fringe packet "tracker" (Pedretti et al. 2004b, in
preparation). Lastly, the data pipeline applied a correction for the variable flux ratios for each
baseline by using a transfer matrix (based on single-beam flux measurements; e.g., Coud´e
du Foresto et al. 1997; Monnier 2001). We have studied the absolute calibration accuracy
by observing single stars of known size and found a systematic error of ∆V 2= 0.10 (e.g., 5%
error in visibility for unresolved sources) under the worst observing conditions; this error
has been combined (in quadrature) with the statistical error for fitting purposes. Improved
analysis methods should achieve <
∼1% errors eventually, since we are using single-mode fibers.
We followed the method of Baldwin et al. (1996) for calculating the complex triple
-- 4 --
amplitude in deriving the closure phase, explicitly guaranteeing "fringe frequency closure"
(νAB + νBC + νCA = 0). While photon-noise bias correction is not necessary for the IONIC-3
("pair-wise") combiner, the instrumental closure phase must be determined empirically. The
miniature dimensions of the IO component have minimized drifts to less than 1 degree over
many hours; chromaticity effects, however, limit our absolute precision when the calibrator
and source are not of the same spectral type due to different "effective wavelengths." En-
gineering tests indicate that the instrumental closure phase (ΦCP ) varies systematically by
1.4 ± 0.3 degrees between a hot star (B8) and a cool star (M3) when using the broadband
H filter. While we minimize these errors by using calibrators matched to the spectral type
of the target, a conservative systematic error of ∆ΦCP = 0.75 degs was adopted here to ap-
proximate our expected residual level of miscalibration. Lastly, the sign of our closure phase
(ΦAB + ΦBC + ΦCA for baseline triangle connecting telescopes A→B→C→A) was calibrated
using the well-known binary stars Capella and Matar (Hummel et al. 1994, 1998).
3. Results and Discussion
With calibrated V 2 and closure phases for each target, a static binary star model (sepa-
ration, position angle, brightness ratio) was fitted to the data for each given epoch, assuming
the calibration errors given in the previous section and ignoring bandwidth-smearing effects.
Table 2 contains the results of our fits based on a maximum-likelihood approach giving equal
weights to the visibility data and closure phases. Explicitly, the probability of a given set
of parameters was approximated by: Probability ∝ exp(−
CP)). This
normalization is equivalent to increasing the data errors to force the χ2/DOF= 1, and con-
servatively estimates errors in the context of an incomplete error model. The most probable
values with 1-sigma errors bars (68.4% confidence interval) were calculated for each param-
eter (separation, position angle, brightness ratio) by integrating over the other variables, a
standard practice referred to as marginalizing over these variables.
Vis2 ) −
minimum(χ2
minimum(χ2
χ2
Vis2
χ2
CP
Based on their brightness and color temperatures, the component stars of both bina-
ries are expected to be unresolved here (<
∼0.5 mas); this assumption, as well as neglecting
bandwidth-smearing, may not remain valid if the data precision is improved significantly.
The V 2 and closure phases for a subset of the data are shown in Figure 2 along with the
best-fitting model prediction. We note that λ Vir, which has a ∼ 207 day period, is not
completely static during each observing "epoch." With complete orbital phase coverage, this
deficiency will be corrected by solving for the full orbital elements by fitting directly to the
visibilities, closure phases, and radial velocities, a procedure outlined in detail by Hummel
et al. (1998) and Boden et al. (1999).
-- 5 --
The radio community developed interferometric imaging techniques (e.g., hybrid map-
ping) capable of reobtaining lost individual phases (e.g., Haniff et al. 1987). As demonstrated
by Baldwin et al. (1996), these methods can be applied successfully to optical interferometers
as well. We present the first aperture synthesis images from IOTA3 in Figure 3, depicting
λ Vir for each epoch. The images are constructed iteratively starting with zero initial phases
and using the measured closure phases for self-calibration. CLEAN deconvolution was ap-
plied using an elliptical Gaussian CLEAN beam. Full details of the image-making process
for this source (as well as the binary Capella) can be found in Kraus (2003). The presented
maps are consistent with the binary parameters found through direct fitting, and demon-
strate the imaging capability of the IOTA3+IONIC-3 system, currently the most sensitive
infrared imaging interferometer in the world.
With this new imaging capability, we begin to study binary stars which have never been
resolved before and which represent unique stellar systems whose properties can shed new
light on rare and/or short-lived phases of stellar evolution.
λ Vir is a well-known spectroscopic binary (period 207 days, Stickland 1976) with similar
components of spectral type early Am (both are metallic-lined, Abt 1961). The stars exhibit
the curious property that one is sharp-lined and the other broad-lined, which makes them
easily distinguishable spectroscopically. IOTA3 observations will allow a determination of
the absolute properties of the components once the full orbit is established. Fundamental
stellar parameters, such as the system distance (independent of Hipparcos) and the individual
masses and absolute near-IR luminosities, can be determined through a combination of the
radial velocity orbit, astrometric orbit, and brightness ratio. Further, these properties can
be compared directly against models of stellar evolution to derive the age of the system. The
very different projected rotational velocities of the components is of special interest for these
Am stars, enabling unique tests of current theories of tidal evolution (Tassoul & Tassoul
1997; Zahn & Bouchet 1989, and references therein).
WR 140 is a prototypical colliding-wind binary source consisting of a Wolf-Rayet (WC7)
and an O4-5 star in a highly-eccentric ∼7.9 yr orbit, and has been extensively studied in the
radio, infrared, optical, ultra-violet, and X-rays (e.g., Moffat et al. 1987; Williams et al. 1990;
White & Becker 1995). Notably, infrared emission appears and fades in conjunction with
periastron passage, explained by transient dust formation in the dense shock-compressed gas
at the wind-wind interface (Williams et al. 1990; Usov 1991; Tuthill et al. 1999). This process
was captured in a series of high-resolution near-infrared images by Monnier et al. (2002),
revealing an eastward moving arc of dust consistent with the O-star being roughly east of
the WR star at periastron. The O-star is expected to be on (nearly) the opposite side of the
WR star now as compared to periastron, which can be reconciled with the position angle
-- 6 --
found here (PA 152◦) only if the WR star is fainter than the O-star at 1.65µm (as expected
by some workers; e.g. Williams et al. 1990). We intend to combine future high-precision
interferometer observations with new radial velocity data (e.g., Marchenko et al. 2003) to
derive component masses and a distance estimate, fundamental parameters for these systems
needed for definitive interpretation of data across the electromagnetic spectrum.
4. Conclusions
We have reported first scientific results from the IOTA3 interferometer: newly-resolved
separation, position angle, and flux ratio information for double-lined spectroscopic binary
stars λ Vir and WR 140. Using a preliminary data reduction pipeline, calibration is sufficient
for model-fitting and aperture-synthesis imaging using hybrid mapping techniques. New
observations are being pursued in order to derive precise component masses and distance
estimates by combining interferometric and radial velocity data. All the visibility and closure
phase data here have been converted to the new FITS format for Optical Interferometry data
(OI-FITS) 1 and are available upon request.
The authors gratefully acknowledge critical support from SAO, NASA, and NSF (AST-
0138303). EP was supported by a SAO Predoctoral fellowship, JDM by a CfA fellowship,
and RM-G, J-PB, and SR through NASA Michelson Fellowships. GT acknowledges partial
support from NASA's MASSIF SIM Key project (JPL 1240033). In addition, we acknowledge
useful contributions from B. Arezki, A. Delboulbe, C. Gil, S. Gluck, E. Laurent, R. B.
Metcalf, and E. Tatulli. IONIC-3 was developed by LAOG and LETI in the context of the
IONIC collaboration (LAOG, IMEP, LETI), funded by the CNRS and CNES (France). This
publication makes use of data products from the 2MASS survey, which is a joint project of
the University of Massachusetts and IPAC/Caltech, funded by NASA and the NSF. This
research has also used the SIMBAD database, NASA's ADS Service, and services of the
Michelson Science Center at Caltech (http://msc.caltech.edu).
REFERENCES
Abt, H. A. 1961, ApJS, 6, 37
Baldwin, J. E., et al., R. W. 1996, A&A, 306, L13+
1http://www.mrao.cam.ac.uk/∼jsy1001/exchange/
Benson, J. A., et al. , C. S. 1997, AJ, 114, 1221
-- 7 --
Berger, J., et al. 2003, in Interferometry for Optical Astronomy II. Edited by Wesley A.
Traub . Proceedings of the SPIE, Volume 4838, pp. 1099-1106 (2003)., 1099 -- 1106
Berger, J. P., et al. 2001, A&A, 376, L31
Boden, A. F., et al. 1999, ApJ, 527, 360
Cornwell, T. J. & Wilkinson, P. N. 1981, MNRAS, 196, 1067
Coud´e du Foresto, V., Ridgway, S., & Mariotti, J.-M. 1997, A&AS, 121, 379
Haniff, C. A., Mackay, C. D., Titterington, D. J., Sivia, D., & Baldwin, J. E. 1987, Nature,
328, 694
Hummel, C. A., Armstrong, J. T., Quirrenbach, A., Buscher, D. F., Mozurkewich, D., Elias,
N. M., & Wilson, R. E. 1994, AJ, 107, 1859
Hummel, C. A., Mozurkewich, D., Armstrong, J. T., Hajian, A. R., Elias, N. M., & Hutter,
D. J. 1998, AJ, 116, 2536
Jennison, R. C. 1958, MNRAS, 118, 276+
Kern, P., Malbet, F., Schanen-Duport, I., & Benech, P. 1997, in Integrated Optics for
Astronomical Interferometry (1996 Oct 15-16, Grenoble, France), 195 -- +
Kraus, S. 2003, Master's thesis, University of Massachusetts, Amherst
Marchenko, S. V., et al. 2003, ApJ, 596, 1295
Millan-Gabet, R., Schloerb, F. P., & Traub, W. A. 2001, ApJ, 546, 358
Moffat, A. F. J., Lamontagne, R., Williams, P. M., Horn, J., & Seggewiss, W. 1987, ApJ,
312, 807
Monnier, J. D. 2000, in Principles of Long Baseline Stellar Interferometry (JPL Publication
00-009), 203 -- +
Monnier, J. D. 2001, PASP, 113, 639
Monnier, J. D., Tuthill, P. G., & Danchi, W. C. 2002, ApJ, 567, L137
Perrin, G. 2003, A&A, 398, 385
-- 8 --
Readhead, A. C. S., Walker, R. C., Pearson, T. J., & Cohen, M. H. 1980, Nature, 285, 137
Schloerb, F. P. 1990, in Amplitude and intensity spatial interferometry; (Tucson, AZ, Feb.
14-16, 1990) SPIE A91-30676 12-89, 154 -- 165
Stickland, D. J. 1976, in IAU Colloq. 32: Physics of Ap Stars (eds. W. W. Weiss, H. Jenkner
and H. J. Wood), 701 -- +
Tassoul, M. & Tassoul, J. 1997, ApJ, 481, 363
Traub, W. A., et al. 2003, in Interferometry for Optical Astronomy II. Edited by Wesley A.
Traub. Proceedings of the SPIE, Volume 4838, pp. 45-52 (2003)., 45 -- 52
Tuthill, P. G., Monnier, J. D., & Danchi, W. C. 1999, Nature, 398, 487
Usov, V. V. 1991, MNRAS, 252, 49
White, R. L. & Becker, R. H. 1995, ApJ, 451, 352+
Williams, P. M., van der Hucht, K. A., Pollock, A. M. T., Florkowski, D. R., van der Woerd,
H., & Wamsteker, W. M. 1990, MNRAS, 243, 662
Zahn, J.-P. & Bouchet, L. 1989, A&A, 223, 112
This preprint was prepared with the AAS LATEX macros v5.2.
-- 9 --
Table 1. Observing Log for λ Vir and WR 140
Date
(UT)
Interferometer
Configuration(a)
Calibrator Names
(Adopted UD Diameter)
2003 Feb 16,17
λ V ir (A1V + A); H mag = 4.28±0.21
A35-B05-C10
2003 Feb 20,22,23 A25-B15-C10
2003 Mar 21
A35-B07-C25
A35-B07-C10
2003 Mar 22
A35-B15-C10
2003 Mar 23-24
2003 Jun 12-17
A35-B15-C10
HD 126035 (G7III, 0.78±0.24 mas)
HIP 71957 (F2III, 1.20±0.22 mas)
HD 126035
HD 126035
HD 126035; HD 158352 (A8V, 0.44±0.10)
HD 126035; HD 158352
HD 126035
WR 140 (WC7 + O4/5); H mag = 5.429±0.023
2003 Jun 17
A35-B15-C10
HD 192985 (F5IV, 0.46±0.10 mas)
HD 193631 (K0, 0.44±0.10 mas)
Notes: H-band photometry from 2MASS All-Sky Release; calibrator diameters estimated
using getCal software (http://msc.caltech.edu).
aConfiguration refers to the location of telescopes A,B,C on the NE, SE and NE arm
respectively; see §2 for more details.
-- 10 --
Table 2. Binary Parameters for λ Vir and WR 140 at 1.65µm
Source
λ Vir
WR 140
Epoch
(U.T.)
Separation
Position Angle(a)
(milliarcseconds)
(degrees)
Flux
Ratio
χ2/DOF(b)
V 2
CP
2003 Feb 16-23
2003 Mar 21-24
2003 Jun 12-17
2003 Jun 17
18.2 ± 0.6
17.8 ± 0.5
19.5 ± 0.6
12.9+0.5
−0.4
200 ± 3
184 ± 3
22 ± 3
151.7+1.8
−1.3
1.69+0.29
−0.21
1.72 ± 0.07
2.05+0.50
−0.35
1.10+0.06
−0.05
0.45
0.33
0.31
0.13
12.8
2.2
4.8
4.7
aPosition angles (East of North) are measured from the bright to the faint component.
bThe χ2 per DOF (degree of freedom), the reduced χ2, is reported for the model fit to the
V 2 and closure phase (CP) measurements separately.
-- 11 --
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
2003 Mar 21-24
)
m
(
e
t
i
a
n
d
r
o
o
C
V
)
m
(
e
t
i
a
n
d
r
o
o
C
V
20
0
-20
-40
20
0
-20
-40
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
2003 Feb 16-23
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
Lam Vir
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
2003 Jun 12-17
30 20 10 0 -10 -20
U Coordinate (m)
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
WR 140
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
2003 Jun 17
30 20 10 0 -10 -20
U Coordinate (m)
Fig. 1. -- (u,v)-plane coverage of science binaries. Each diamond symbol represents a single
observation with the IOTA3 interferometer, corresponding to approximately 100-200 inter-
ferogram scans. The overlaid image in each panel represents the V 2 of the best fit binary
model for each epoch discussed in text.
-- 12 --
Lam Vir Data from UT2003Mar23
-1
0
Hour Angle (hours)
WR 140 Data from UT2003Jun17
-0.5
0.0
0.5
1.0
Hour Angle (hours)
1
2
1.5
2.0
2.5
d
e
r
a
u
q
S
-
s
V
i
d
e
r
a
u
q
S
-
s
V
i
d
e
r
a
u
q
S
-
s
V
i
B
A
-
C
B
-
C
A
-
e
s
a
h
P
e
r
u
s
o
C
l
)
s
g
e
d
(
-
C
B
A
-
1.0
0.8
0.6
0.4
0.2
0.0
1.0
0.8
0.6
0.4
0.2
0.0
1.0
0.8
0.6
0.4
0.2
0.0
-70
-80
-90
-100
-110
-120
1.0
0.8
0.6
0.4
0.2
0.0
1.0
0.8
0.6
0.4
0.2
0.0
1.0
0.8
0.6
0.4
0.2
0.0
0
-50
-100
-150
d
e
r
a
u
q
S
-
s
V
i
d
e
r
a
u
q
S
-
s
V
i
d
e
r
a
u
q
S
-
s
V
i
B
A
-
C
B
-
C
A
-
e
s
a
h
P
e
r
u
s
o
C
l
)
s
g
e
d
(
-
C
B
A
-
Fig.
2. -- (top panel): This figure shows example binary model fits for λ Vir on
UT 2003 March 23. The top three panels show the V 2 for baselines connecting telescopes
A-B, B-C, and A-C, respectively. The bottom panel shows the calibrated closure phase for
the A-B-C triangle. (bottom panel): Same as top panel for UT 2003 June 17 observations
of WR 140. These curves represent the best-fit model parameters as described in §3. For
all panels, the error bars shown represent only the statistical error and do not include the
systematic error discussed in §2.
-- 13 --
UT2003Feb16-23
UT2003Mar21-24
UT2003Jun12-17
40
20
0
-20
-40
40
20
0
-20
-40
dRA [mas]
Fig. 3. -- First aperture synthesis images from the upgraded IOTA 3-telescope interferom-
eter, depicting λ Vir based on 1.65µm observations at three separate epochs. The binary
components are well-resolved from each other but individually are unresolved; orbital motion
is clearly evident, showing nearly a ∼180◦ rotation between the first and last epoch. Each
panel includes a contour map showing 5% intervals in surface brightness scaled to the peak,
as well as the the CLEAN restoring beam as inset. North is up and East is left.
|
0803.4097 | 2 | 0803 | 2008-06-17T12:18:11 | Detecting gravitational wave emission from the known accreting neutron stars | [
"astro-ph",
"gr-qc"
] | Detection of gravitational waves from accreting neutron stars (NSs) in our galaxy, due to ellipticity or internal oscillation, would be a breakthrough in our understanding of compact objects and explain the absence of NSs rotating near the break-up limit. Direct detection, however, poses a formidable challenge. Using the current data available on the properties of the accreting NSs in Low Mass X-Ray Binaries (LMXBs), we quantify the detectability for the known accreting NSs, considering various emission scenarios and taking into account the negative impact of parameter uncertainty on the data analysis process. Only a few of the persistently bright NSs accreting at rates near the Eddington limit are detectable by Advanced LIGO if they are emitting gravitational waves at a rate matching the torque from accretion. A larger fraction of the known population is detectable if the spin and orbital parameters are known in advance, especially with the narrow-band Advanced LIGO. We identify the most promising targets, and list specific actions that would lead to significant improvements in detection probability. These include astronomical observations (especially for unknown orbital periods), improvements in data analysis algorithms and capabilities, and further detector development. | astro-ph | astro-ph |
Mon. Not. R. Astron. Soc. 000, 000 -- 000 (0000)
Printed 4 June 2018
(MN LATEX style file v2.2)
Detecting gravitational wave emission from the known
accreting neutron stars
Anna L. Watts1,2, Badri Krishnan3, Lars Bildsten4 and Bernard F. Schutz3
1 Max Planck Institut fur Astrophysik, Karl-Schwarzschild-Strasse 1, 85741 Garching, Germany
2 Astronomical Institute "Anton Pannekoek", Kruislaan 403, 1098 SJ Amsterdam, The Netherlands; [email protected]
3 Albert-Einstein-Institut, Max-Planck-Institut fur Gravitationsphysik, Am Muhlenberg 1, 14476 Golm, Germany
4 Kavli Institute for Theoretical Physics, Kohn Hall, University of California at Santa Barbara, CA 93106, USA
4 June 2018
ABSTRACT
Detection of gravitational waves from accreting neutron stars (NSs) in our galaxy, due
to ellipticity or internal oscillation, would be a breakthrough in our understanding
of compact objects and explain the absence of NSs rotating near the break-up limit.
Direct detection, however, poses a formidable challenge. Using the current data avail-
able on the properties of the accreting NSs in Low Mass X-Ray Binaries (LMXBs),
we quantify the detectability for the known accreting NSs, considering various emis-
sion scenarios and taking into account the negative impact of parameter uncertainty
on the data analysis process. Only a few of the persistently bright NSs accreting at
rates near the Eddington limit are detectable by Advanced LIGO if they are emitting
gravitational waves at a rate matching the torque from accretion. A larger fraction
of the known population is detectable if the spin and orbital parameters are known
in advance, especially with the narrow-band Advanced LIGO. We identify the most
promising targets, and list specific actions that would lead to significant improvements
in detection probability. These include astronomical observations (especially for un-
known orbital periods), improvements in data analysis algorithms and capabilities,
and further detector development.
Key words: accretion, accretion disks -- gravitational waves -- stars: neutron -- stars:
rotation -- X-rays: binaries -- X-rays: bursts
1
INTRODUCTION
A number of interferometric gravitational wave observato-
ries have been built with the intention of opening a new ob-
servational window for studying astrophysical objects. These
include the LIGO1, GEO2, VIRGO3 and TAMA4 detec-
tors. The TAMA and LIGO Scientific Collaborations have
demonstrated their ability to reach sensitivity goals, take
year-long stretches of data with good duty cycles, and anal-
yse the data to set, with confidence, upper limits on the
emission from a number of possible sources. At present the
large LIGO and VIRGO detectors are performing a signif-
icant sensitivity upgrade. This is therefore a good time to
carry out a realistic study of the challenges of searching for
gravitational waves from one class of sources that future up-
grades may render detectable: spinning neutron stars (NSs)
in Low Mass X-ray Binaries (LMXBs).
1 http://www.ligo.caltech.edu
2 http://geo600.aei.mpg.de/
3 http://wwwcascina.virgo.infn.it
4 http://tamago.mtk.nao.ac.jp
c(cid:13) 0000 RAS
For the last ten years, ever since a paper by one of us
(Bildsten 1998) suggested that LMXBs could be steady bea-
cons of gravitational waves, the study of LMXBs (and espe-
cially of the source Scorpius X-1) has been one of the scien-
tific goals of the development of detectors with greatly im-
proved sensitivity. Since then, X-ray astronomers have gath-
ered a wealth of new data on these sources, astrophysicists
have built better models, and gravitational wave scientists
have gained considerable experience of their ability to ex-
tract weak signals from data streams. It is timely, therefore,
to revisit these estimates.
The expectation that LMXBs could be strong steady
sources of gravitational waves originates in one of the
most important outstanding questions about NSs: why
they all spin at frequencies much less than the break-up
limit of 1 kHz (Lattimer & Prakash 2007). Simple estimates
(Cook, Shapiro & Teukolsky 1994) of the spin-up timescales
for accreting NSs in LMXBs suggest that there should be no
difficulty in reaching at least 1 kHz. The fastest rotating ac-
creting pulsar is at 599 Hz (Galloway et al. 2005) and many
NS rotate much more slowly (Chakrabarty et al. 2003). The
2 Watts et al.
millisecond radio pulsars, the likely offspring of LMXBs,
also spin at slower rates than expected. The current record
stands at 716 Hz (Hessels et al. 2006) and surveys in nearby
globular clusters continue to reveal a paucity of rapid rota-
tors (Ransom et al. 2005; Hessels et al. 2007; Freire et al.
2008). Whether the apparent spin limit is genuine remains
to be resolved. Nevertheless, given the current sample, it
certainly appears that there is some brake that prevents ac-
creting NSs from reaching the break-up limit.
The candidate mechanisms fall into two main camps. In
the first, accretion torques are reduced and eventually bal-
anced by the interaction between the accretion disk and the
NS's magnetic field (Ghosh & Lamb 1978; White & Zhang
1997; Andersson et al. 2005). The second possibility, and
our focus here, is the loss of angular momentum via the emis-
sion of gravitational radiation (Papaloizou & Pringle 1978;
Wagoner 1984; Bildsten 1998). There are many conceivable
ways for an accreting NS to develop a quadrupolar asymme-
try that leads to gravitational wave (GW) emission: crustal
mountains (Bildsten 1998; Ushomirsky, Cutler & Bildsten
2000; Melatos & Payne
2006;
Haskell, Jones & Andersson 2006), magnetic deformation
(Cutler 2002; Haskell et al. 2008) or internal r-mode oscil-
lations (Bildsten 1998; Andersson, Kokkotas & Stergioulas
1999;
2000;
Heyl
Andersson, Jones & Kokkotas
2002;
2002; Wagoner
2006;
Bondarescu, Teukolsky & Wasserman 2007). The exciting
prediction from all GW emission scenarios is the possible
direct detection of an accreting NS by a ground-based
interferometric gravitational wave detector.
2005; Payne & Melatos
Andersson et al.
Nayyar & Owen
Levin
2002;
1999;
Searches for periodic GWs from NSs have already been
performed with the LIGO and GEO detectors. These include
searches of known radio pulsars (Abbott et al. 2004, 2005b,
2007b), assuming the phase of the GW signal to be locked to
the known NS rotation, and wide parameter surveys for hith-
erto unknown NSs (Abbott et al. 2005a, 2007a, 2008). There
have also been searches for GWs from the accreting NS Scor-
pius X-1. The first (Abbott et al. 2007a) used a coherent
statistic on a template grid utilising 6 hours of data (limited
by computational requirements) from the second science run
of the two LIGO 4 km interferometers, followed by a coinci-
dence analysis between candidates from the two detectors.
The second search (Abbott et al. 2007c) used a method of
cross-correlating the outputs of the two LIGO 4 km detec-
tors using about 20 days of coincident data taken during the
fourth science run. Data from resonant bar detectors have
also been used in these searches; see e.g. Astone et al. (2002,
2005) for a blind all-sky search in a narrow frequency band
using data from the EXPLORER detector.
These early searches, while so far yielding only upper
limits, have served to develop and prove data analysis meth-
ods that will be used on data from future, more sensitive
searches. The VIRGO detector is approaching its design sen-
sitivity and the LIGO detectors have just completed a full
run at their first-stage design sensitivity. An initial upgrade
of LIGO is about to commence, followed by a full upgrade
to Advanced LIGO5 in the next decade. VIRGO expects to
parallel these developments. These impending improvements
5 http://www.ligo.caltech.edu/advLIGO
in GW sensitivity, coupled with the experience of perform-
ing realistic data analysis, make the time right to assess
what the relevant accreting NS properties tell us about the
ultimate detectability of NS sources.
This paper explains the prospects for detecting GWs
from accreting NSs, and identifies the most promising tar-
gets. We start in Section 2 with the best case scenario,
highlighting the data analysis challenges and the need for
detailed information on the accreting NS properties, which
are presented in Section 3. Section 4 explains fully the GW
search data analysis challenges and the direct consequences
of parameter uncertainty on integration times and detection
statistics, allowing us to assess future detectability in Sec-
tion 5. We close in Section 6 by highlighting where progress
can be made in the short-term on NS source properties, and
the implications for current and future gravitational wave
searches.
2 OVERVIEW AND BEST CASE ESTIMATES
The potential strength of a periodic GW signal at frequency
M was esti-
ν from specific LMXBs with accretion rates
mated by Bildsten (1998) (hereafter B98) in the mountain
scenario. In this model a quadrupole moment Q, that is sta-
tionary in the rotating frame of the star (a 'mountain'), leads
to the emission of gravitational waves with a predominant
frequency ν = 2νs, νs being the spin frequency of the NS.
The assumption is that the accretion torque on a NS of mass
M and radius R
Na = M (GM R)1/2 ,
is balanced by the GW torque
Ngw = −
32GQ2Ω5
s
5c5
,
(1)
(2)
where Ωs = 2πνs is the angular frequency of the star. The
accretion rate is estimated from the bolometric X-ray flux F
by assuming that the luminosity is L ≈ GM M /R, yielding
(3)
4πRd2F
M =
,
GM
to the
the distance
where d is
source. The pre-
dicted gravitational wave amplitude h0 (as defined in
Jaranowski, Kr´olak & Schutz (1998))6 can be written in
terms of the GW luminosity,
Egw = NgwΩs, as
h2
0 =
5G
2π2c3d2ν 2
Egw.
(4)
Under the condition of torque balance,
h0 = 3×10−27F 1/2
−8 „ R
10km«3/4„ 1.4M⊙
M «1/4„ 1 kHz
νs «1/2
,
(5)
where F−8 = F/10−8erg cm−2 s−1. Equation (5) makes it
clear that the GW signal strength depends on two observ-
ables, the flux on the sky from the LMXB and the NS spin
rate (Wagoner 1984; Bildsten 1998). This amplitude is then
6 Note that flux formulae are often given in terms of an angu-
lar and time-averaged amplitude h rather than in terms of h0.
In Owen et al. (1998) and Andersson, Kokkotas & Stergioulas
(1999), for example, the quoted amplitude h2 = 2h2
0/5.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
3
compared directly to the best case detectable amplitude in
a long search, hsens
, which we will derive shortly.
0
These amplitudes are sufficiently weak that
long
stretches of data must be folded, using predicted signal tem-
plates. This is called matched filtering, and its sensitivity
improves with the square-root of the observation time, pro-
vided that the template manages to keep phase with the real
signal to within about one radian over the entire duration.
Where spin and orbital parameters are poorly constrained,
many templates must therefore be searched. Analysis can
become statistically and computationally untenable, and
computational loads may limit integration times and thereby
detectability. Contrast for example the most recent searches
for radio pulsars, where a precise timing ephemeris is avail-
able (Abbott et al. 2007b), with searches for the closest
accreting neutron star Scorpius X-1 (Abbott et al. 2007c).
Poor constraints on the spin and orbital parameters for Sco
X-1 necessitate multiple templates. This reduces the feasible
integration time substantially, resulting in upper limits that
are approximately two orders of magnitude larger than those
obtained for the best radio pulsar. We begin our discussion
there.
2.1 Spin Frequencies of Accreting NSs
The discoveries and studies with the Rossi X-ray Timing
Explorer (RXTE) over the last ten years have dramati-
cally improved our understanding of stellar spin rate, νs
(Strohmayer & Bildsten 2006; van der Klis 2006). Having
some constraint on the spin is, as will become clear in later
Sections, the most important factor determining the feasibil-
ity of gravitational wave searches. For this reason we concen-
trate only on sources for which there is some measurement
or estimate of νs. We now summarize the three relevant cat-
egories of accreting NSs: accreting millisecond pulsars, burst
oscillation sources, and kiloHertz Quasi-Periodic Oscillation
(kHz QPO) sources.
The spin is measured directly in the accreting millisec-
ond pulsars, where fixed hotspots, presumably at the mag-
netic footpoints, are a permanent asymmetry. In 1998 only
one such object was known; there are now ten members
of this class, three of which show only intermittent pulsa-
tions. The other measures of spin are indirect. Probably the
most reliable are burst oscillations, seen during Type I X-ray
bursts (when accreted material burns in an unstable ther-
monuclear flash). In 1998 there were six burst oscillation
sources: there are now twelve stars with burst oscillations
seen in multiple bursts (including three pulsars), and seven
stars with tentative detections.
For
the three accreting millisecond pulsars
that
also show burst oscillations,
the burst oscillation fre-
quency is at or very close to the known spin fre-
quency (Chakrabarty et al. 2003; Strohmayer et al. 2003;
Altamirano et al. 2008). It would therefore seem reason-
able to equate the burst oscillation frequency with the
spin frequency for the non-pulsing sources (as was done
in B98). The frequency for a given source is highly con-
sistent from burst to burst, implying that there is at least a
strong dependence on stellar spin (Strohmayer et al. 1998a;
Muno, Ozel & Chakrabarty 2002). The detection of highly
coherent oscillations lasting several hundred seconds dur-
ing a superburst adds further support to this hypothesis
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
(Strohmayer & Markwardt 2002). There are, however, some
complicating factors. Firstly, burst oscillations can exhibit
frequency drifts of up to a few Hz. Secondly, there are some
differences in the properties of the burst oscillations of the
non-pulsing LMXBs as compared to the pulsars (see for
example Watts & Strohmayer (2006)), suggesting that the
mechanism may differ. In models that involve global modes
of the surface layers, for example, the observed frequency
would be offset from spin frequency by several Hz (Heyl
2004; Piro & Bildsten 2005).
The third class of sources to be considered are those
which exhibit twin kHz QPOs. Early observations suggested
that although the frequencies could shift, their separation
remained relatively constant, implying a link to stellar spin.
We now know that separation varies (often quite substan-
tially) as accretion rate changes, and the cause of the kHz
QPOs is still not understood. However, most models still
depend in some way on the stellar spin, either directly or
via the influence on the space-time in the inner regions of
the accretion disk (see van der Klis (2006) for a recent and
comprehensive review). Observations of twin kHz QPOs in
two of the accreting millisecond pulsars have done little to
resolve the situation: in one case separation is rather close to
the spin frequency; in the other it is slightly less than half the
spin (Wijnands et al. 2003a; Linares et al. 2005). Whether
there is any link between kHz QPO separation and stellar
spin will doubtless emerge in due course. However, given
that most models still predict some relation, we follow B98
and include these stars in our analysis.
2.2 Optimal Gravitational Wave Detection
Before being scaled down for their current sensitivity up-
grades, the first generation LIGO interferometers operated
for nearly two years at or better than their design sensitivity,
by the end of which the first generation Virgo detector was
not far behind. The "Enhanced" LIGO and VIRGO interfer-
ometers should begin operation in 2009, and the Advanced
LIGO interferometers are now funded and should be opera-
tional by 2014. Advanced VIRGO is expected on the same
schedule. Beyond this, there are ambitious plans for a third
generation Einstein Telescope (ET) in Europe.
The sensitivity of a gravitational wave detector is de-
termined by the power spectral density of its instrumental
strain noise, normalized to an equivalent gravitational-wave
amplitude h(t). This is just the Fourier power spectrum of
h(t) and is called Sh(ν): the noise power per unit frequency.
It is conventional to plot h(ν) = [Sh(ν)]1/2, which allows
comparison of the noise to the signal's amplitude if one
knows the signal's bandwidth. The sensitivity curves of cur-
rent and future detectors are shown in Figure 1. The design
noise curves for Initial LIGO, Virgo and ET are taken from
analytic models available, for example, in the LIGO software
repository7. The most recent science runs of the LIGO de-
tectors have in fact reached their design goals over a broad
frequency range of interest, above ∼ 40 Hz. The Enhanced
LIGO noise curve in Figure 1 is a realistic estimate of what
can be achieved; thus at low frequencies (below ∼ 40 Hz),
7 http://www.lsc-group.phys.uwm.edu/daswg/projects/lal.html
4 Watts et al.
the Enhanced LIGO noise curve lies above the Initial LIGO
design curve.
The Advanced LIGO detector configurations have not
yet been finalized, and we therefore need to consider differ-
ent possibilities. Advanced LIGO can potentially be oper-
ated in a narrow band mode where sensitivity is gained in
a relatively narrow frequency range at the expense of broad
band sensitivity (see e.g. Meers (1988); Buonanno & Chen
(2002)). This could be particularly relevant for periodic sig-
nals where the frequency is well known. The first panel of
Figure 1 shows an example of such a narrow band noise
curve8 The noise curve can be tuned to target a broad range
of frequencies by changing a number of interferometer pa-
rameters, and for our purposes, the value of the noise at the
minimum is especially important. Thus, the lower panel of
Fig. 1 shows the lower envelope of the narrow band noise
curves above 100 Hz. Between 100 Hz and about 400 Hz,
the narrow band curves are limited by thermal noise, and by
quantum noise at higher frequencies9. See Section 6.4 for ad-
ditional discussion. It is important to keep in mind that the
narrow-band "envelope" does not represent any particular
interferometer configuration but is rather a superposition of
many configurations. It is only useful for targeting narrow-
band signals with frequency uncertainties of, say, O(10) Hz
in which case we can choose the appropriate element from
the set of configurations used to produce the envelope. Fi-
nally, note that possible designs for the third generation de-
tectors are still being explored, and thus the noise curve for
the Einstein Telescope is much more preliminary.
To start addressing the question of detecting GWs from
LMXBs, we start by asking how strong the signal would
need to be for detection if we knew all source parameters
to sufficient accuracy that only one template was needed
(a coherent fold). Computational cost is no issue, and we
can easily integrate for long periods Tobs. Such a fully co-
herent search with D detectors of comparable sensitivity
is a best case and defines a signal-to-noise ratio (SNR) as
SNR2 = h2
0TobsD/Sn; thus, the SNR-squared builds up lin-
early with the observation time (here Sn(ν) is the power-
spectral density of the detector noise). Conversely, for a
given choice of SNR threshold for detectability, this leads
to a minimum detectable signal amplitude h0 (following the
notation of Jaranowski, Kr´olak & Schutz (1998)):
h0 ≈ 11.4r Sn
DTobs
(6)
which is a useful indicator of the sensitivity of such a search.
8 The various Advanced LIGO noise curves shown here have
been obtained using version 6.2 of the Matlab script "Bench".
It is worth mentioning that a newer version of Bench, v7.0, is
currently under development which includes an improvement in
coating thermal noise; it is probably worth re-calculating the Ad-
vanced LIGO noise curves once this and later versions become
available.
9 To generate the lower envelope of the narrow band noise curves,
we have used Bench to calculate the narrow band noise curves for
a range of choices for a few interferometer parameters. These pa-
rameters are the phase and transmittance of the signal recycling
cavity, and the transmittance the input test mirror and the power
recycling mirror. The envelope is then obtained by calculating the
convex hull of the minima of the various noise curves.
1e-22
)
/
2
1
-
z
H
(
i
D
S
P
e
s
o
N
n
a
r
t
S
i
1e-23
1e-24
ET
E - L I G O
A-LIGO
A-LIGO NB TH
I
- L I G O
V I R G O
O N B EX
A-LIG
100
1000
Frequency (Hz)
I
- L I G O
V I R G O
E - L I G O
1e-25
10
1e-22
)
2
/
1
-
z
H
(
i
D
S
P
e
s
o
N
n
a
r
t
S
i
1e-23
A-LIGO NB TH
A-LIGO
A-LIGO NB ENV
1e-24
ET
1e-25
10
100
1000
Frequency (Hz)
Figure 1. The noise curves for Initial LIGO, Virgo, Advanced
LIGO, and the third generation Einstein Telescope (ET) interfer-
ometers. The Initial, Enhanced and Advanced LIGO curves are
labelled "I-LIGO", "E-LIGO" and "A-LIGO" respectively; "ET"
is Einstein Telescope. There are three curves for Advanced LIGO
in each panel. The top panel shows a nominal broad band config-
uration (the so called "zero de-tuned" configuration), an example
of a narrow band curve ("A-LIGO NB EX"), and the total ther-
mal noise ("A-LIGO TH"), i.e. the sum of the suspension and
mirror thermal noise curves. The lower panel shows the lower en-
velope of the narrow band curves ("A-LIGO NB ENV") instead
of the narrow band example. The thermal noise is shown because
it is sometimes taken as a theoretical lower bound on the narrow
banding for frequencies above, say, 100 Hz; as seen from the lower
panel, this is not a good approximation at higher frequencies.
The factor of 11.4 corresponds to an SNR threshold which
would lead to a single trial false alarm rate of 1% and a
false dismissal rate of 10% (Abbott et al. 2007a) and a uni-
form averaging over all possible source orientations and sky
positons (Jaranowski, Kr´olak & Schutz 1998).
Figure 2 compares the predicted and detectable am-
plitudes for this best-case scenario using the long-term flux
average derived in Section 3 and summarised in Table 1, and
assuming that each NS is in perfect spin balance (so that we
can neglect spin derivatives), with gravitational wave torque
balancing that of accretion. We assume that the spin fre-
quency for the kHz QPO sources lies in the middle of the
known range of separations, and take Tobs = 2 years as a
reference value.
Figure 2 naively implies that the kHz sources are the
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
5
- L I G O
I
E - L I G O
Pulsars
Bursters
kHz QPO
A-LIGO
A-LIGO NB
E T
A-LIGO NB TH
1e-25
1e-26
1e-27
1e-28
0
h
e
d
u
t
i
l
p
m
A
1000
Gravitational wave frequency (Hz)
500
Figure 2. Best case detectability for the mountain scenario for Tobs = 2 years, balancing the long-term average flux, and assuming
that all parameters are known (single template search). The bursters are divided into two groups: those for which the frequency is
confirmed (filled) and those for which the frequency requires confirmation (open), see Section 3. The frequency at which the kHz QPO
symbols appear is derived from the centre of the measured range of separations: the predicted amplitude would be higher if the frequency
were lower, and vice versa. We show detectability thresholds for Initial LIGO (I-LIGO), Enhanced LIGO (E-LIGO), Advanced LIGO
(A-LIGO), and the Einstein Telescope (ET). We also show two detectability curves for Advanced LIGO Narrow Band: the expected
envelope for the narrow band detector that includes all sources of noise (A-LIGO NB), and a curve showing only the thermal noise floor
(A-LIGO NB TH).
most easily detected sources. However, they offer a specific
challenge, as we do not know many of their spins or orbital
parameters. Detecting the GW signal requires a knowledge
of the GW phase evolution which depends crucially on the
orbital parameters. Ignorance or inaccurate knowledge of the
orbit could then require a search over a significant number
of parameters and as we shall see, this can have a dramatic
impact on the sensitivity of the GW search. Hence assessing
the detectability of GWs requires a more careful descrip-
tion of the mechanics of the gravitational wave data analysis
process. Most of this paper is therefore focused on clearly
assessing the data analysis challenges for present and future
generations of GW detectors that takes into account the
limitations imposed by incomplete astrophysical knowledge
and finite computational resources.
3 ACCRETING NEUTRON STAR
PROPERTIES
In this Section we summarize our knowledge of the rel-
evant search parameters for all accreting NSs in LMXBs
where there is some estimate of the spin: the accret-
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
ing millisecond pulsars, the burst oscillation sources, and
the kHz QPO sources. This sample contains both atoll
and Z sources, a classification determined by the spec-
tral and timing properties (Hasinger & van der Klis 1989;
Muno, Remillard & Chakrabarty 2002; Gierli´nski & Done
2002; van der Klis 2006). As explained in Section 2 we re-
quire flux histories, spin frequencies and orbital parameters,
some of which are measured, whilst others can only be esti-
mated. We also need to gauge the uncertainty on each quan-
tity. This sets the required search parameter space, and in
Section 5 we use this information to compute the number of
signal templates required for each source.
This Section is rather lengthy, as we give full details of
the provenance of all of the values used in our study. There
are two main reasons for this: firstly, to make clear the link
between the astronomical observations and the consequences
for gravitational wave searches. Secondly, many of the val-
ues that we derive involve assumptions, or draw on old or
uncertain measurements: where this is the case we wanted
to make it explicit, in order to drive future astrophysical
modelling and observations. The information in this Section
should also be a useful resource for anyone intending to carry
6 Watts et al.
out a search for gravitational waves from these objects, or for
future detectability studies. Readers who are not concerned
with the details of the source properties and uncertainties
can however skip this Section, and will find the key results
summarized in Tables 1 - 4.
3.1 Constraining Fluxes and Accretion Rates
The observed flux, F , sets the expected GW signal strength
(Equation 5). For persistent sources we record only the long-
term average flux, Fav, whereas for transient sources we also
record the outburst flux, Fob (averaged over outbursts and
quiescent periods). Bolometric outburst fluences have been
computed for the accreting millisecond pulsars, but for the
majority of sources this is not the case. For the remaining
sources we determine F using data from the RXTE All Sky
Monitor (ASM), which provides a near-continuous history of
source activity from 1996 to the present in the 2-10 keV band
(Levine et al. 1996). Converting the 2-10 keV flux to a bolo-
metric flux requires detailed spectral modelling. For most
burst oscillation sources we use the results of Galloway et al.
(2008): these authors carry out spectral modelling using the
pointed RXTE PCA data (2.5-25 keV) to estimate bolo-
metric flux. We compare the calculated fluxes to the ASM
countrate at the time of the observation to establish this
relationship. For the transient sources we only include in
our integrated ASM histories the times when the source is
in outburst and detectable above a 3σ level. This avoids
contamination from other sources in the field which would
overestimate the long-term flux. This method of estimating
flux history introduces some errors, since spectral shapes
(and the correction from 2-10 keV flux to bolometric flux)
will change: but it gives a reasonable estimate. This uncer-
tainty should however be borne in mind in Section 6 for
those sources that are on the margins of detectability.
We also record position and distance. Source position
must be known to a certain degree of precision for the long
folds that this type of analysis necessitates (Section 4.1). We
have listed the most accurate and up to date position known
for the X-ray source or its optical, infra-red or radio coun-
terpart. Source distance, which is relevant for the emission
modelling in Section 5, can be estimated in several different
ways. Only for the closest source in our sample, Sco X-1,
can the distance be measured via parallax. For the other
sources different methods are used: location of the source
in a globular cluster; the presence of radius expansion X-ray
bursts (where luminosity reaches the Eddington limit10); in-
ferences about mass transfer from the long-term X-ray flux,
assuming that the binary orbit evolves due to gravitational
wave emission; absorption and spectral modelling.
We give details of pulsar frequency, burst oscillation
frequency, and twin kHz QPO separation, as measured with
RXTE's Proportional Counter Array (PCA). For sources
with only a burst oscillation frequency, we assume that νs
lies within ±5 Hz of the burst oscillation frequency νb. For
the kHz QPO sources we assume that the spin lies within
10 The Eddington limit depends on the composition. Sometimes
composition can be inferred from burst properties, but this is
not always the case, leading to additional uncertainty. See also
Galloway, Ozel & Psaltis (2008).
the reported range of kHz QPO separations. We review both
of these assumptions in more detail in later Sections.
3.2 Orbital Uncertainties
We must know (or presume) the orbital parameters in or-
der to fold long stretches of gravitational wave data, most
importantly, the orbital period Porb and eccentricity, e. We
also need a reference time within the orbit. Depending on
the original reference we quote either T90, the time of infe-
rior conjunction of the companion star, or Tasc, the time of
ascending node, when the Doppler shifted frequency of the
neutron star is at its lowest. Note that T90 = Tasc + Porb/4.
The third parameter is the projected semi-major axis of the
neutron star, ax sin i, which we denote ap. Depending on
the measurement, we may instead quote the amplitude of
the projected orbital velocity of the neutron star, vx sin i
(referred to as K1 in the optical literature). The two quanti-
ties are related by vx sin i = 2πax sin i/Porb. The orbital pa-
rameters are measured directly for the accreting millisecond
pulsars via X-ray timing. The situation is more challenging
for the non-pulsing and intermittent sources.
Some high inclination systems show eclipses in the X-
ray lightcurve, providing both Porb and Tasc. There are also
systems that show dips rather than full eclipses: the dips
occur when the NS is obscured by the bulge where the
mass stream from the donor star joins the accretion disk
(White & Swank 1982). Dips certainly tell us Porb, and re-
strict the inclination to lie in the range 60◦−75◦. What they
do not necessarily yield is Tasc: in the two systems that show
both dips and eclipses, dips occur at various offsets from the
eclipse times (Cominsky & Wood 1984; Parmar et al. 1986;
Motch et al. 1987; Smale et al. 1992). Detailed modelling
is therefore required to determine the relationship between
Tasc and the dip time Tdip. Throughout this Section we list
the most recent orbital ephemeris for each system. In sev-
eral cases the ephemerides are sufficiently out of date that we
should consider Tasc to be totally unconstrained. However,
we presume that it would be straightforward to obtain a new
ephemeris, with an error no worse that that of the existing
measurement. We therefore use all existing measurements in
our initial assessment of detectability.
The orbital parameters can also be measured in wave-
bands other than the X-ray, particularly the optical. A num-
ber of systems show photometric variability at the orbital
period. Maximum optical light occurs when the NS is at
inferior conjunction and we observe reprocessed emission
from the heated face of the donor star, yielding both Porb
and Tasc, but not the projected semi-major axis. An alter-
native method that can provide all three pieces of infor-
mation is phase-resolved optical spectroscopy. LMXBs ex-
hibit many emission lines, some from heated face of the
donor star (Steeghs & Casares 2002), others from the ac-
cretion disk very close to the compact object. By mea-
suring the orbital Doppler shifts of these lines, and using
techniques such as Doppler tomography to check the emis-
sion location, major progress has been made in comput-
ing orbital parameters. Additional constraints on the sys-
tem are possible (assuming Roche lobe overflow and tidal
locking) if one can detect rotational broadening of absorp-
tion lines from the donor star (Horne, Wade & Szkody 1986;
Casares, Charles & Kuulkers 1998).
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
7
Unfortunately, for many of the systems of interest, one
or all of the orbital parameters are unknown. We can how-
ever still place various constraints on the systems to reduce
the number of templates required. For those systems with
measured orbital periods, for example, we can assume that
the donor star fills its Roche lobe. This fixes ¯ρd, the mean
density of the donor star:
¯ρd ≈ 110„ Porb
1hr «−2
g cm−3.
(7)
be persistent or transient at the inferred accretion rate were
also considered (King, Kolb & Burderi 1996; Dubus et al.
1999; in't Zand, Jonker & Markwardt 2007): unfortunately
for most of the systems in our study this added very little
in the way of tighter constraints. For systems where there
are no better constraints on orbital period, we will assume
that Porb lies between 10 minutes and 240 hours. The pre-
sumed lack of eclipses in such systems sets i < 60◦, so from
Equation (8) we obtain an upper limit on ax sin i of 22.6 lt-s.
∼ 10 hours, ¯ρd <
∼ 1 g cm−3, the companion
∼ 80 minutes,
The range of donor types with this ¯ρ limits the donor mass
Md = mdM⊙. LMXB donors include main sequence stars,
evolved hydrogen-burning stars, helium-burning stars, and
brown or white dwarfs (Podsiadlowski, Rappaport & Pfahl
2002). For Porb >
must be an evolved hydrogen-burning star such as a sub-
giant. For ultracompact systems with Porb <
the companion must be a white dwarf, a helium star, or
a highly evolved helium-rich secondary (Rappaport & Joss
1984; Deloye & Bildsten 2003; Nelemans, Jonker & Steeghs
2006). For intermediate orbital periods more options are pos-
sible, maximum donor mass being by the main sequence star
- but evolved, less massive companions or even brown dwarfs
may be possible (Tout et al. 1996; Chabrier & Baraffe 2000;
Chabrier et al. 2000; Bildsten & Chakrabarty 2001)11. Ad-
ditional constraints on donor properties may come from
X-ray burst properties or spectral type: more evolved,
lower mass, donors will have later spectral types than
the main sequence star with the same ¯ρd (Baraffe & Kolb
2000; Kolb, King & Baraffe 2001). Having finally estab-
lished the range of likely donor masses, and knowing that
the NS mass Mx = mxM⊙ is in the range 1.2 − 2.4M⊙
(Lattimer & Prakash 2007), we then estimate ax sin i using
ax sin i = 1.174„ Porb
1hr «2/3
md
(md + mx)2/3 sin i lt − s . (8)
∼ 60◦),
For systems that do not show dips or eclipses (i <
and where there is no other limit on inclination, Equation
(8) gives only an upper limit. For dipping systems we as-
sume 60◦ <
For eclipsing systems an additional constraint comes from
the half-angle of the X-ray eclipse θx (Bradt & McClintock
1983):
∼ 75◦, giving both upper and lower limits.
∼ i <
sin i =
1
cos θx "1 −„0.38 − 0.2 log
mx
md«2#1/2
.
(9)
For systems with no measured orbital period, we make
the standard LMXB assumption that md/mx < 0.8. Time-
averaged accretion rate (as estimated from the X-ray flux)
can give general constraints if we assume that mass transfer
is driven by gravitational radiation: however magnetic brak-
ing may also play a role in mass transfer and the contribution
is hard to quantify. The conditions required for the system to
11 Rappaport & Joss (1984) showed that helium-rich donor stars
would have higher masses for a given Porb than expected for
main sequence stars. However, the binary evolution simulations
of Podsiadlowski, Rappaport & Pfahl (2002) did not generate any
such stars at intermediate orbits, so we neglect this possibility un-
less there is overwhelming evidence for the presence of helium-rich
material (from X-ray bursts, for example).
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
3.3 Accreting millisecond pulsars
The ten accreting millisecond pulsars are the only systems
where we clearly know the NS spin frequency νs. All are
transient, but their pulsation characteristics differ. Seven
of these systems, which have short outbursts (weeks) and
long periods of quiescence (years), show persistent pulsa-
tions throughout their outburst phases. The other three sys-
tems are rather different. HETE J1900.1-2455 went into out-
burst in 2005 and is still active, but only showed pulsations
during the first two months of the outburst. The other two
systems have shown pulsations only intermittently during
outburst. We now discuss the sources in order of spin fre-
quency, from highest to lowest.
3.3.1 IGR J00291+5934 (νs = 599 Hz)
This source was discovered in 2004 during a 13 day out-
burst with total fluence (0.1-200 keV) 1.8 × 10−3 ergs cm−2
(Galloway et al. 2005). We assume that this is a typical
outburst. The RXTE ASM history suggests a recurrence
time of 3 years, giving Fav = 1.8 × 10−11 ergs cm−2 s−1.
The most accurate position for the source, given by the op-
tical/near infra-red counterpart, is RA = 00h29m03.s05 ±
0.s01, Dec = +59◦34′18.′′93 ± 0.′′05 (Torres et al. 2008).
This position accords with the earlier optical position of
(Fox & Kulkarni 2004) and the Chandra X-ray (Paizis et al.
2005), but is offset by 3.2σ in RA from the radio posi-
tion (Rupen, Dhawan & Mioduszewski 2004). Mass trans-
fer arguments suggest a minimum distance of ≈ 4 kpc
(Galloway et al. 2005), and the lack of bursts implies a
maximum distance of 6 kpc (Galloway 2006). There have
been three studies of the spin and orbital parameters.
Galloway et al. (2005) used data from the start of the out-
burst and the optical position of Fox & Kulkarni (2004).
Falanga et al. (2005); Burderi et al. (2007) used additional
data from later in the outburst and the radio position12. The
studies agree on the orbital parameters: Porb = 8844.092 ±
0.006 s, T90 = 53345.1875164 ± 4 × 10−7 MJD (TDB),
ax sin i = 64.993 ± 0.002 lt-ms (1σ uncertainties, Galloway,
private communication), and e < 2× 10−4 (3σ upper limit).
There are however small differences in spin parameters. In
our initial assessment of detectability we use the most recent
values of Burderi et al. (2007): νs = 598.89213053±2×10−8
Hz, νs = (8.5±1.1)×10−13 Hz/s at epoch MJD 53346.184635
(1σ uncertainties).
12 Although an offset position can generate an apparent spin
derivative, the offset is not sufficiently large to account for the
reported values of νs.
8 Watts et al.
3.3.2 Aql X-1 (1908+005) (νs = 550 Hz)
Aql X-1 is a transient atoll source with quasi-regular out-
bursts, ∼ 10 in the RXTE era. Using the spectral modelling
of Galloway et al. (2008) we set 5 ASM cts/s = 3.1 × 10−9
ergs cm−2 s−1 (bolometric). The long-term average ASM
countrate is 1.8 cts/s, while the average countrate during the
bright 45 day outburst of late 2000 was 34 cts/s. The most
recent position, from VLA observations of the radio counter-
part, is RA = 19h11m16.s01± 0.s03, Dec = +00◦35′06.′′7± 0.′′3
(J2000, 1σ error bars) (Rupen, Mioduszewski & Dhawan
2004). Galloway et al. (2008) use PRE bursts to estimate
a distance of 3.9 ± 0.7 kpc or 5.0 ± 0.9 kpc depending
on composition. Casella et al. (2008) have reported the
detection of intermittent accretion-powered pulsations at
550.2745 ± 0.0009 Hz. The orbital period, determined from
the optical lightcurve in outburst, is Porb = 18.9479±0.0002
hours (Chevalier & Ilovaisky 1998) (the quiescent period
agrees at the 0.02% level). The time of minimum opti-
cal light determined by Garcia et al. (1999) gives T90 =
2450282.220 ± 0.003 HJD. Both time of minimum light
and orbital period were confirmed in later analysis by
Welsh, Robinson & Young (2000). Cornelisse et al. (2007)
attempted to measure the projected orbital velocity of
the neutron star directly using phase-resolved optical spec-
troscopy: they report a preliminary value of vx sin i = 68± 5
km/s. Although the fit quality is not good and this result re-
quires confirmation, we will use this value in our preliminary
assessment of detectability.
3.3.3 SAX J1748.9-2021 (νs = 442 Hz)
This is a transient source located in the globular cluster
NGC 6440, with three outbursts in the RXTE era. Using
the spectral modelling of Galloway et al. (2008) we set 13
ASM cts/s = 5 × 10−9 ergs cm−2 s−1 (bolometric). The
long-term average ASM countrate is 0.2 cts/s, while the
average countrate during the bright 60 day outburst of
2005 was 10 cts/s. The position of the optical counterpart
(Verbunt et al. 2000) (which accords with the Chandra po-
sition of Pooley et al. (2002)) is RA = 17h48m52.s14, Dec =
-20◦21′32.′′6 (J2000), with an error of 0.′′5. Kuulkers et al.
(2003) estimate distance to the cluster as 8.4+1.5
−1.3 kpc, in
good agreement with the value of 8.1± 1.3 kpc derived from
PRE bursts (Galloway et al. 2008). Altamirano et al. (2008)
discovered intermittent accretion-powered pulsations in two
outbursts from this source (see also Gavriil et al. (2007)).
Patruno et al. (2008) have now carried out a detailed phase-
connected timing study to determine the spin and orbital pa-
rameters. They find νs = 442.36108118±5×10−8 Hz, Porb =
8.76525 ± 3 × 10−5 hours, Tasc = 52191.507190 ± 4 × 10−6
MJD/TDB, and ax sin i = 0.38760 ± 4 × 10−5 lt-s, with
e < 1.3 × 10−4 (1σ uncertainties and upper limits).
3.3.4 XTE J1751-305 (νs = 435 Hz)
The 2004 outburst of this source, which lasted ≈ 10 days,
had total fluence (2-200 keV) (2.5 ± 0.5) × 10−3 ergs cm−2
(Markwardt et al. 2002). Assuming that this outburst is typ-
ical, the mean recurrence time of 3.8 years yields Fav =
2 × 10−11 ergs cm−2 s−1. The most accurate position for
the source, measured with Chandra, is RA = 17h51m13.s49±
0.s05, Dec = −30◦37′23.′′4±0.′′6 (J2000), where the uncertain-
ties are the 90% confidence limits (Markwardt et al. 2002).
Mass transfer arguments yield a lower limit on the distance
of 6 kpc (Markwardt et al. 2002). The most recent timing
study by Papitto et al. (2008), using the Chandra position,
gives Porb = 2545.342 ± 0.002 s, Tasc = 52368.0129023 ±
4 × 10−7 MJD (TDB), ax sin i = 10.125 ± 0.005 lt-ms, e <
1.3×10−3 (90% confidence level uncertainties and upper lim-
its). The spin parameters are νs = 435.31799357 ± 4 × 10−8
Hz, and the study also suggests that νs may be non-zero.
3.3.5 SAX J1808.4-3658 (νs = 401 Hz)
Galloway & Cumming (2006) analyse the five known out-
bursts from this source, and find a mean outburst fluence
of (6.0 ± 0.2) × 10−3 ergs cm−2 (0.1-200 keV). In estimat-
ing transient gravitational wave signal we use an outburst
duration of 20 days: we neglect the extended flaring phase
often seen at the end of outbursts in this source, since emis-
sion during this phase is at a much lower level. Mean re-
currence time is 2.2 years, giving Fav = 8.6 × 10−11 ergs
cm−2 s−1. The most recent and precise position, for the op-
tical counterpart, is RA = 18h08m27.s62, Dec = -36◦58′43.′′3
(J2000), with an uncertainty of 0.′′15 (Hartman et al. 2008).
Galloway & Cumming (2006) have derived a distance of 3.4-
3.6 kpc using both mass transfer arguments and burst prop-
erties. Using the refined optical position, Hartman et al.
(2008) report the following values for the orbital param-
eters: Porb = 7249.156961 ± 1.4 × 10−5 s at time Tasc =
Porb = (3.48 ±
52499.9602477 ± 1.0 × 10−6 MJD (TDB),
0.12) × 10−12 Hz/s, ax sin i = 60.28132 ± 2.4 × 10−4 lt-ms
(1σ errors). The eccentricity e < 0.00021 (95% upper limit).
The spin rate was tracked across multiple outbursts, and is
given relative to a reference frequency ν0 = 400.975210 Hz.
In the 1998 outburst, νs − ν0 = 0.371 ± 0.018 µHz, with νs
in the range (-7.5, 7.3)×10−14 Hz/s (95% confidence lim-
its). In the 2000 outburst, νs − ν0 = 0.254 ± 0.012 µHz,
with νs in the range (-1.1, 4.2)×10−14 Hz/s. In the 2002
outburst, νs − ν0 = 0.221 ± 0.006 µHz, with νs in the range
(-1.2, 2.5) ×10−14 Hz/s. In the 2005 outburst, νs − ν0 =
0.190 ± 0.015 µHz with νs in the range (-0.5, 2.3)×10−14
Hz/s. Fitting the frequency evolution across all four out-
bursts gives νs = (5.6 ± 2.0) × 10−15 Hz/s. In our initial
assessment of detectability, we use the spin solution for the
2002 outburst.
3.3.6 HETE J1900.1-2455 (νs = 377 Hz)
This source was first detected in June 2005, and has
remained in outburst ever since (Degenaar et al. 2007).
Galloway (2006) and Galloway et al. (2007a) report an av-
erage (bolometric) outburst flux of ≈ 9 × 10−10 ergs cm−2
s−1. We assume an outburst duration of 2 years, and (given
that no previous outbursts are known) a recurrence time
of at least 10 years (Kaaret et al. 2006), yielding Fav ≈
2 × 10−10 ergs cm−2 s−1. The position of the optical coun-
terpart, is RA = 19h00m08.s65, Dec = -24◦55′13.′′7 (J2000),
with an estimated uncertainty of 0.′′2 (Fox 2005). The dis-
tance estimated by Galloway et al. (2008) using RXTE ob-
servations of radius expansion bursts,
is 4.7 ± 0.6 kpc.
This accords with the earlier estimate of 5 kpc made by
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
9
Kawai & Suzuki (2005) using HETE burst data. Timing
analysis by Kaaret et al. (2006), using RXTE data from
June 16 - July 7 2006, resulted in the following orbital pa-
rameters: Porb = 4995.258 ± 0.005 s, T90 = 53549.145385 ±
7× 10−6 MJD (TT), ax sin i = 18.41± 0.01 lt-ms, e < 0.002.
The spin was νs = 377.296171971 ± 5 × 10−9 Hz. All errors
and upper limits are 1σ uncertainties. On July 8th there
was an apparent jump in spin rate of ∆νs/νs ∼ 6 × 10−7,
to νs = 377.291596 ± 1.6 × 10−5 Hz. Thereafter pulsations
ceased and spin has not been tracked since despite the fact
that the source has remained in outburst (Galloway et al.
2007a). In our initial assessment of detectability we do not
take into account the apparent jump in spin.
3.3.7 XTE J1814-338 (νs = 314 Hz)
This pulsar has had only one outburst in the RXTE era,
lasting ≈ 50 days. Galloway (2006) estimate a bolometric
outburst fluence of (3.0 ± 0.1) × 10−3 ergs cm−2 s−1. Given
a recurrence time of at least 7.5 years, this yields an up-
per limit on Fav of 1.3 × 10−11 ergs cm−2 s−1. The posi-
tion derived from X-ray and optical spectroscopy is RA =
18h13m39.s04, Dec = -33◦46′22.′′3 (J2000), 90 % confidence
error circle of 0.′′2 (Krauss et al. 2005). These authors use
X-ray spectroscopy to infer a minimum distance of 3.8 kpc.
The upper limit on the distance, derived from burst proper-
ties, is 8.0± 1.6 kpc (Strohmayer et al. 2003). Timing analy-
sis by Papitto et al. (2007), using data from the whole 2003
outburst, leads to the following orbital parameters: Porb =
15388.7229±0.0002 s, Tasc = 52797.8101698±9×10−7 MJD
(TDB), ax sin i = 390.633 ± 0.009 lt-ms, e < 2.4 × 10−5
(3σ upper limit). The associated spin parameters are νs =
314.35610879± 1× 10−8 Hz, νs = (−6.7± 0.7)× 10−14 Hz/s.
Uncertainties are at the 90% confidence level.
3.3.8 XTE J1807-294 (νs = 191 Hz)
There has been only one recorded outburst from this source,
for which Galloway (2006) computes a bolometric fluence of
(3.1 ± 0.2) × 10−3 ergs cm−2. For a recurrence time of at
least 7.1 years, Fav is at most 1.4 × 10−11 ergs cm−2 s−1.
In estimating transient gravitational wave signal we use an
outburst duration of 50 days, neglecting the prolonged low
flux tail at the end of the outburst. The most accurate posi-
tion, measured with Chandra, is RA = 18h06m59.s8, Dec =
-29◦24′30′′ (J2000), with an uncertainty due to systematic
errors of 1′′ (Markwardt, Juda & Swank 2003). Using mass
transfer estimates, Galloway (2006) derives a lower limit on
the distance of 4.7 kpc. Determination of spin and orbital
parameters in this source is complicated by extreme varia-
tions in pulse profile. A recent study by Riggio et al. (2007)
finds Porb = 2404.41665 ± 0.00040s, ax sin i = 4.819 ± 0.004
lt-ms, Tasc = 52720.675603 ± 6 × 10−6 MJD/TDB and
νs = 190.62350694 ± 8 × 108 Hz (1σ uncertainties). The
eccentricity e < 0.0036 (2σ upper limit). Chou et al. (2008)
report similar values apart from for Tasc, where the values
found by the two studies differ by more than the quoted un-
certainties. Both Chou et al. (2008) and Riggio et al. (2008)
suggest a non-zero νs ∼ 10−14 − 10−13 Hz/s in outburst.
In our initial assessment of detectability we use the values
and uncertainties of Riggio et al. (2007), although clearly
the 'true' uncertainty on Tasc is larger.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
3.3.9 XTE J0929-314 (νs = 185 Hz)
This source has had one outburst in the RXTE era, for
which Galloway (2006) estimates a bolometric fluence of
(5.4 ± 0.3) × 10−3 ergs cm−2. Outburst duration, which we
assume to be typical, was ≈ 60 days. Given a recurrence time
of at least 6.3 years, Fav is at most 2.7 × 10−11 ergs cm−2
s−1. The most accurate position, given by the optical coun-
terpart, is RA = 9h29m20.s19, Dec = -31◦23′03.′′2 (J2000)
with error circle 0.′′1 (Giles et al. 2005). Galloway (2006)
uses mass transfer and recurrence time estimates to infer
a lower limit to the distance of 3.6 kpc (revising an earlier
value of 5 kpc in Galloway et al. (2002)). Timing analysis by
Galloway et al. (2002), using an earlier optical position from
Giles et al. (2002), leads to the following orbital parameters:
Porb = 2614.746 ± 0.003 s, T90 = 52405.49434 ± 1 × 10−5
MJD (TDB), ax sin i = 6.290 ± 0.009 lt-ms. The associated
spin parameters are νs = 185.105254297 ± 9 × 10−9 Hz,
νs = (−9.2 ± 0.4) × 10−14 Hz/s. All errors are 1σ uncer-
tainties (Galloway, private communication). The eccentric-
ity e < 0.007 (2σ limit).
3.3.10 SWIFT J1756.9-2508 (νs = 182 Hz)
This source has had one outburst in 2006 lasting 13 days,
with total fluence (1-10000 keV) (4.5±0.8)×10−4 ergs cm−2.
No previous outbursts are known (although there are gaps
in coverage), but this suggests a recurrence time of at least
10 years. The best position for the source, from SWIFT, is
RA = 17h56m57.s35, Dec = -25◦06′27.′′8 (J2000), with un-
certainty 3.′′5. The distance is not well constrained, but is
thought to be ≈ 8 kpc. The orbital parameters are Porb =
3282.104 ± 0.083 s, Tasc = 54265.28707 ± 6 × 10−5 MJD
(TDB), ax sin i = 5.942± 0.027 lt-ms, e < 0.026 (95% upper
limit). The spin rate is νs = 182.065804253 ± 7.2×10−8 Hz,
with νs < 1× 10−12 Hz/s. All errors are 90% confidence, up-
per limits are 95% confidence. All information on this source
is taken from Krimm et al. (2007).
3.4 Burst oscillation sources
We list all sources for which burst oscillations have been
reported, in order of decreasing burst oscillation frequency
νb. For some sources, oscillations have been detected at the
same frequency in multiple bursts: these results can be re-
garded as secure. In some cases, however, oscillations have
only been seen in a single burst. Given the number of NSs
whose bursts have now been searched for oscillations (a fac-
tor not included in quoted statistical significances), these
results should be regarded as tentative until confirmed in a
second burst. We include them in our survey (marked with
an asterisk) since if they turn out to be promising gravi-
tational wave sources this would provide added impetus to
confirm or alternatively rule out the candidate burst oscil-
lation detection.
3.4.1 XTE J1739-285 (νb = 1122 Hz)*
This transient atoll source has had four outbursts in the
RXTE era. Using the spectral modelling of Kaaret et al.
(2007) we set 10 ASM cts/s = 1.4× 10−9 (2-20 keV) and ap-
ply a bolometric correction factor of 1.34 (the average factor
10 Watts et al.
found by Galloway et al. (2008) for burst sources). The long-
term average ASM countrate is 0.3 cts/s, and the average
countrate during the bright 200 day outburst of 2005 was 4
cts/s. The most precise source position, measured by Chan-
dra, is RA = 17h39m53.s95, Dec = -28◦29′46.′′8 (J2000), with
a 90% error radius of 0.′′6 (Krauss et al. 2006). The absence
of an optical/IR counterpart sets an upper limit on distance
of 12 kpc (Torres et al. 2006). The absence of PRE bursts
sets a more stringent upper limit of 10.6 kpc (Kaaret et al.
2007). A candidate 1122 Hz burst oscillation was detected in
part of one burst recorded by RXTE during the 2005 out-
burst. The orbital parameters of the source are unknown,
and there are no measured constraints on the properties of
the companion.
3.4.2 4U 1608-522 (νb = 620 Hz)
This transient atoll source has had several outbursts dur-
ing RXTE's lifetime, and seems to be active at a low level
(ASM countrate > 3σ) even when not in outburst. Using
the spectral modelling of Galloway et al. (2008) we set 70
ASM cts/s = 4.8 × 10−8 ergs cm−2 s−1 (bolometric). The
long-term average ASM countrate is 3.6 cts/s, while the av-
erage countrate during the bright 100 day outburst in 2005
was 30 cts/s. The best position, from the optical counter-
part, is RA = 16h08m52.s2, Dec = -52◦17′43′′ (B1950), with
errors of ±0.′′5 (Grindlay & Liller 1978). No more up to date
position is available. Source distance, derived under the as-
sumption that radius expansion bursts reach the Edding-
ton limit for pure helium, is 4.1 ± 0.4 kpc (Galloway et al.
2008). Burst oscillations at ≈ 620 Hz have been detected in
multiple bursts (Hartman et al. 2003; Galloway et al. 2008).
The last (unsuccessful) search for persistent pulsations from
this source, using data from 1989-1991, searched only up
to 512 Hz (Vaughan et al. 1994). The orbital parameters
are not known, although a number of tentative periodic-
ities have been reported (Lochner & Roussel-Dupre 1994;
Wachter et al. 2002). The spectral type of the companion is
that of a late F/early G type main sequence companion star,
but would also match that of a more evolved K/M type star
(Wachter et al. 2002). The mean density range for donors of
this type suggests that the orbital period lies in the range
10 - 125 hours. Given that the system is non-eclipsing we
assume i < 60◦ and hence obtain, for the assumed orbital
periods, ax sin i < 12.3 lt-s.
3.4.3 SAX J1750.8-2900 (νb = 601 Hz)
This weak transient atoll source has had two outbursts
in the RXTE era. Following the spectral modelling of
Galloway et al. (2008) we set 8 ASM cts/s = 3.4 × 10−9
ergs cm−2 s−1 (bolometric). The long-term average ASM
countrate is 0.08 cts/s, while the average countrate during
the brighter 110 day outburst in 1997 was 2.8 cts/s. The
most precise position, from BeppoSAX, is RA = 17h50m24s,
Dec = -29◦02′18′′ (J2000), with a 99% error radius of 1′
(Natalucci et al. 1999). The distance, estimated from radius
expansion bursts, is 6.79 ± 0.14 kpc (Galloway et al. 2008).
Burst oscillations at 601 Hz have been detected in multi-
ple bursts (Kaaret et al. 2002; Galloway et al. 2008). The
orbital parameters are not known, and there are no obser-
vations of the companion star.
3.4.4 GRS 1741.9-2853 (AX J1745.0-2855) (νb = 589
Hz)
This is a transient source in the crowded Galactic Centre,
with three outbursts in the RXTE era. Unfortunately the
fact that the field is crowded means that there is no reliable
ASM flux history for this source. The most precise position
for the source, derived by Chandra, is RA = 17h45m2.s33,
Dec = -28◦54′49.′′7 (J2000), with an uncertainty of 0.′′7
(Muno, Baganoff & Arabadijs 2003). Analysis of radius ex-
pansion bursts suggests a distance 6.0± 1.6 kpc (Eddington
limit assuming cosmic abundances) or 8± 2 kpc (Eddington
limit for pure He) (Galloway et al. 2008). Burst oscillations
at 589 Hz have been detected in 2 bursts (Strohmayer et al.
1997)13. The orbital parameters are not known, and there is
no information on the properties of the companion star.
3.4.5 4U 1636-536 (νb = 581 Hz)
This is a persistent atoll source. Using the spectral mod-
elling of Galloway et al. (2008) we set 18 ASM cts/s =
8.4 × 10−9 ergs cm−2 s−1 (bolometric). The long-term av-
erage ASM countrate is 10 cts/s. The best position for
the source, from the optical counterpart V801 Ara, is RA
= 16h36m56.s41, Dec = -53◦39′18.′′1 (B1950), with an er-
ror circle of less than 1′′ (Bradt & McClintock 1983). No
more up to date position is available. The distance, esti-
mated from a large sample of radius expansion bursts, is
6.0 ± 0.5 kpc (Galloway et al. 2006). Burst oscillations at
≈ 581 Hz have been seen in multiple bursts and a super-
burst (Strohmayer et al. 1998a; Strohmayer & Markwardt
2002; Galloway et al. 2008). The most recent unsuccess-
ful search for persistent pulsations, using data from 1987,
searched only up to 512 Hz (Vaughan et al. 1994). The
latest ephemeris, derived from phase-resolved optical spec-
troscopy, gives Porb = 0.15804693 ± 1.6 × 10−7 days and
T90 = 2452813.531 ± 0.002 HJD (Casares et al. 2006).
Augusteijn et al. (1998) set an upper limit on the orbital
period derivative of Porb/ Porb 6 3 × 105 years. No tighter
limit has yet been reported. Orbital Doppler shifts on burst
oscillations during a superburst lead to limits 90 < vx sin i <
113 km/s, (Strohmayer & Markwardt 2002; Casares et al.
2006). The quoted range corresponds to varying the refer-
ence phase of the ephemeris across the ±1σ range.
3.4.6 X 1658-298 (MXB 1659-29) (νb = 567 Hz)
This transient source, which has eclipses and dips, was ac-
tive from 1976-1979, and then again from 1999-2001. Us-
ing the spectral modelling of Galloway et al. (2008) we
set 2.4 ASM cts/s = 6.7 × 10−9 ergs cm−2 s−1 (bolo-
metric). Over the most recent outburst, which lasted 870
days, the average ASM countrate was 2.7 cts/s. Assum-
ing a recurrence time of ≈ 23 years, the long-term aver-
age ASM countrate is 0.3 cts/s. The best position for the
source, from the optical counterpart V2134 Oph, is RA =
17h02m06.s42, Dec = -29◦56′44.′′33 (J2000), with an uncer-
tainty of 0.′′1 (Wachter & Smale 1998). Assuming that the
13 The bursts were originally attributed to a different source,
MXB 1743-29, due to source confusion in the crowded field
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
11
bright radius expansion bursts reach the Eddington limit for
pure He, Galloway et al. (2008) derive a distance of 12 ± 3
kpc. Burst oscillations at ≈ 567 Hz have been detected
in multiple bursts (Wijnands, Strohmayer & Franco 2001).
The most up to date X-ray ephemeris, by Oosterbroek et al.
(2001), gives Porb = 0.2965045746 ± 3.4 × 10−9 days,
and T90 = 2443059.225826 ± 0.000093 JD/TDB. The pro-
jected semi-major axis has not been measured, but we
can set bounds on it using the constraints outlined at
the start of this section. For the above orbital period,
Wachter, Smale & Bailyn (2000) find X-ray eclipse dura-
tion half-angles 6.34◦ ± 0.01◦ (slightly lower than the value
reported by Cominsky & Wood (1984)). We can therefore
use Equation (9) to restrict sin i. If we assume Roche
lobe overflow, then from Equation (7) the mean density
of the donor star is 2.2 g/cm3, suggesting a main se-
quence or evolved companion. Maximum donor mass oc-
curs if the donor is on the main sequence. In this case,
maximum Md = 0.78M⊙ and would be of spectral type
K0 (Wachter & Smale 1998). Whilst the data are consis-
tent with this spectral type, there is some indication that
the spectral type is later, suggesting a more evolved (lower
mass) companion with spectral type perhaps as late as M2
(Wachter, Smale & Bailyn 2000). We will therefore adopt a
minimum companion mass of 0.1M⊙ (Baraffe & Kolb 2000;
Podsiadlowski, Rappaport & Pfahl 2002). For the neutron
star we will consider masses in the range 1.2 − 2.4M⊙
(Lattimer & Prakash 2007). These assumptions suggest that
ax sin i lies in the range 0.24 -- 2.03 lt-s.
3.4.7 A 1744-361 (νb = 530 Hz)*
This transient dipping source has irregular outbursts, three
in RXTE's lifetime. Based on the spectral modelling of
Bhattacharyya et al. (2006) we set 1.2 ASM cts/s = 3 ×
10−10 ergs cm−2 s−1 (3 - 14 keV) flux and apply a bolo-
metric correction factor of 1.34 (the mean correction factor
for converting 2.5 - 25 keV flux to bolometric flux found by
Galloway et al. (2008) for other burst sources). The long-
term average ASM countrate is 0.16 cts/s, and during the
bright 100 day outburst in 2003 the average countrate was
6 cts/s. The best position, for the radio counterpart, is RA
= 17h48m13.s148 ± 0.s014, Dec = -36◦07′57.′′02 ± 0.′′3 (J2000)
(Rupen, Dhawan & Mioduszewski 2003). This is within the
error circle of the Chandra position (Torres et al. 2004).
The lack of radius expansion in the one burst detected
by RXTE sets an upper limit on the distance of ≈ 9 kpc
(Bhattacharyya et al. 2006). These authors reported a burst
oscillation at ≈ 530 Hz during the rising phase of the one
burst detected by RXTE. They also reported a possible or-
bital period of 97 ± 22 minutes, traced by dips in the X-ray
lightcurve, (Bhattacharyya et al. 2006). In our initial assess-
ment we will assume that this is indeed the orbital period. In
this case the assumption of Roche lobe overflow (Equation
7) gives a mean donor density of 28 − 70 g/cm3, consistent
with a main sequence star or slightly evolved donor. The
maximum donor mass (for a star on the main sequence) is
0.22 M⊙. For a dipping source we can assume an inclination
in the range 60 − 75◦. For neutron star masses in the range
1.2 − 2.4M⊙ we can therefore set an upper limit on ax sin i
of 0.2 lt-s. The minimum donor mass, of 0.07M⊙, is set by
the most evolved hydrogen burning star possible. Together
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
with the inclination constraint, this gives a lower limit on
ax sin i of 0.05 lt-s.
3.4.8 KS 1731-260 (νb = 524 Hz)
This transient atoll source was in outburst from 1988 (and
possibly earlier) until 2001. Based on the spectral modelling
of Galloway et al. (2008) we set 18 ASM cts/s = 9.7 × 10−9
ergs cm−2 s−1 (bolometric). The average countrate while
the source was still in outburst was 9 cts/s: the long-term
average countrate over the RXTE lifetime is 4 cts/s. The
most precise position, for the optical counterpart, is RA
=17h34m13.s47, Dec = -26◦05′18.′′8, with an error of 0.′′4
(Wijnands et al. 2001). This is within the error circle of the
Chandra position (Revnivtsev & Sunyaev 2002). Assuming
that the radius expansion bursts reach the limit for pure
He, Galloway et al. (2008) derive a distance of 7.2± 1.0 kpc.
Burst oscillations at ≈ 524 Hz have been detected in multi-
ple bursts (Smith, Morgan & Bradt 1997; Muno et al. 2000;
Galloway et al. 2008). The orbital parameters have not been
measured. A study of the scatter of asymptotic burst oscil-
lation frequencies by Muno et al. (2000) suggested that the
vx sin i might be as high as 340± 100 km/s, but this has not
been revisited. Identification of the counterpart was ham-
pered by high reddening along the galactic plane, but it has
now been detected (Mignani et al. 2002). If the companion
is on the main sequence, it has to be of spectral type later
than F: if it has evolved off the main sequence then it is not a
red giant. For this to be the case in a Roche lobe overflowing
system we require Porb > 2 hours.
3.4.9 4U 0614+09 (νb = 415 Hz)*
the IR counterpart,
This is a bursting atoll source, persistent but highly vari-
able. Based on spectral modelling by Ford et al. (2000),
s−1
we set 9 ASM cts/s = 3.3 × 10−9 ergs cm−2
(bolometric). The long-term average ASM countrate is
from a
3.3 cts/s. The best and most recent position,
Spitzer observation of
is RA =
6h17m07.s35 ± 0.s03, Dec = +09◦08′13.′′60 ± 0.′′05 (J2000)
(Migliari et al. 2006)14. Brandt et al. (1992) infer an up-
per limit on the distance of 3 kpc from an X-ray burst.
Strohmayer, Markwardt & Kuulkers (2008) detected burst
oscillations at 415 Hz in one burst recorded by the
SWIFT Burst Alert Telescope. The orbital parameters
are unknown, but this is a candidate ultra-compact bi-
nary (Juett, Psaltis & Chakrabarty 2001). Nelemans et al.
(2004) have shown that the companion is most likely
a C/O white dwarf. Using the white dwarf models of
Deloye & Bildsten (2003) this would imply Porb = 15-20
minutes and hence (assuming that i < 60◦ due to the lack
of dips and eclipses), ax sin i 6 0.014 lt-s.
3.4.10 4U 1728-34 (GX 354+00) (νb = 363 Hz)
This is a persistent atoll source. Based on the spectral
modelling of Galloway et al. (2008) we set 4 ASM cts/s =
14 This is consistent with the older position for the optical coun-
terpart V1055 Ori (Bradt & McClintock 1983).
12 Watts et al.
1.2 × 10−9 ergs cm−2 s−1 (bolometric). The long-term av-
erage ASM countrate is 7.3 cts/s. The most precise posi-
tion for this source is that of the radio counterpart, RA
= 17h31m57.s73 ± 0.s02, Dec = -33◦50′02.′′5 ± 1.′′1 (J2000),
where the errors are 1σ uncertainties (Mart´ı et al. 1998).
Assuming that the bright PRE bursts reach the He limit,
Galloway et al. (2008) infer a distance of 5.2±0.5 kpc. Burst
oscillations have been detected in multiple bursts at ≈ 363
Hz (Strohmayer et al. 1996; Galloway et al. 2008). An un-
successful search for persistent pulsations using Ginga data
was carried out by Vaughan et al. (1994). The orbital pa-
rameters are not known, although Strohmayer et al. (1998b)
infer vx sin i < 20.7 km/s from the scatter of asymptotic fre-
quencies of burst oscillations from a series of bursts from
1996-7. We will use this constraint in our initial assessment
and use it to estimate Porb. If we assume Roche lobe over-
flow, then we find that a main sequence or evolved star can-
not satisfy the various relations. The companion must be ei-
ther a white dwarf or a helium star. This is consistent with
the properties of the X-ray bursts from this source, which
suggest a hydrogen poor donor (Galloway et al. 2008). We
will therefore assume that Porb < 10 hours.
3.4.11 4U 1702-429 (νb = 329 Hz)
This is a persistent atoll source. Based on the spectral
modelling of Galloway et al. (2008), we set 2 ASM cts/s =
7.8×10−10 ergs cm−2 s−1 (bolometric). The long-term aver-
age ASM countrate is 3.2 cts/s. The best position, measured
by Chandra, is RA = 17h06m15.s314, Dec = -43◦02′08.′′60
(J2000), with an uncertainty of 0.′′6 (Wachter et al. 2005).
Assuming that the PRE bursts reach the limit for pure He,
Galloway et al. (2008) derive a distance of 5.5 ± 0.2 kpc.
Burst oscillations are detected in multiple bursts at ≈ 329 Hz
(Markwardt, Strohmayer & Swank 1999a; Galloway et al.
2008). The orbital parameters are not known but there are
no dips or eclipses.
3.4.12 MXB 1730-335 (Rapid Burster) (νb = 306 Hz)*
The Rapid Burster is a transient globular cluster source with
regular outbursts that occur around every 200 days. It is un-
usual in being the only system to show both Type I and Type
II X-ray bursts, the latter being driven by spasmodic accre-
tion. Using the spectral modelling of Galloway et al. (2008)
we set 10 ASM cts/s = 6.7 × 10−9 ergs cm−2 s−1 (bolo-
metric). The long-term average ASM countrate is 1 ct/s,
and during a typical 25 day outburst the average countrate
is 8.8 cts/s. The most accurate position, given by the ra-
dio counterpart, is RA = 17h33m24.s61, Dec = -33◦23′19.′′8
(J2000), with an error of 0.′′1 (Moore et al. 2000). This is
within the error circle of Chandra observations (Homer et al.
2001). The distance to the host globular cluster, Liller 1, is
8.8+3.3
−2.4 kpc (Kuulkers et al. 2003). Averaging the burst rise
phase of 31 X-ray bursts recorded by RXTE revealed a weak
candidate burst oscillation frequency of ≈ 306 Hz (Fox et al.
2001). The orbital parameters are not known, and no opti-
cal counterpart has been detected because of crowding in
the host globular cluster.
3.4.13 IGR J17191-2821 (νb = 294 Hz)*
This transient X-ray source was discovered only recently,
with one recorded outburst. The average ASM countrate
during the 11 day outburst was 3.7 cts/s, giving a long-term
average countrate during the RXTE era of 0.01 cts/s. Based
on Klein-Wolt et al. (2007a) we assume 3.5 ASM cts/s =
1.2 × 10−9 ergs cm−2 s−1 (2-10 keV), and apply a bolo-
metric correction factor of 2. The most precise position for
the source, from SWIFT, is RA = 259.81306◦ , Dec = -
28.29919◦ (J2000), with an accuracy of 4′′ (Klein-Wolt et al.
2007b). Burst oscillations at ≈ 294 Hz have been detected
in one burst (Markwardt et al. 2007). The peak flux of the
X-ray bursts sets an upper limit to the distance of ∼ 11
kpc (Markwardt et al. 2007). Orbital parameters are not yet
known, and there is no information on the companion.
3.4.14 4U 1916-053 (νb = 270 Hz)*
This is a persistent source in an ultracompact binary
with an H-poor donor star. Following the spectral mod-
elling of Galloway et al. (2008) we set 0.5 ASM cts/s =
2.7 × 10−10 ergs cm−2 s−1 (bolometric). The long-term av-
erage ASM countrate is 1.3 cts/s. The most recent posi-
tion, given by Chandra, is RA = 19h18m47.s871, Dec = -
05◦14′17.′′09 (J2000), with an error 0.′′6 (Iaria et al. 2006)15.
Galloway et al. (2008) derive a distance of 8.9 ± 1.3 kpc or
6.8 ± 1.0 kpc, depending on composition, from PRE bursts.
Burst oscillations at ≈ 270 Hz have been detected in one
X-ray burst (Galloway et al. 2001). The orbital period has
been the subject of much debate in the literature, due in
part to differences between the X-ray and orbital periods
determined by dipping and photometry respectively. The
most recent papers on this topic seem to resolve the issue
(Chou, Grindlay & Bloser 2001; Retter et al. 2002) by de-
termining that the system displays superhumps rather than
being a hierarchical triple. Chou, Grindlay & Bloser (2001)
use X-ray dip times to derive Porb = 3000.6508 ± 0.0009 s
(in agreement with the value of Porb = 3000.6452 ± 0.0043
s reported by Wen et al. (2006) using the RXTE All Sky
Monitor), with Porb < 2.06 × 10−11 (2σ upper limit). The
X-ray dip ephemeris is Tdip = 50123.00944±1.4×10−4 MJD.
Whilst it is not entirely clear how the dip time relates to T90
we will assume in our initial analysis that this could be de-
termined, and use the uncertainty in Tdip as an estimate of
the uncertainty in T90. The projected semi-major axis has
not been measured, but we can impose some constraints.
For the known orbital period, the assumption of Roche lobe
overflow (Equation 7) gives the mean mass of the donor
as 158 g/cm3, implying a dwarf companion. X-ray burst
properties imply a helium rich donor. Optical spectroscopy
shows large amounts of N, suggesting that the companion
is a helium white dwarf rather than a helium star or an
evolved secondary (Nelemans, Jonker & Steeghs 2006). Us-
ing the models of Deloye & Bildsten (2003) this implies a
companion mass in the range 0.008 - 0.03 M⊙ (depending
on core temperature). For dips we expect an inclination in
the range 60− 75◦. If we assume that the neutron star mass
15 This position has a substantial offset in declination from the
optical counterpart (V1405 Aql) reported by (Gottwald et al.
1991), with RA = 19h18m47.s91, Dec = -05◦14′08.′′7 (J2000).
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
13
is in the range 1.2−2.4M⊙, Equation (8) implies that ax sin i
must lie in the range 4 -- 25 lt-ms.
3.4.15 XB 1254-690 (νb = 95 Hz)*
This source is persistent, at a steady level, with dips.
Galloway et al. (2008) carry out spectral modelling and find
a persistent flux level of 9 × 10−10erg cm−2 s−1 (bolomet-
ric). The Chandra position reported by Iaria et al. (2007)
is RA = 12h57m37.s153, Dec = -69◦17′18.′′98, with a 90%
uncertainty radius of 0.′′616. Radius expansion in the precur-
sor to a superburst leads to a distance estimate of 13 ± 3
kpc (in't Zand et al. 2003). Bhattacharyya (2007) reported
tentative evidence of a burst oscillation at 95 Hz in one
burst from this source. Motch et al. (1987) used optical pho-
tometry to find Porb = 3.9334 ± 0.0002 hours This accords
with the period of 3.88 ± 0.15 hours derived from the X-
ray lightcurve (Courvoisier et al. 1986), although the X-
ray dips are not always present (Smale & Wachter 1999).
Barnes et al. (2007) have recently updated the ephemeris
derived by Motch et al. (1987) and report an X-ray dip time
Tdip = 2453151.647 ± 0.003. Motch et al. (1987) showed
that the X-ray dips in this source occurred at phase 0.84
(with zero at optical minimum), so they therefore derive T0
= JD 2453151.509 ±0.003. Barnes et al. (2007) have used
phase-resolved spectroscopy of the He II λ4686 emission line
(thought to be emitted in the inner accretion disk, close to
the compact object) to estimate the velocity of the compact
object, and find a velocity semi-amplitude vx sin i = 130±16
km/s. Only the lower portion of this range is consistent with
a main sequence or undermassive companion star: however
we will use this value in our initial assessment of detectabil-
ity.
3.4.16 EXO 0748-676 (νb = 45 Hz)
This system, which shows both dips and eclipses, has been
persistently active since 1985. Based on the spectral mod-
elling of Galloway et al. (2008), we set 0.6 ASM cts/s =
3.6×10−10 ergs cm−2 s−1 (bolometric). The long-term aver-
age countrate is 0.76 cts/s. The best position, for the optical
counterpart UY Volantis, is RA = 7h48m25.s0 ± 0.s1, Dec =
-67◦37′31.′′7 ± 0.′′7 (B1950) (Wade et al. 1985). No more up
to date position is available. The detection of PRE bursts
that seem to be He rich implies a distance of 7.4 ± 0.9 kpc
(Galloway et al. 2008). Villarreal & Strohmayer (2004) dis-
covered burst oscillations at 45 Hz by averaging together
spectra from 38 separate bursts detected between 1996 and
early 2003. Timing of the eclipses constrains the orbital pa-
rameters. Attempts to compute an orbital ephemeris, how-
ever, have been complicated. The most recent study by
Wolff et al. (2002) finds a large apparent period change of
8 ms over the period 1985 to 2000 - much larger than ex-
pected from orbital models - and intrinsic jitter that cannot
be explained by any simple ephemeris. The reason for this
variability has yet to be resolved, and it is not clear whether
this represents genuine evolution in the binary period or not.
16 This position is 2′′ away from the older optical counterpart
position reported by Bradt & McClintock (1983).
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Wolff et al. (2002) consider various models in their analy-
sis, with orbital periods Porb = 0.1593378 ± 1 × 10−7 days,
mid-eclipse times T90 = 46111.0739 ± 0.0013 MJD/TDB,
and Porb . 10−11. More recent analysis by Wen et al.
(2006), using ASM data, finds Porb = 0.1593375 ± 6 × 10−7
days. The projected semi-major axis has not been mea-
sured directly, but can be constrained. Wolff et al. (2002)
find eclipse durations 497.5 ± 6 s, which gives a eclipse half-
angle θx = (6.543 ± 0.015)◦. This constrains the inclination
via Equation (9). Roche lobe overflow (Equation 7) implies
a mean density for the donor star of 7.5 g/cm3, suggesting a
main sequence or evolved companion. Maximum donor mass
corresponds to a main sequence donor, with Md = 0.42M⊙.
Donor mass can be lower if the companion is evolved, so
following Hynes et al. (2006) we take a minimum plausible
companion mass of 0.07 M⊙. For the NS we consider masses
in the range 1.2 − 2.4M⊙. This implies that ax sin i lies in
the range 0.11 -- 0.84 lt-s.
3.5 Kilohertz QPO sources
These remain the most difficult sources, because of the un-
certainty in the precise relationship between kHz QPO sep-
aration (which varies) and spin frequency. Sources where a
wide range of accretion rates have been sampled show varia-
tion, those where only a few accretion rates have been sam-
pled (including the pulsars that have kHz QPOs) do not. To
gauge the uncertainty, consider the kHz QPO separations
recorded for those sources where we have either a spin fre-
quency or a burst oscillation frequency, illustrated in Figure
3:
• Aql X-1: 550 Hz intermittent pulsar, twin kHz QPO
separation 278 ± 18 Hz (Barret, Boutelier & Miller 2008).
• SAX J1808.4-3658: 401 Hz pulsar. Twin kHz QPOs
observed once during the 2002 outburst. Wijnands et al.
(2003a) report a separation of 195 ± 6 Hz, later analysis by
van Straaten, van der Klis & Wijnands (2005) using a dif-
ferent fitting technique gives a separation of 182 ± 8 Hz.
• XTE J1807-294: 191 Hz pulsar. Twin kHz QPOs de-
tected on several occasions during the 2003 outburst. The
measured separations were consistent with the spin fre-
quency, although the weighted average separation of 205± 6
Hz exceeds the spin at the 2.3σ level (Linares et al. 2005).
• 4U 1608-522: 619 Hz burst oscillations. Twin kHz
QPOs observed in the 1996 and 1998 outbursts, separations
in the range 225 − 325 Hz (M´endez et al. 1998a,b).
• SAX J1750.8-2980: 601 Hz burst oscillations. Tenta-
tive detection of twin kHz QPOs with a separation of 317±9
Hz (Kaaret et al. 2002).
• 4U 1636-536: 582 Hz burst oscillations. Twin
kHz QPOs detected on multiple occasions, with sep-
arations varying from 240 − 325 Hz (Wijnands et al.
1998;
1997a;
Jonker et al. 2002; di Salvo, M´endez & van der Klis 2003;
Barret, Olive & Miller 2005).
M´endez, van der Klis & van Paradijs
• KS 1731-260: 524 Hz burst oscillations. Twin kHz
QPOs with separation 260.3±9.6 Hz seen in one observation
(Wijnands & van der Klis 1997).
• 4U 0614+09: 415 Hz burst oscillations. Twin kHz
QPOs have been seen on several occasions (Ford et al.
1997; van Straaten et al. 2000, 2002). Separations vary from
14 Watts et al.
700
600
500
400
300
200
100
)
z
H
(
n
o
i
t
a
r
a
p
e
s
O
P
Q
z
H
k
0
0
100
200
300
400
Spin/burst oscillation frequency (Hz)
500
600
700
Figure 3. A comparison of twin kHz QPO separation and spin
frequency or burst oscillation frequency for those sources that
show both phenomena. The dotted line indicates equality of the
two measures. For some objects kHz QPO separation is consistent
with being constant: these are shown as single points with error
bars. Note however that this may be due to poor sampling of
source states. For five objects kHz QPO separation varies: these
are shown as two points with a line indicating the range.
• 4U 1728-34:
238± 7 Hz up to 382± 7 Hz (perhaps even as low as 213± 9
Hz although this figure is tentative).
363 Hz burst oscillations. Twin
kHz QPOs detected on multiple occasions, with sep-
arations
in the range 275 -- 350 Hz (Strohmayer et al.
1996; M´endez & van der Klis 1999; di Salvo et al. 2001;
Migliari et al. 2003).
• 4U 1702-429: 330 Hz burst oscillations. Twin kHz
QPOs detected once in 1997, with a separation of 333 ± 5
Hz (Markwardt, Strohmayer & Swank 1999a).
• IGR J17191-2821: 294 Hz burst oscillations. Twin
kHz QPOs with separation 330 Hz (Klein-Wolt et al. 2007c).
• 4U 1916-053: 270 Hz burst oscillations. Twin kHz
QPOs detected several times in 1996. Separation was con-
sistent with being constant at 348± 12 Hz on four occasions;
on the fifth it was 289 ± 5 Hz (Boirin et al. 2000).
It is far from clear that there is a direct (or indeed any)
relationship between kHz QPO separation and spin, partic-
ularly for the high frequency sources. See Yin et al. (2007)
and M´endez & Belloni (2007) for an extended discussion of
this issue.
We now summarize the properties for the kHz QPO
sources. Note that we exclude from our data set the
peculiar X-ray binary Cir X-1 (1516-559), since it is
not clear that our accretion torque model applies. Al-
though this source has twin kHz QPOs (Boutloukos et al.
2006),
to be a high mass X-ray bi-
nary (Jonker, Nelemans & Bassa 2007) with a highly
eccentric orbit
(Murdin et al. 1980; Oosterbroek et al.
1995) where accretion disk formation is only sporadic
(Johnston, Fender & Wu 1999; Johnston et al. 2001).
thought
it
is
found that the long-term average flux is 1.1 × 10−8 ergs
cm−2 s−1. The best position for this source, from the opti-
cal counterpart, is RA = 21h42m36.s91, Dec = +38◦05′27.′′9,
with an accuracy of 0.′′5 (B1950)(Giacconi et al. 1967;
Bradt & McClintock 1983). No more up to date high preci-
sion position is available. Galloway et al. (2008) use the PRE
bursts to estimate a distance of 10± 2 kpc or 14± 3 kpc de-
pending on composition. However optical observations sug-
gest a distance of only 7.2 ± 1.1 kpc (Orosz & Kuulkers
1999). Wijnands et al. (1998a) detected twin kHz QPOs
in RXTE data from a few hours on July 2 1997, when
the source was on the horizontal branch. Peak separa-
tion was 346 ± 29 Hz. Kuznetsov (2002) re-analysed the
same data and found a separation of 366 ± 18 Hz. Un-
successful searches for persistent pulsations were carried
out using Ginga data from 1987-1989 (Wood et al. 1991;
Vaughan et al. 1994). Kuulkers et al. (1995) set upper lim-
its on the presence of burst oscillations for frequencies be-
low 256 Hz using EXOSAT data from 1983-1985, and Smale
(1998) set upper limits on burst oscillations in the 200-
600 Hz range for one burst observed with RXTE. The bi-
nary orbit can be constrained by optical observations of
the companion star V1341 Cyg. The most recent spectro-
scopic observations, by Casares, Charles & Kuulkers (1998)
give Porb = 9.8444 ± 0.0003 days, T90 = 2449339.50 ± 0.03
HJD, and e = 0.024 ± 0.015 (consistent at the 2σ level with
being zero, which we assume in our initial analysis). Error
bars are 1σ uncertainties. By measuring the projected veloc-
ity of the secondary star and its rotational broadening, then
assuming tidal locking and Roche lobe overflow, they infer
vx sin i = 29.9 ± 3.6 km/s. We use this value in our initial
assessment of detectability.
3.5.2 GX 340+0 (1642-455)
This is a persistent Z source that has not shown X-ray
bursts. Based on spectral modelling by Ford et al. (2000) we
set 25 ASM cts/s = 2.3 × 10−8 ergs cm−2 s−1 (bolometric).
The long-term average countrate is 30 cts/s. The best posi-
tion, given by the radio counterpart, is RA = 16h45m44.s60±
0.s02, Dec = -45◦37′53.′′6±0.′′3 (J2000) (Penninx et al. 1993).
Christian & Swank (1997) use Einstein observations to es-
tablish an upper limit on the distance of 11 ± 3.3 kpc17.
Twin kHz QPOs have been detected in RXTE data from
1997-8, with separations in the range 275±24 Hz to 413±21
Hz (Jonker et al. 1998, 2000). However, for the sample as a
whole, the separation is formally consistent with being con-
stant at 339± 8 Hz. Unsuccessful searches for persistent pul-
sations have been carried out using Ginga data (Wood et al.
1991; Vaughan et al. 1994). The orbital parameters are not
known.
3.5.3 4U 1735-44
This is a bright persistent atoll source that has both
bursts and superbursts. Using the spectral modelling of
Galloway et al. (2008) we set 18 ASM cts/s = 9.1 × 10−9
3.5.1 Cyg X-2 (2142+380)
This is a persistent Z source that has X-ray bursts.
Galloway et al. (2008) carried out spectral modelling and
17 Ford et al. (2000) quote a lower distance for this source, but
the value given does not tally with that in the original reference
that they cite.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
15
ergs cm−2 s−1 (bolometric). The long-term average coun-
trate is 14 cts/s. The best position, for the optical counter-
part V926 Sco, is RA = 17h35m19.s28, Dec = -44◦25′20.′′3,
uncertainty < 1′′(Bradt & McClintock 1983). No more up
to date position is available. The distance inferred from the
radius expansion bursts (assuming He limit) is 8.5 ± 1.3
kpc (Galloway et al. 2008). Twin kHz QPOs were seen in
RXTE observations on May 30-31 1998. Peak separation
varied from 296 ± 12 Hz up to 341 ± 7 Hz, inconsistent
at the 3.1σ level with being constant (Ford et al. 1998).
Jongert & van der Klis (1996) placed upper limits on the
presence of burst oscillations (up to 256 Hz) using EXOSAT
data: no figures have been reported for the RXTE burst sam-
ple. The orbital parameters can be constrained by optical
observations. Casares et al. (2006) report a recent spectro-
scopic ephemeris, with Porb = 0.19383351± 3.2× 10−7 days,
in agreement with the most recent photometric ephemeris
(Augusteijn et al. 1998). The time of inferior conjunction
of the donor star T90 = 2452813.495 ± 0.003 HJD. Further
measurements and source modelling suggest that velocity
semi-amplitude of the donor star is in the range 215-381
km/s, with a mass ratio md/mx in the range 0.05-0.41. This
would suggest vx sin i < 156 km/s and we use this limit in
our initial estimate of detectability.
3.5.4 GX 5-1 (1758-250)
This is a persistent non-bursting Z source, located in the
highly absorbed galactic bulge region. Based on spectral
modelling by Ford et al. (2000), we set 50 ASM cts/s =
3.8 × 10−8 ergs cm−2 s−1 (bolometric). The long-term
average ASM countrate is 70 cts/s. The best position,
is RA = 18h01m08.s233, Dec
for the radio counterpart,
= −25◦04′42.′′044 (J2000), positional uncertainty 0.′′040
(Berendsen et al. 2000). Distance to this source is poorly
constrained, although Christian & Swank (1997) give an
upper limit of 9 ± 2.7 kpc18. Twin kHz QPOs have
been reported in RXTE observations from 1996 - 2000
(Wijnands et al. 1998b; Jonker et al. 2002). Peak separa-
tion is not constant, but varies from 232 ± 13 Hz up to
344 ± 12 Hz. Unsuccessful searches for persistent pulsations
were carried out using Ginga data from 1987 (Wood et al.
1991; Vaughan et al. 1994). The orbital parameters are un-
known.
3.5.5 4U 1820-30
This is a persistent bursting atoll source with regular dip-
ping cycles. Using the spectral modelling of Galloway et al.
(2008) we set 32 ASM cts/s = 2.1 × 10−8 ergs cm−2
s−1 (bolometric). The long-term average ASM countrate
is 20.6 cts/s. The best position, from Hubble Space Tele-
scope observations of the optical counterpart,
is RA =
18h23m40.s453 ± 0.s012, Dec = -30◦21′40.′′08 ± 0.′′15 (J2000,
1σ errors), (Sosin & King 1995)19. The source is located in
18 The distance of 6.4 ± 0.6 kpc derived by Penninx (1989),
cited incorrectly as 9.2 ± 0.7 kpc by Fender & Hendry (2000);
Migliari & Fender (2006) is highly model-dependent.
19 Obtaining a more accurate radio position is complicated by
the presence of a nearby pulsar (Migliari et al. 2004).
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
the globular cluster NGC 6624, and optical observations im-
ply a cluster distance of 7.6 ± 0.4 kpc (Heasley et al. 2000).
The distance estimated from X-ray bursts is ≈ 6.6 kpc
(Vacca, Lewin & van Paradijs 1986; Kuulkers et al. 2003).
Twin kHz QPOs have been detected by RXTE on several
occasions (Smale, Zhang & White 1997; Zhang et al. 1998;
Altamirano et al. 2005). Measured separations vary in the
range 220-350 Hz, although the uncertainties are such that
the separation is consistent with being constant at ≈ 275
Hz. Dib et al. (2005) have placed extremely stringent limits
on the presence of persistent pulsations up to 2000 Hz using
archival RXTE data, for all ax sin i < 16.8 lt-ms (superced-
ing earlier work by Wood et al. (1991) and Vaughan et al.
(1994)). Jongert & van der Klis (1996) placed upper limits
on the presence of burst oscillations up to 256 Hz for EX-
OSAT bursts. The most up to date X-ray ephemeris, us-
ing all data from Ariel 5, Ginga and RXTE, gives Porb =
Porb = (−7.54 ± 3.21) × 10−13 s/s,
685.0119 ± 1.02 × 10−4 s,
and the time of maximum X-ray light is 2442803.63564 ±
2.2 × 10−4 HJD (Chou & Grindlay 2001). Noting that an
ephemeris of similar accuracy can also be derived from UV
data (Anderson et al. 1997), we will assume that the refer-
ence time could if required be related to a known phase in
the orbit. There is also a super-orbital periodicity of ≈ 176
days, most likely due to perturbations of the orbital eccen-
tricity (up to e = 0.004 by a third member of the system
(Chou & Grindlay 2001; Zdziarski, Wen & Gierli´nski 2007),
but we neglect this in our initial assessment. The projected
semi-major axis has not been measured, but can be con-
strained. The assumption of Roche lobe overflow (Equa-
tion 7) gives a mean donor density of 3300 g/cm3, im-
plying a white dwarf companion. The donor must also be
helium-rich to explain the X-ray burst properties. Helium
white dwarf models of Deloye & Bildsten (2003) suggest
a donor mass in the range 0.07 − 0.08M⊙ (depending on
temperature). Anderson et al. (1997) inferred an inclination
in the range 35 − 50◦ from observations of the UV coun-
terpart. Ballantyne & Strohmayer (2004), analysing super-
burst data, inferred a slightly lower value, so we will consider
a minimum inclination of 30◦. Then for neutron star masses
in the range 1.2 − 2.4M⊙ we predict ax sin i in the range 7 --
20 lt-ms. Recent modelling by Zdziarski, Wen & Gierli´nski
(2007) suggests that the super-orbital variability may even-
tually pose even tighter constraints on the system.
3.5.6 Sco X-1 (1617-155)
This the closest accreting neutron star in our study, and
is a persistent Z source. Using the spectral modelling
of Ford et al. (2000) we set 920 ASM cts/s = 4 × 10−7
ergs cm−2 s−1 (bolometric). The long-term average ASM
countrate is 892 cts/s. The most accurate position for
the source, given by VLBA measurements,
is RA =
16h19m55.s0850, Dec = -15◦38′24.′′9, with an uncertainty of
0.′′5 (Bradshaw, Fomalont & Geldzahler 1999). The source
has also has a measurable proper motion, which we neglect
in this initial analysis (Bradshaw, Fomalont & Geldzahler
1999). The distance, measured by parallax, is 2.8 ± 0.3 kpc
(Bradshaw, Fomalont & Geldzahler 1999). Twin kHz QPOs
are observed, with separations in the range 240 -- 310 Hz
1997; M´endez & van der Klis
(van der Klis et al.
2000). There have been
for
unsuccessful
1996,
searches
16 Watts et al.
persistent pulsations up to 256 Hz using EXOSAT
(Middleditch & Priedhorsky 1986) and up to 512 Hz
using Ginga data (Wood et al. 1991; Hertz et al. 1992;
Vaughan et al. 1994). Photometric observations of
the
counterpart V818 Sco imply an orbital pe-
optical
s
(Gottlieb, Wright & Liller
riod of 68023.84 ± 0.08
of RXTE ASM data by
1975),
although analysis
Vanderlinde, Levine & Rappaport
that
the true period could in fact be slightly longer, at
68170 s. Analysis by Steeghs & Casares (2002) indicates
vx sin i = 40 ± 5 km/s. Assuming the orbital period of
Gottlieb, Wright & Liller (1975), Steeghs & Casares (2002)
derive an ephemeris with T90 = 2451358.568± 0.003 HJD20.
suggests
(2003)
3.5.7 GX 17+2 (1813-140)
This is a persistent Z source with X-ray bursts. Using the
spectral modelling of Galloway et al. (2008) we set 40 ASM
cts/s = 1.6 × 10−8 ergs cm−2 s−1 (bolometric). The long-
term average ASM countrate is 45 cts/s. The most accurate
position, from VLA observations of the radio counterpart,
is RA = 18h16m1.s389 ± 0.′′06, Dec = -14◦02′10.′′62 ± 0.′′04
(J2000), 1σ uncertainties in position (Deutsch et al. 1999).
This is within the Chandra error circle for the X-ray po-
sition (Callanan et al. 2002). Analysis of X-ray bursts sug-
gests a distance of either 9.8 ± 0.4 kpc or 12.8 ± 0.6 kpc
(Galloway et al. 2008). However, there are questions over
how to correct for the super-Eddington persistent flux, and
the true distance could be lower (Kuulkers et al. 2002).
Twin kHz QPOs have been observed on multiple occasions
(Wijnands et al. 1997b; Homan et al. 2002). Separation,
which varies from 239± 17 Hz up to 308± 14 Hz, is not con-
stant at the 97% confidence level. Upper limits on the pres-
ence of persistent pulsations in Ginga data were reported by
Wood et al. (1991) and Vaughan et al. (1994). Upper lim-
its on the presence of burst oscillations in EXOSAT and
RXTE data have been reported by Kuulkers et al. (1997)
and Kuulkers et al. (2002). The orbital parameters are not
known.
3.5.8 XTE J2123-058
This is a transient bursting atoll source in the Galactic Halo,
with one recorded outburst in 1998. Using the spectral mod-
elling of Galloway et al. (2008) we set 6.4 ASM cts/s =
2.1 × 10−9 ergs cm−2 s−1 (bolometric). The average ASM
countrate over the 50 day outburst was 2.7 cts/s, yielding
a long-term average countrate over RXTE's lifetime of 0.03
cts/s. The most accurate position, as measured by Chan-
dra, is RA = 21h23m14.s54, Dec = -05◦47′53.′′2 (J2000, un-
certainty 0.′′6) (Tomsick et al. 2004). Tomsick et al. (2001)
infer a distance of 8.5± 2.5 kpc, consistent with the distance
of 9.6 ± 1.3 kpc inferred by Casares et al. (2002). Twin kHz
20 Abbott et al. (2007a) use a slightly larger uncertainty on T90
in their searches for gravitational waves from Sco X-1, to account
for the time elapsed since the Steeghs & Casares (2002) measure-
ment. For consistency with the rest of our analysis we use the
smaller uncertainty, assuming that the measurement could be re-
done today to the same level of accuracy.
QPOs were detected during one observation, with separa-
tions in the range 255 ± 14 to 276 ± 9 Hz (Homan et al.
1999; Tomsick et al. 1999). Casares et al. (2002) use spec-
troscopic and photometric measurements to derive Porb =
21447.6 ± 0.2 s (1σ errors). Tomsick et al. (2002), how-
ever, derive Porb = 21442.3 ± 1.8 (1σ errors). This discrep-
ancy, and results from earlier photometric measurements by
Tomsick et al. (1999); Zurita et al. (2000), have yet to be
resolved. The most recent time of minimum optical light
T90 = 2451779.652 ± 0.001 HJD. The projected velocity
vx sin i is also constrained. Casares et al. (2002) attempt to
measure this directly, and find vx sin i = 140 ± 27 km/s.
Tomsick et al. (2001, 2002) measure the projected orbital
velocity and rotational velocity of the companion. Assum-
ing a Roche lobe filling and tidally locked companion they
infer the mass ratio, and hence a projected orbital velocity
for the neutron star of vx sin i = 110+54
−36 km/s. Shahbaz et al.
(2003) use the projected orbital velocity of the companion
measured by Casares et al. (2002) and use more sophisti-
cated models of the system to give the mass ratio. The re-
sulting projected orbital velocity is vx sin i = 103+46
−7 km/s
(90% confidence). In our initial analysis, however, we use
the direct measurement by Casares et al. (2002).
3.5.9 GX 349+2 (1702-363, Sco X-2)
This persistent Z source does not show X-ray bursts. Us-
ing the modelling of Zhang, Strohmayer & Swank (1998),
we set 43 ASM cts/s = 1.4 × 10−8 ergs cm−2 s−1 (2-
10 keV). There is no detailed spectral modelling avail-
able for this source, so we adopt a bolometric correc-
tion factor of 2. The long-term average ASM countrate
is 50.2 cts/s. The best position, from VLA measurements
of the radio counterpart, is RA = 17h02m22.s93, Dec = -
36◦21′20.′′3 (B1950, accuracy 0.′′5) (Cooke & Ponman 1991).
Iaria et al. (2004) infer a distance of 3.6-4.4 kpc from
BeppoSAX observations. Twin kHz QPOs were detected
by RXTE in January 1998, with a separation of 266 ±
13 Hz (Zhang, Strohmayer & Swank 1998), confirmed by
O'Neill et al. (2002). The source has however rarely been ob-
served in the state where kHz QPOs are prevalent. A search
for persistent pulsations in Ginga data from 1989 was unsuc-
cessful (Vaughan et al. 1994). The binary period, measured
using optical photometry and spectroscopy, has been the
subject of some debate (Southwell, Casares & Charles 1996;
Wachter & Margon 1996; Barziv et al. 1997), but is now es-
tablished as Porb = 22.5 ± 0.1 hours (1σ error) (Wachter
1997). The other binary parameters have not been mea-
sured, although we can constrain ax sin i. The assumption
of Roche lobe overflow (Equation 7) gives a mean donor
density 0.2 g/cm3, which requires a donor that has evolved
off the main sequence. In the absence of better constraints
we will assume a mass ratio md/mx < 0.8. For the range of
neutron star masses considered (1.2 − 2.4M⊙) this implies
ax sin i < 7 lt-s.
4 SEARCHING WITH PARAMETER
UNCERTAINTIES
Let us summarize briefly the gravitational wave emission
from a neutron star in a binary system. We assume that
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
17
the centre of mass of the binary is not accelerating in the
solar system barycentre (SSB) frame. The timing model for
the arrival times of the wave-fronts of the GW is taken to
be the usual one (Taylor & Weisberg 1989). Let T be the
arrival time of the wave at the SSB, τ the proper time of
emission in the rest frame of the neutron star, and t the
time in the rest frame of the gravitational wave detector.
The quantities T and τ are related by
T − T0 = τ + ∆R + ∆E + ∆S
(10)
where ∆R is the Roemer time delay accounting for the light
travel time across the binary, ∆E and ∆S are respectively
the orbital Einstein and Shapiro time delays in the binary,
and T0 is a reference time. There are no additional timing
delays due to dispersion. It turns out that, for our purposes,
the Roemer delay is the most significant contribution. If r
is the vector joining the centre of mass of the binary system
with the neutron star, and n is the unit vector pointing from
the SSB to the source, then
∆R = −
r · n
c
.
(11)
There is then a similar relation between T and the arrival
time t at the earth based detector, and we assume that this
can be corrected for since the sky-position is known.
In the models that we are considering, the intrinsic grav-
itational wave frequency ν depends on the spin frequency νs.
The phase of the GW at the SSB is
where
φ(t) = Φ0 + Φ(t)
Φ(T ) = 2πν„T −
n · r(T )
c
« .
(12)
(13)
Inclusion of frequency derivatives in this phase model is
straightforward, and we do not write it down explicitly. Since
the gravitational wave amplitudes are expected to be very
weak and the output of the GW detectors dominated by
noise, knowledge of the waveform, especially its phase, is
crucial for detection.
The phase Φ(t) depends on the orbital parameters in-
Porb, Tasc, ax sin i, and e. In
troduced in Section 3.2: Porb,
addition, there are 2 parameters specifying the orientation
of the orbital plane: the inclination angle ι and the argument
of periapsis ω. Of these 7 parameters, only 6 are required
to define the phase model because of the projection along
the line of sight n; see Dhurandhar & Vecchio (2001) for
further details. Taking the spin frequency νs and its time
derivative νs into account, we therefore have a total of 8
parameters which determine the frequency evolution of the
signal: (νs, νs, ax sin i, e, Porb,
Porb, Tasc, ω).
This is clearly a very large parameter space, and a
search over all these parameters using a sufficiently large
data volume will be a big data analysis challenge. Let
us therefore make some simplifying assumptions:
νs = 0,
Porb = 0, and e = 0; i.e. we assume a neutron star spin per-
fectly balanced between accretion and gravitational radia-
tion, and a circular orbit which does not decay appreciably
over the course of the observation time. These assumptions
may not hold for the sources and for the large observation
times that we are considering, and an actual search might
very well have to take some or all of these effects into ac-
count. However, for assessing the detection prospects as we
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
want to do here, this simplification is useful, since adding
the extra parameters will further increase the number of
templates. For some of the more promising sources at or
near the detection threshold, these assumptions will need to
be revisited in greater detail. Some of these extra parame-
ters may need to included, and the resulting search might
again become computationally difficult; this will be studied
in greater detail in future work.
In the case when the orbit is circular (e = 0), which
we shall assume in the rest of this paper, the argument of
periapsis and the initial orbital phase combine additively
into a single parameter so that we are left with only 4 search
parameters: λ = (νs, ax sin i, Porb, Tasc); we shall denote the
components of λ by λi with i = 0 . . . 3.
4.1 Template counting
To determine the computational cost involved in searching
the parameter space described above, we need to calculate
the number of templates required. A calculation of the re-
quired number of templates to search a portion of the pa-
rameter space is based on demanding a certain maximum
mismatch between the templates at neighbouring points in
parameter space. This also guarantees that the true signal
will not have more that the given mismatch to at least one
of the search templates. The mismatch between waveform
templates is measured simply as the fractional loss in the
signal-to-noise ratio (SNR) when one waveform is filtered
(folded) by the other. This fractional loss can be regarded as
a distance measure between points in parameter space, and
this leads naturally to the definition of a parameter-space
metric gij (Sathyaprakash & Dhurandhar 1991, 1994; Owen
1996; Prix 2007b). Using the metric, we write the proper dis-
tance squared (the "mismatch") between two infinitesimally
separated parameter space points as
m = gij dλidλj
(14)
The size of the parameter space is then given by the volume
measure determined by the metric in the usual way. Then,
assuming that we cover this parameter space by a lattice
grid, the number of templates is the total parameter space
volume divided by the volume of each unit cell which makes
up the lattice.
The optimal choice of the lattice is determined by a so-
lution to the so called sphere-covering problem (Prix 2007).
For our purposes, we shall use a simple cubic grid, and there
are two reasons why it is acceptable to use this approxi-
mation. First, the dimensionality of the reduced parameter
space that we are looking at is low enough that the im-
provement in the template placement efficiency is not more
than a factor of about 2 or 3 (Prix 2007). Furthermore, this
improvement in the efficiency does not actually lead to a
corresponding factor of 2-3 improvement in the sensitivity;
the gain in the coherent integration time afforded by this im-
provement is much smaller because the computational cost
typically scales as a large power of the coherent integration
time. Finally, the size of each unit cell is chosen based on
the fractional loss in SNR, i.e. the mismatch m, that we are
willing to tolerate; we shall use a reference value of m = 30%
in this paper.
The first detailed study of the parameter space met-
ric for a neutron star in a binary orbit was carried out
18 Watts et al.
in Dhurandhar & Vecchio (2001) for a coherent matched
filter search. The search for gravitational radiation from
Sco X-1 reported in Abbott et al. (2007a) was the first
and so far, only application of this study. The aim of this
section is mainly to collect some template counting equa-
tions for later use. These equations can all be derived in
a more or less straightforward manner from the results of
Dhurandhar & Vecchio (2001). The main difference is that
Dhurandhar & Vecchio (2001) use notation and variables
targeted towards gravitational wave data analysis, while
here we choose to use notation more familiar to an astron-
omy/astrophysics audience.
max − λi
The first issue is the number of parameters which must
be searched. Let us denote by ∆λi = λi
min the uncer-
tainty in λi from astronomical observations; we assume the
region to be rectangular. Since the proper length of the line
√giidλi, a useful upper bound
on this proper length is to use the maximum value of gii, i.e.
)1/2 ∆λi. This proper distance can then be compared
(gmax
with our reference mismatch m, and we get a measure of the
number of templates required in the λi direction:
in the λi direction is R λi
max
min
λi
ii
Nλi = ∆λir gmax
ii
m
.
(15)
If Nλi < 1, it indicates the λi direction can be covered by
just a single template, and the effective dimensionality of our
parameter space is reduced by 1. The number of templates
for the frequency are:
Nν = ∆ν
πTobs√3m
.
(16)
The uncertainty ∆ν relates directly to the uncertainty in
spin ∆νs. We shall ignore the correlations of ν with the other
parameters; this approximation will suffice for our purposes.
For the other directions, the expressions for Nλi have simple
expressions in two regimes: Tobs ≪ Porb and Tobs ≫ Porb.
have:
In the limit of large observation times, Tobs ≫ Porb, we
Nap = ∆ap
NTasc = ∆Tasc
NPorb = ∆Porb
√2πνmax√m
,
√8π2amax
p
orb √m
P min
√8π2amax
νmax
,
p
νmaxTobs
√3m(P min
orb )2
.
In the other limiting case Tobs ≪ Porb, we get
Nap = ∆ap
,
2π3νmaxT 2
obs
√45m(P min
orb )2
8π5apνmaxT 3
obs
√175m(P min
orb )4
4π3apνmaxT 2
obs
√45m(P min
orb )3
,
.
NTasc = ∆Tasc
NPorb = ∆Porb
(17)
(18)
(19)
(20)
(21)
(22)
For a search over all three parameters at once, the total
number of templates in (ap, Porb, Tasc) space is, for Tobs ≫
Porb:
Nap PorbTasc =
π5ν 3
maxTobs
√8m3
∆[a3
p]∆[P −2
orb]∆[Tasc] ,
(23)
while for Tobs ≪ Porb it is:
√48π13ν 3
maxT 9
obs
√7027611500m3
NapPorbTasc =
∆[a3
p]∆[P −10
orb ]∆[Tasc] .
(24)
Note that Nap NPorb NTasc 6= NapPorbTasc . This happens be-
cause the correlations between the different parameters (i.e.
the off-diagonal terms in the metric) can be very important,
especially for short observation times.
Similarly, we shall require the equations in the case
when one of the coordinates can be ignored and the search
can be performed in a 2-dimensional subspace. The equa-
tions for Tobs ≫ Porb are:
NapTasc =
p]∆[Tasc] ,
∆[a2
(25)
π3ν 2
max
Porbm
π4(amax
p
)2ν 2
m√3
maxTobs
m√3
π3ν 2
maxTobs
∆[P −2
orb]∆[Tasc] ,
∆[a2
p]∆[P −1
orb] .
NPorbTasc =
NapPorb =
Finally, for Tobs ≪ Porb we get:
NapTasc =
∆[a2
p]∆[Tasc] ,
orb
obs(amax
p
2π8ν 2
maxT 5
obs
45√35mP 6
maxT 5
4π8ν 2
135√35m
maxT 6
8π8ν 2
obs
2835√5m
)2
∆[P −6
orb]∆[Tasc] ,
(29)
∆[P −6
orb]∆[a2
p] .
(30)
NPorbTasc =
NapPorb =
(26)
(27)
(28)
We take the total number of templates to be Nν (if it
exceeds unity) multiplied by the number of templates in
(ap, Porb, Tasc) space.
While these equations might not seem very illuminat-
ing, two important features are worth remembering. First,
and probably most importantly, the scaling of the number
of templates with Tobs is very different in the two regimes
Tobs ≪ Porb and Tobs ≫ Porb. For example, in Equations
(23) and (24), the scaling is O(Tobs) when Tobs ≫ Porb,
while it is O(T 9
obs) for small Tobs. This will have important
consequences for GW data analysis, as we shall see later.
Second, computational cost issues become more important
at higher frequencies and for tighter orbits, because the num-
ber of templates typically increases when νmax increases or
Porb decreases.
Finally, we say a few words about the positional ac-
curacy required for the GW searches. We do not wish to
consider searches over sky-position, and thus it is impor-
tant to know the position sufficiently accurately before-
hand. The periodic wave searches get their sky-position in-
formation from the Doppler pattern of the frequency evo-
lution. The sky-position accuracy ∆θ depends strongly on
the coherent observation time Tcoh. For short observation
times (≪ 1 yr) ∆θ increases as roughly O(T 2
obs)
(Brady et al. 1998; Prix 2007b). This increase in the sky-
resolution eventually saturates when Tcoh becomes compa-
rable to a year. The ultimate limit on ∆θ is the diffrac-
tion limit, with Earth's orbit being the aperture size. Thus,
the smallest error box for position will be ∆θ ∼ λgw/1 AU
where λgw is the wavelength of the GW. This corresponds
to about 8′′ at 50 Hz and it is inversely proportional to fre-
quency; this requirement is easily met for all the sources
we are considering. It might in fact be possible to use a sky-
obs) or O(T 3
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
19
position mismatch as a veto to rule out potential candidates.
For a candidate with given values of the frequency and or-
bital parameters, we could calculate the detection statistic
at mismatched sky-positions and verify that the SNR does
decrease as expected. The work of (Prix & Itoh 2005) (see
also Appendix A of Krishnan et al. (2004)) which studies
correlations in frequency and sky-position mismatch might
be useful for this purpose.
5 FUTURE DETECTABILITY
5.1 Best case detectability for various emission
models
We will start by looking at the best case scenarios for de-
tection, and ask what would happen if we knew all of the
parameters to sufficient accuracy that we only had to search
one template for each source, using the flux information col-
lected in Section 3 and summarized in Table 1. In this case
computational cost is not an issue, and we can integrate
for long periods. We will assume that the system is in per-
fect spin balance (so that we can neglect spin derivatives),
and that gravitational wave torques are the only negative
torques operating in the system. This means that we ne-
glect any possible spin-down effects due to the interaction
of the neutron star magnetosphere with the accretion disk,
or any magnetic dipole spin-down.
In Section 2 we gave an overview of the spin balance
model and calculated the best case detectability assuming
gravitational wave emission due to a 'mountain', balancing
the long-term average flux. The results were shown in Figure
2. We also however need to consider whether the quadrupole
Q required for spin balance is feasible.
Q = 7.4 × 1035g cm2„ R3
M1.4«1/4
10
dkpcF 1/2
−8 „ 1 kHz
νs «5/2
(31)
where dkpc = d/(1 kpc). Figure 4 shows the required
quadrupole (scaled by 1045 g cm2, the approximate mo-
ment of
inertia of a neutron star), assuming the long-
term average flux. We also show the maximum sus-
tainable quadrupole for an accreted crust computed by
Haskell, Jones & Andersson (2006) (note that these authors
compute Q22 = (15/8π)1/2Q). The values required for
the slower spinning bursters are at the upper boundary
of what is thought to be feasible, and if the spin of the
kHz QPO sources is slower than the measured separations,
the required quadrupole may exceed this value. However,
magnetic confinement could support larger quadrupoles
(Melatos & Payne 2005; Payne & Melatos 2006).
It is clear from Figure 2 that if the kHz QPO sources
do have spins in the range inferred from the kHz QPO sep-
aration (or higher) then Sco X-1 is the only source that is
in principle marginally detectable by Enhanced LIGO. Pre-
dicted amplitude would of course rise if the spin frequency
were substantially lower, and if this were the case both Sco
X-1 and GX 5-1 could be within detectable range for En-
hanced LIGO (although the inferred quadrupole would be
large, approaching the maximum thought possible).
For Advanced LIGO, several of the kHz QPO sources
are in principle detectable for a single template search in
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
a
i
t
r
e
n
i
f
o
t
n
e
m
o
M
/
e
l
o
p
u
r
d
a
u
Q
1e-05
1e-06
1e-07
1e-08
1e-09
1e-10
Pulsars
Bursters
kHz QPO
1000
Gravitational wave frequency (Hz)
500
Figure 4. The quadrupole Q required for spin balance in the
mountain scenario given the long-term average flux. Q is scaled
by 1045 g cm2 (the approximate moment of inertia of a neutron
star). The frequency at which the kHz QPO symbols appear is the
centre of the measured range of separations. The solid line illus-
trates how the requisite Q would vary for Sco X-1 if this is not the
spin frequency. The dashed line shows the maximum feasible non-
magnetic quadrupole calculated by Haskell, Jones & Andersson
(2006). The uncertainty in the inferred quadrupole is not shown,
but can be substantial since it depends on the distance to the star
(see Table 1).
the broad band configuration. Two of the burst oscillation
sources (XB 1254-690 and 4U 1728-34) are also marginally
detectable within the given narrow band envelope. Several
other burst oscillation sources might be detectable if it were
possible to push the narrow band envelope further down
towards the thermal noise floor. If the proposed Einstein
Telescope reaches its design specification, several more of the
burst and kHz QPO sources might be reachable, although
the pulsars remain undetectable.
One of the major uncertainties in our modelling is
the response to variations in accretion rate. In order to
gauge this effect we therefore consider an alternative model
for the transients, one in which the gravitational wave
torque balances the accretion torque in outburst. This as-
sumes that the gravitational wave emission mechanism re-
sponds very rapidly to the accretion. Although response
timescales are not well studied, this scenario is not unrea-
sonable - an accretion-induced mountain or unstable oscil-
lation, for example, may well grow during outburst and de-
cay during quiescence. Figure 5 compares the best case de-
tectability for the transients if we balance the outburst flux
rather than the long-term average flux (Table 1). The in-
tegration time in outburst is taken to be either the typ-
ical outburst duration, or the maximum integration time,
whichever is larger. The required quadrupoles are all be-
low the Haskell, Jones & Andersson (2006) limit. Many of
the sources show an improvement: the reduction in Tobs
is more than compensated for by the increase in flux. The
burst oscillation sources X 1658-298 and KS 1731-260, for
example, which are undetectable if we consider time-average
flux,
lie within this scenario on or just above the Ad-
vanced LIGO Narrow Band envelope. The intermittent pul-
sar HETE J1900.1-2455 also becomes a more promising
source for the Eintein Telescope. Further theoretical con-
20 Watts et al.
)
e
d
u
t
i
l
p
m
a
e
l
b
a
t
c
e
t
e
d
/
e
d
u
t
i
l
p
m
a
d
e
t
c
i
d
e
r
P
(
0
1
g
o
L
0
-1
-2
0
-1
-2
0
-1
-2
0
Long-term average
Outburst only
Pulsars
Bursters
kHz QPO
A-LIGO BB
A-LIGO NB
ET
500
1000
0
500
1000
Gravitational wave frequency (Hz)
Figure 5. Best case detectability (single template) for the transients, in the mountain scenario. The left-hand panels show the detectability
if gravitational wave emission balances the long-term average accretion rate, for Tobs = 2 years. The right-hand panels show the
detectability if we if we consider the quadrupole necessary to balance the accretion torque during outburst. Tobs for each source is now
either the outburst duration or the maximum integration time (2 years), whichever is longer. In order to compare sources with different
integration times, we show log10 of the ratio of the predicted to the detectable amplitude. The three rows are for different detector
configurations: Top - Advanced LIGO broad band; Middle - Advanced LIGO Narrow Band Envelope; Bottom - Einstein Telescope.
sideration should clearly be given to the issue of torque re-
sponse timescales.
We also consider the situation that would result if
the accretion torque is balanced by spin-down due to
an internal r-mode rather than a mountain. We first
make the assumption that the star has a spin rate and
temperature such that it can sustain an unstable r-mode
(so that it lies in the 'r-mode instability window', see
Andersson & Kokkotas 2001). This assumption may not
be warranted, particularly for some of the more slowly
rotating stars. We also assume that the r-mode amplitude
is steady: r-mode unstable stars may well experience duty
cycles with short-lived periods of strong spin-down (Levin
1999; Andersson et al. 2000; Heyl 2002; Kinney & Mendell
2003), but there are some scenarios in which r-mode emis-
(Andersson, Jones & Kokkotas
sion may be persistent
2002; Wagoner
2006;
Bondarescu, Teukolsky & Wasserman 2007). The grav-
itational wave
torque associated with the dominant
l = m = 2 r-mode (l and m being the standard angular
quantum numbers) is
Nayyar & Owen
2002;
Ngw = −3 J Ωsα2M R2/τg
(32)
(Andersson & Kokkotas 2001). The frequency of the emit-
ted gravitational radiation is not a harmonic of the spin
frequency: for the l = m = 2 r-mode it is 4/3 times the spin
frequency21. The quantity α is the mode amplitude. For an
n = 1 polytrope model of the neutron star, J = 1.635×10−2
and the radiation reaction timescale τg is given by
τg = 47M −1
1.4 R−4
νs «6
10 „ 1 kHz
s
(33)
(Andersson & Kokkotas 2001). The associated luminosity is
Egw = Ngwωm/m, where ωm = ωgw is the angular frequency
21 We assume that this calculation of the relationship between
mode frequency and spin is exact. In practice there is some the-
oretical uncertainty, which would further increase the number of
templates to be searched.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
21
1e-02
1e-04
1e-06
e
d
u
t
i
l
p
m
a
e
d
o
m
R
-
1e-08
Pulsars
Bursters
kHz QPO
500
Gravitational wave frequency (Hz)
1000
Figure 7. The r-mode amplitude α required for spin balance
in the r-mode scenario, using the long-term average flux for
all sources (persistent and transient). The dashed line shows
the maximum (saturation) amplitude calculated by Arras et al.
(2003). The uncertainty in the inferred amplitude is not shown,
but can be substantial since it depends on the distance to the
star. The solid line indicates the r-mode amplitude that would
be required for Sco X-1 if the spin frequency differs from that
inferred from kHz QPO separation.
of the mode. Assuming spin balance, Na = Ngw, the result-
ing gravitational wave amplitude is
h0 = 3.7 × 10−27„ R3
M1.4«1/4
10
F 1/2
−8 „ 1 kHz
νs «1/2
(34)
In Figure 6 we compare the predicted and detectable ampli-
tudes for the r-mode scenario for Tobs = 2 years.
We now consider whether the required r-mode ampli-
tude is feasible. Assuming spin balance, the mode amplitude
is
α = 6.9 × 10−8`M 5
1.4R9
10´−1/4
dkpcF 1/2
−8 „ 1 kHz
νs «7/2
(35)
Figure 7 shows the amplitude required for spin balance for
each source for the long-term average flux. The required am-
plitudes should be compared to the maximum saturation
amplitude αs computed by Arras et al. (2003):
αs ≈ 8 × 10−3" νs
1 kHz"5/2
(36)
The values required for the two slowest spinning
bursters exceed the amplitude limit, as would the kHz QPO
sources if the spin is much slower than the measured separa-
tions. However for all other sources the required amplitude
is below the Arras et al. (2003) limit.
The gravitational wave frequency is lower within the r-
mode scenario than in the mountain scenario. Spin balance
therefore requires a higher GW amplitude, making the r-
mode emission model rather more optimistic. For Advanced
LIGO many of the kHz QPO sources lie above the broad
band noise curve. Two of the burst oscillation sources with
reasonable values of α, 4U 1636-536 and 4U 1728-34, sit well
above the Advanced LIGO narrow band envelope, while 4U
1702-429 and KS 1731-260 lie on the curve (XB 1254-690,
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
which would also be marginal for the narrow band config-
uration, has an unfeasibly high α ). For the Einstein Tele-
scope, however, these sources are well above the noise curve,
and the intermittent pulsar Aql X-1 is also marginally de-
tectable.
If we consider the transient scenario (balancing out-
burst torque rather than time-average torque) the r-mode
model also makes rather more optimistic predictions than
the mountain model. Three burst oscillation sources that are
marginally detectable at best when balancing long-term flux
now come within range of Advanced LIGO's Narrow Band
configuration: XB 1658-298, KS 1731-260 and 4U 1608-522.
By the time we reach the sensitivity of the Einstein Tele-
scope the intermittent pulsar HETE J1900.1-2455 may also
be detectable in outburst.
5.2 Data analysis techniques
space
(either
two kinds of
searches over our param-
We consider
eter
can also be
used as parts of an optimized multi-stage hierarchi-
cal scheme (Brady & Creighton 2000; Cutler et al. 2005;
Meinshausen, Bickel & Rice 2007)):
these methods
of
(i) A full coherent matched-filter search using all avail-
able data, possibly from multiple detectors. This involves
a demodulation of the data for a given parameter space
point using the so called F-statistic; details can be found in
(Jaranowski, Kr´olak & Schutz 1998; Cutler & Schutz 2005).
Such a coherent demodulation statistic can be augmented
by the technique of combining sidebands as described in
Messenger & Woan (2007) (adapted from similar techniques
used in radio pulsar searches, Ransom et al. (2003)).
(ii) A power folding method where the available data is
broken up into smaller segments (often called "stacks" in
the GW literature); the duration of each stack is Tcoh and
Nstacks denotes the number of stacks. If there were no gaps
in the data, we would have Tobs = NstacksTcoh, but duty
factors of ∼ 80% are more likely. Each segment is coher-
ently demodulated using (i) and excess power from each
segment is combined without maintaining phase coherence.
Examples of such power folding methods are: a) stack-slide
(Brady & Creighton 2000) where the relevant statistic is
simply a weighted sum of F-statistic values (at the appro-
priate frequency bins)
ρ =
Nstacks
Xk=1
wkF(k) ,
(37)
or b) the Hough transform method (Krishnan et al. 2004)
where one adds weighted binary number counts
ρ =
Nstacks
Xk=1
wkn(k)
(38)
where n(k) is either 0 or 1 depending on whether the F
exceeds a certain threshold. In each case, the weights wk are
chosen to optimize the sensitivity.
For both (i) and (ii), the computational cost is propor-
tional to the total number of templates Ntemp and to the
amount of data available. Thus, for (i), it is approximately
22 Watts et al.
1e-25
0
h
e
d
u
t
i
l
p
m
A
1e-26
1e-27
1e-28
I - L I G O
E T
E-LIGO
Pulsars
Bursters
kHz QPO
A-LIGO
A-LIGO NB
A-LIGO NB TH
500
Gravitational wave frequency (Hz)
1000
Figure 6. Best case detectability (single template search) using the long-term average flux, for the r-mode scenario. Symbols and lines
as for Figure 2.
ATobsNcoh for some constant A, while for (ii) it is propor-
tional to NstacksTcohNtemp. The precise value of the sensi-
tivity of these searches clearly depends on the details of the
analysis method and software, and the quality of data, such
as the duty cycle of the detector (which might reduce the
amount of data actually available), the presence of noise
artifacts such as spectral disturbances, uncertainties in the
calibration of the detector, and so on. It is thus not possible
to estimate the sensitivity to better than, say, ∼ 5 − 10%
without actually carrying out the search, and it is in fact
even pointless to try and do so in this paper. It is of course
possible to get semi-realistic estimates and this is what we
shall do, but these uncertainties should always be kept in
mind.
Starting with the statistical factor, note that if the
threshold corresponding to a single trial false alarm rate is α,
the probability that the threshold is crossed at least once in
Ntrials independent trials is F A = 1−(1−α)Ntrials ≈ αNtrials
when αNtrials ≪ 1 (Jaranowski, Kr´olak & Schutz 1998). So
we choose α = F A/Ntrials to ensure that the total false
alarm probability is F A. We equate the number of trials
with the total number of templates that must be searched.
Strictly speaking, this is not true because the different tem-
plates are not completely independent. This is usually not
a significant effect as long as the mismatch m used to con-
struct the template bank is not too small; we shall use a
reference value of m = 0.3. To make matters more compli-
cated, while there are reliable calculations for the template
counting for (i) as presented in Sec. 4.1, there are as yet no
reliable estimates for (ii). We shall therefore consider only
the number of templates in each coherent segment. Fortu-
nately, the dependence of the statistical factor is not very
steep (in fact slower than logarithmic) with the number of
trials, so this does not make a significant difference to our
results.
Some details of the statistical calculation are in order.
The choice of α determines the thresholds for both the co-
herent F-statistic search and the semi-coherent search. The
exact relation is however different. The F-statistic follows a
χ2-distribution with 4 degrees of freedom (it is actually 2F
which is χ2 distributed). For the semi-coherent statistic ρ is,
to a reasonable approximation, simply Gaussian (assuming
Nstacks to be sufficiently large); it is actually χ2 with 4Nstacks
degrees of freedom. The assumption of gaussianity may be
questionable here because we are after all dealing with the
tail of the distribution, and the central limit theorem may
not be reliable especially when Nstacks is not particularly
large. A further complication arises when ρ is not simply a
sum of the F-statistic values but is perhaps a Hough statistic
(Krishnan et al. 2004) when the distribution of ρ is closer to
a binomial (which can also be approximated by a Gaussian).
We shall nevertheless assume gaussianity for our purposes
and leave a more detailed study for future work. The rela-
tions between α and the corresponding threshold Fth and
ρth are then different in the two cases:
α = (1 + Fth)e−Fth ,
α =
1
2
erfc„ ρth√2σ«
(39)
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
23
40
30
20
10
r
o
t
c
a
f
l
a
c
i
t
s
i
t
a
t
S
0
0
Gaussian
F statistic
10
20
Log10 (Number of trials)
30
40
The next question we need to address is the computa-
tional cost and whether it is necessary or worthwhile to do
the semi-coherent search (ii). Clearly, for a given amount of
data while we ideally want to make the coherent integration
time Tcoh as large as possible, and the number of segments
Nstacks correspondingly small, the choice of these parameters
is dictated by the computational cost. This computational
burden is in fact, by far, what limits the search sensitivity
(As we just saw, Fstat has a very weak dependence on the
number of templates; we need to change the number of tem-
plates by orders of magnitude before it has an appreciable
effect). The bigger effect of reducing the number of tem-
plates by better astrophysical modelling or observations is
that it allows us to have a larger coherent integration time
Tcoh.
Figure 8. Dependence on the number of independent trials (tem-
plates) of the statistical factor Fstat that appears in the de-
tectability equation (40). Note that Fstat = 11.4 when Ntrials = 1
which reproduces equation (6).
where erfc is the complementary error function, and σ is
the standard deviation of ρ. In the presence of a signal with
amplitude h0, the mean values of F and ρ are increased by
an amount proportional to the SNR2, and thus to h2
0. Tak-
ing a fixed value of the false dismissal rate β determines the
minimum value of h0 needed to exceed the thresholds ρth
and Fth; we shall always take β = 0.1 as the reference value.
Folding in a uniform averaging over all possible pulsar orien-
tations along with the statistical factor in a single parameter
Fstat, we get the following expression for the sensitivity
hsens
0 ≈
Fstat
N 1/4
stacksr Sh
DTcoh
.
(40)
where Fstat varies with the number of trials required,
as shown in Figure 8. This factor was also discussed in
(Abbott et al. 2007a) and (Abbott et al. 2005a) which are
examples of (i) and (ii) respectively. In each case, Fstat in-
creases slower than logarithmically with Ntrials, and the sta-
tistical factor is worse for the χ2 distribution as expected be-
cause it is not as sharply peaked as the Gaussian. Finally, it
is perhaps worth mentioning that averaging over the pulsar
orientation cos ι may not always be appropriate, especially
if there happen to be independent estimates of the neutron
star orientation. If the value of cos ι happens to be anywhere
in the top 1 − p percentile of a uniform distribution (i.e. if
we average over all cos ι > p), then
Fstat
.
(41)
Fstat →
p1 + 11(p + p2)/16 + (p3 + p4)/16
So for example, if p = 0.9, we get Fstat → Fstat/1.5. This
could be important for sources which happen to be near the
detection threshold.
So far we have incorporated the statistical and geo-
metrical factors in Equation (40). For template bank based
searches, the sensitivity is further degraded because of the
discreteness of the template grid. For a template bank cre-
ated with a maximum mismatch of m, assuming a cubic
grid, the average degradation in the SNR is (1− m/3). This
corresponds to a degradation of p1 − m/3 in h0 which is
≈ 0.95 for m = 0.3.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
For a given source, we would first like to know if a
semi-coherent search would be useful. The obvious answer
is: whenever computational cost is an issue, but a more
quantitative answer is also easy to get. Let us compare the
sensitivities of (i) and (ii) for a given computational cost.
Let us assume that the cost for the coherent analysis scales
as a power of Tobs: Ccoh = AT k
obs for some constants A
and k. So if we assume a fixed value of the computational
cost C0, then Tobs = (C0/A)1/k. Thus, the sensitivity is
0 ∝ T −1/2
obs = (C0/A)−1/2k. For a semi-coherent search
hcoh
on the other hand, we have C0 ≈ ANstacksT k
coh (this is
true if the cost of combining the different stacks is negli-
gible). In this case we get Tcoh = (C0/ANstacks)1/k. Using
hsemi−coh
0
coh we get
∝ N −1/4T −1/2
hsemi−coh
0
hcoh
0
= N (2−k)/4k
stacks
.
(42)
> hcoh
This tells us that a semi-coherent search is not effective (i.e.
hsemi−coh
) for k 6 2, and it gets more and more
0
effective for larger k (this conclusion is robust: it is not af-
fected by the approximation of neglecting the cost of the
semi-coherent step).
0
In our present case, recall from equations (17) -- (28) that
we have very different scalings in the regimes Tobs ≫ Porb
and Tobs ≪ Porb. So, for binary systems for which the pa-
rameters have been sufficiently constrained astronomically
and we can afford large integration times, we shall assume
that we are better off doing a simple coherent search. For
others, a semi-coherent strategy has a much larger impact.22
For each potential source, our strategy is to first estimate the
number of templates that are required for a coherent search.
With reasonable estimates of available computational re-
sources, and assuming Tobs = 2 years, this determines the
maximum coherent integration time Tcoh that can be anal-
ysed. If Tcoh is not much smaller than Porb, then by the
previous argument we assume that the gain in using a semi-
coherent search is not significant, and we restrict ourselves
to a pure coherent search. On the other hand, if this Tcoh
does turn out to be much less than Porb, then we consider a
semi-coherent search and estimate Tcoh and Nstacks (assum-
ing here the cost of the semi-coherent step to be comparable
22 Semi-coherent searches might start becoming more important
as soon as we need to start including other parameters, especially
frequency derivatives and Porb.
24 Watts et al.
4
.
1
1
/
r
o
t
c
a
f
l
a
c
i
t
s
i
t
a
t
S
4
3
2
1
Pulsars
Bursters
kHz QPO
1e-28
1e-27
Model amplitude
1e-26
1e-25
Figure 9. The factor by which Fstat increases over the single
template value (Fstat = 11.4) plotted against the predicted am-
plitude for the mountain scenario (Equation 5). The brightest and
most promising sources are the most poorly constrained, and this
is reflected in the effect on Fstat. The difference in Fstat between
the two strongest sources, Sco X-1 and GX 5-1, is primarily due
to the fact that Sco X-1 has much better orbital constraints.
0
to the coherent step). Equation ((40)) then yields an esti-
mate of the search sensitivity hsens
for each potential source.
We take the statistical factor Fstat in Equation ((40)) using
just the number of templates in the coherent step. There
is as yet no systematic study of the semi-coherent metric
for binary systems, and this estimate should be updated as
soon as these calculations are available. We expect that this
approximation will not make a qualitative difference to our
final results because of the weak dependence of Fstat on the
number of templates, but this needs to be verified. In the
previous section, we estimated the estimated GW amplitude
hexp
can be a use-
0
0
ful "detectability" ranking for each source.
from each source, and the ratio hexp
0 /hsens
In the cases when Porb is essentially unknown, the
ranges of Tasc and ax sin i that we have chosen depend on
Porb. This means that the parameter space region is not rect-
angular and thus the template counting equations of Section
4.1 will over-estimate the number of templates significantly,
though not by orders of magnitude.
5.3 Numbers of templates
We can now compute the effect on our searches of param-
eter uncertainty. We will start by considering the effect on
the statistical factor Fstat, neglecting computational cost is-
sues. In other words, we assume that it is feasible to do a
coherent fold. In Tables 2, 3 and 4 we summarize the param-
eter ranges assumed for the pulsars, bursters and kHz QPO
sources respectively. The Tables show the associated Nλi
for each parameter (for Tobs = 2 years), the resulting total
number of templates that must be searched Ntemp, and the
effect on Fstat. Figure 9 shows the change in the statistical
factor for each source.
In Figure 2, were we showed results for a single template
search, each source had Fstat = 11.4. This meant that we
could plot a single detectability threshold curve for each de-
tector. When each source has a different Fstat, however, the
detectability threshold curves differ for each source. One way
of comparing detectability for sources with different noise
threshold curves is to plot the ratio of emitted to detectable
amplitude: this is the approach that we took for the tran-
sients (Figure 5), and that we will adopt when we come to
consider computational cost. What we can also do, however,
to see the effect, is to scale the predicted amplitude h0 by
11.4/Fstat. Predicted amplitude does not really change, of
course, but it is a useful way of visualising the impact of the
statistical factor.
Figure 10 shows the impact on the mountain scenario: it
should be compared to Figure 2. Only for the most tightly
constrained pulsars is detectability largely unchanged: for
the majority of sources the detectable amplitude falls by a
factor of 2-3 compared to a single template search. Although
this does not sound like a great deal, it is sufficient to push
all sources except Sco X-1 below the detection threshold for
Advanced LIGO broad band23. The effect on the r-mode sce-
nario, summarized in Tables 2-4, is slightly less severe, leav-
ing Sco X-1, GX 5-1, GX 349+2 and 4U 1820-30 above the
detection threshold for Advanced LIGO broad band. The sit-
uation is better for the narrow band configuration: although
none of the burst oscillation sources remain within range (for
any of the emission scenarios considered), several of the kHz
QPO sources are still viable. The spins for these sources are
poorly constrained. However, the anticipated narrow band
configurations (see top panel of Figure 1) have a reasonably
broad bandwidth, leaving ample scope for searches.
We can now take the final step and look at the impact
of computational constraints on searches involving multi-
ple templates. The available computational power sets the
length of data Tobs that can feasibly be analysed within a
given amount of time. In Table 5 we summarize the impact in
the situation where we assume a maximum analysis time of
2 years, assuming that for Advanced LIGO we have compu-
tational power 50 times greater than that currently available
within the LSC, while for the Einstein Telescope we assume
100 times more computing than at present. These assump-
tions are, of course, arbitrary, since the computing power
that can be applied to future searches depends not just on
technology (e.g. Moore's Law) but also on improvements in
data analysis techniques, and of course also on how much
money is spent on computing by future projects. We have
taken numbers we feel are defensible, but they may turn out
to be conservative, especially for the Einstein Telescope.
For the pulsars, computing constraints have little effect
on the detectability of the sources; however, these are not
likely to be detectable in any case. But for the bursters and
the kHz QPO sources, computational constraints lead to a
major reduction in the Tobs that it is feasible to analyse. For
many sources we have to resort to a semi-coherent search,
as a coherent fold is no longer possible. Figures 11 and 12
show the effects on detectability for the mountain and r-
mode scenarios respectively, taking into account both the
change in Fstat and the reduced observation times for our
assumed computing power.
The additional impact of computational limitation is
substantial. For Advanced LIGO's broad band configura-
tion, only Sco X-1 and 4U 1820-30 (the latter in the r-mode
23 Sco5 X-1 would also be marginal for Enhanced LIGO, over a
restricted frequency range.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
25
- L I G O
I
E - L I G O
Pulsars
Bursters
kHz QPO
A-LIGO
A-LIGO NB
E T
A-LIGO NB TH
1e-25
1e-26
1e-27
1e-28
0
h
e
d
u
t
i
l
p
m
A
1000
Gravitational wave frequency (Hz)
500
Figure 10. Effect on detectability for the mountain scenario (assuming long-term average flux and a coherent fold with Tobs = 2 years),
taking into account the effect on Fstat associated with the fact that Ntemp > 1. Compare to the best case detectability shown in Figure
2. As in Figure 2, the noise curves are computed assuming Fstat = 11.4 (the single template value), but we have scaled the predicted
amplitudes to reflect the fact that Fstat is larger. Although this is not strictly "correct" (it is the thresholds that should move, not the
predicted amplitudes) this is a useful way to visualise the impact. See the text for more details.
scenario) remain above the detectability threshold. These
two sources are also the only two left above the narrow band
envelope threshold, although there are other sources just be-
low the envelope that might be detectable if the narrow band
configuration were able to push closer to the thermal noise
floor.
Parameter uncertainty clearly poses a major problem,
even for stars where we have some indication of the spin
rate24. If the parameter space volume can be reduced, how-
ever, then the statistical and computational restrictions will
have less impact. In the following section we will examine
this in more detail and consider how best to solve it.
6 DISCUSSION
6.1 Current and future prospects
It is clear from Section 5 that the detection of gravitational
waves from accreting neutron stars is a difficult task. The
X-ray bright kHz QPO sources suffer from parameter un-
certainty, forcing us to look in addition at the much weaker
24 This problem is of course even more pronounced for the ac-
creting neutron stars where we have no indication of spin rate.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
but better constrained burst oscillation sources. The best
constrained sources, the accreting millisecond pulsars, are
expected to emit at such a low level that they are unlikely
to be detectable by second generation detectors. This is par-
ticularly depressing since our calculations have been carried
out within the context of a fairly optimistic modelling and
analysis scenario. We have not as yet considered any addi-
tional sources of spin-down, and have also neglected physi-
cally reasonable complications such as spin variability which
would increase the number of templates still further. Spin
variability would of course also lead to decoherence of the
signal for sources where we cannot track the spin, reducing
integration times to the decoherence time Tdecoh, defined by
T 2
decoh νs = 1.
(43)
One can get some idea of the worst case scenario by calculat-
ing the νs that would result if the source were spinning up at
the maximal rate due to the accretion torque (Equation 1).
The results are given in Table 5: for the bright kHz sources
this worst case decoherence time could be ∼ 1 week, and
this is something that will need to be considered in future
studies of detectability.
In order to improve prospects for those sources that
are in principle detectable, we clearly need to find ways of
Pulsars
Bursters
kHz QPO
26 Watts et al.
Single template
Multiple templates
A-LIGO BB
A-LIGO NB
ET
)
e
d
u
t
i
l
p
m
a
e
l
b
a
t
c
e
t
e
d
/
e
d
u
t
i
l
p
m
a
d
e
t
c
i
d
e
r
P
(
0
1
g
o
L
2
0
-2
2
0
-2
2
0
-2
0
500
1000
0
500
1000
Gravitational wave frequency (Hz)
Figure 11. Detectability for the mountain scenario (long-term average flux, maximum Tobs = 2 years), taking into account computational
limitations as well as the effect on Fstat. The computational cost of searching multiple templates affects the feasible integration time and
the choice of coherent/semi-coherent search technique (Table 5). In order to compare detectability for sources with different integration
times we plot log10 of the ratio of predicted to detectable amplitude. The three rows are for different detector configurations: Top -
Advanced LIGO broad band; Middle - Advanced LIGO narrow band envelope; Bottom - Einstein Telescope. The left hand panels show
detectability for the best case (single template) search, as in Figure 2: the right hand panels show the situation after including the
statistical and computational limitation. For Advanced LIGO we assume a 50-fold increase in computational efficiency as compared to
the current situation; for the Einstein Telescope we assume a 100-fold increase. In both cases we assume a maximum analysis time of 2
years.
improving source constraints and removing computational
limitations. In the sections that follow we detail the actions
that would lead to the biggest improvements.
6.2 Astronomical observations
The major obstacle to detection by the current and immi-
nent generation of detectors is clearly parameter uncertainty,
with spin uncertainty having the largest impact (see Ta-
bles 2-4). Astronomical observations might help to constrain
source properties.
The single most valuable thing that could be done to
improve the current situation is to determine more precisely
the NS spin in the bright kHz QPO and burst sources. A
substantial archive of high time resolution X-ray data exists
for many of these sources from RXTE and its predecessors.
However the only deep search for pulsations in the literature
is that carried out by Dib et al. (2005) for 4U 1820-30. Se-
rious consideration should be given to carrying out similar
analysis for all of the kHz QPO and burst sources, most par-
ticularly Sco X-1. Thorough searches for intermittent pulsa-
tions, such as those carried out by Casella et al. (2008) and
Altamirano et al. (2008), would also be worthwhile.
It may also be possible to find burst oscillations in
the kHz QPO sources: five of the sources that we have
analysed are known to burst. If a sufficient archive of
bursts can be built up from these sources burst oscilla-
tions may well be detected either in individual bursts or
by stacking power spectra from multiple bursts (the tech-
nique used to find weak burst oscillations from EXO 0748-
676 (Villarreal & Strohmayer 2004). We also need to verify
the burst oscillation frequencies for the seven sources with
tentative or single burst detections, in particular the four
that are above the detection threshold for some of the sce-
narios that we have examined: XB 1254-690, 4U 1916-053,
MXB 1730-335 and 4U 0614+09.
Identifying the orbital period can also make a substan-
tial difference to the number of templates searched. Con-
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
27
Single template
Multiple templates
Pulsars
Bursters
kHz QPO
A-LIGO BB
A-LIGO NB
ET
)
e
d
u
t
i
l
p
m
a
e
l
b
a
t
c
e
t
e
d
/
e
d
u
t
i
l
p
m
a
d
e
t
c
i
d
e
r
P
(
0
1
g
o
L
2
0
-2
2
0
-2
2
0
-2
0
500
Gravitational wave frequency (Hz)
0
500
Figure 12. As Figure 11 but for the r-mode scenario.
sider for example the difference between the two kHz QPO
sources Sco X-1 and GX 5-1. Both are expected to be strong
emitters, but Sco X-1 suffers much less from parameter un-
certainty than GX 5-1 because the former has a well con-
strained orbital period. Three of the most promising kHz
QPO sources have no orbital constraints (GX 340+0, GX
5-1 and GX 17+2), while a fourth (GX 349+2) has only
a relatively weak constraint. Several of the most promising
burst oscillation sources also fall into this category: atten-
tion should focus on 4U 1608-522, KS 1731-260, 4U 1728-
34, 4U 1702-429, and MXB 1730-335, sources that could be
detectable in some scenarios. Immense progress has been
made in recent years in constraining orbital parameters for
LMXBs using optical/IR observations, particularly with the
Bowen technique of spectroscopy. Identification of counter-
parts is often difficult, but there would be a substantial pay-
off in terms of gravitational wave detection prospects.
6.3 Astrophysical modelling issues
Both burst oscillation frequency and kHz QPO frequency
are proxy measures for the stellar spin rate. The precise link
to spin for each measure is not clear because in neither case
do we understand the mechanism. Astrophysical modelling
to pin down the cause of the two phenomena could therefore
shed light on the reliability and accuracy of the spin proxy.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Several models have been suggested for burst oscilla-
tions. These include global oscillations of the surface lay-
ers of the neutron star, large-scale drifting vortices excited
by the passage of the flame fronts, or hotspots linked in
some way to persistent pulsations. All of the models have
shortcomings, and progress towards understanding this phe-
nomenon has to some degree stalled. The situation for the
kHz QPOs is however even worse. There are several mod-
els, summarized in detail in van der Klis (2006). All involve
either orbital motions of material within the disk or disk os-
cillations, and many also require some mechanism to select
preferred radii within the disk. Developing testable predic-
tions that would distinguish between the different models,
however, is very difficult. A substantial body of data exists
within the RXTE archive to test models if such predictions
can be made. The identification of a robust link between kHz
QPO frequency and stellar spin would have a major impact
in reducing parameter uncertainty for the most promising
sources for Advanced LIGO.
6.4 Data analysis and detector issues
In the absence of computational limitations, even taking
into account the effect on statistics, there would be up to
8 sources potentially detectable by Advanced LIGO (Figure
10). We therefore need to improve the data analysis tools
28 Watts et al.
used for the searches. We need significant improvements in
all relevant aspects of the data analysis pipelines: acceler-
ating existing software through improvements to both soft-
ware algorithms and computer hardware, and also develop-
ing other data analysis techniques. The software algorithms
being used in the LIGO data analysis software have im-
proved continually over the years since the first LSC publi-
cation on periodic waves in 2003; we expect these improve-
ments to continue. It is possible that we may get an improve-
ment of more than a order of magnitude in existing codes
over the next several years. The improvements in computer
hardware, even just following Moore's law, will yield an ad-
ditional factor of about 16 in the next 8 years, i.e. by the
time we expect Advanced LIGO to be operating. Further-
more, computing platforms like Einstein@Home allow us to
increase the total number of computers available to do the
analysis. On top of this, it might also be possible to use spe-
cial purpose hardware for the periodic wave searches. Most
of the analysis involves a large number of relatively simple
operations, and it might be quite feasible to design chips
which are efficient for these particular calculations. An ex-
ample of these are the GRAPE special-purpose computers
which have proven to be extremely useful in astrophysical
N-body simulations (Makino & Taiji 1995) and in molecu-
lar dynamics calculations in condensed matter physics. It is
hard to anticipate these developments, but an improvement
of 4 or 5 orders of magnitude in effective computing power
might be feasible.
Apart from these technical improvements, it is also pos-
sible to develop new analysis methods. A good example of
this is the cross-correlation method used in the Sco X-1
search (Abbott et al. 2007c). This was neither a matched
filter nor a standard time-frequency semi-coherent search. It
was instead based on aperture synthesis, i.e. the fact that we
have multiple detectors in operation, and that they all see es-
sentially the same GW signal at any given time. This method
has so far mostly been used in the stochastic background
searches, and is now being adapted to the periodic wave
searches (Dhurandhar et al. 2008). Apart from the compu-
tational efficiency, this method is also relatively insensitive
to the uncertainty in the signal model caused, e.g. by pul-
sar glitches (though this leads to a correspondingly greater
computational cost in any follow-ups that must be done for
verifying a detection and for parameter estimation). Further
searches using this method, in combination with the other
methods discussed in this paper seem to be quite promising,
especially given that by the time of Advanced LIGO there
will be a third interferometer of comparable sensitivity in
Virgo.
In the near term while these improvements are in
progress,
it seems reasonable to search for some of the
brighter sources such as Sco X-1 not at the frequency implied
by the kHz QPOs, but rather at lower frequencies corre-
sponding to the sweet spot of the detectors. Not only are the
instruments more sensitive at these frequencies, but this is
computationally easier and is astrophysically well motivated
if the link between the kHz QPO and the spin frequency is
not validated.
We conclude this section with a brief discussion of pos-
sible Advanced LIGO configurations and some of its impli-
cations for our purposes. Figure 13 shows noise curves for
three cases: an example broad band configuration with some
1e-23
A-LIGO Zero-Detuned
S
N
O B
A-LIG
V
N
B E
O N
A -LI G
)
2
/
1
-
z
H
(
D
S
P
e
s
i
o
N
n
i
a
r
t
S
1e-24
10
100
1000
Frequency (Hz)
Figure 13. Some possible Advanced LIGO configurations. See
text for discussion.
tuning of the signal recycling cavity optimized for binary
neutron star (BNS) inspirals, the envelope of possible nar-
row band configurations, and the zero-detuned broad band
configuration. The zero-detuned curve and the narrow band
envelope are the same as in Fig. 1. The BNS curve is relevant
because binary neutron star inspirals will be among the key
targets for Advanced LIGO and the detector might possibly
be operated in this configuration for significant durations.
Thus, while BNS signals are not our concern in this paper,
it is important to consider the impact of this configuration
for our purposes. It is obvious that the BNS curve affects us
adversely above ∼ 500 Hz which, as we see from Figure 2 is
precisely where the kHz QPO sources are expected to lie.
These curves highlight the importance of spin measure-
ments. For all potential sources with frequencies between,
say, 150 and 2000 Hz, the narrow banding gains us no more
than a factor of 2. The best case for narrow banding is then
a compelling source whose frequency is very well known and
is within this range and is not detectable by the broad-band
curve. The other possibility is to employ the zero-detuned
noise curve for detection and then use the narrow band curve
as a follow up to confirm detection or to increase SNR for ex-
tracting better astrophysical information. Given the present
uncertainties in the spin frequency, this latter possibility
seems to be the best option for detecting periodic waves
from LMXBs. This conclusion could change if there turn
out to be significant improvements in the narrow band noise
curves.
6.5 Conclusions
We have shown that gravitational wave observations of ac-
creting neutron stars will be challenging. It is therefore
worthwhile considering what is to be gained from such an ef-
fort: what new information gravitational wave observations
will bring to the study of these sources.
The input from electromagnetic observations that is re-
quired for these searches is primarily on kinematical param-
eters like spin rates and orbital ephemerides. The gravita-
tional waves that are generated carry information on the
orientation and dynamics of the neutron star and its inter-
action with the accretion disk. The elliptical polarization
of the gravitational waves, which will emerge from anal-
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
29
ysis of the detected signal, determines the orientation of
the neutron-star spin axis. The degree of alignment of this
with the inferred disk orientation, especially for the slower-
spinning (presumably younger) systems, will constrain mod-
els of binary evolution and the formation of the NS.
In systems where observations and modelling constrain
the NS spin to a relatively narrow range, the GW frequency
will determine whether the r-mode or mountain scenario, or
indeed some other scenario, is the appropriate one. The GW
amplitude will then determine the degree of mass asymme-
try (mountains) or the size of the velocity field (r-modes).
In both cases this will open for the first time a wealth of
opportunities for studying the physics of NS interiors.
For all detected systems, the measured GW amplitude
and frequency will tell us how much angular momentum is
being carried away in GWs; by measuring or limiting the
rate of change of the GW frequency we can then infer the
rate at which angular momentum is being accreted from the
disk. This will be an important constraint on models of the
disk and the NS magnetic field.
Clearly, if GW searches turn up signals in unexpected
places in parameter space, this will challenge prevailing mod-
els for these systems. For example, if the GW frequency
equals the X-ray pulsation frequency instead of twice or four
thirds its value, then this might imply that the pulsation fre-
quency is actually twice the true spin frequency.
Even in the worst case, where a sufficiently sensitive
search fails to detect GWs at the expected amplitude, the
negative result could eliminate the GW option for spin
balance and demonstrate that this somehow resulted from
a propeller-type torque from the NS's magnetic field. Of
course, one would have to have confidence that the GWs
were not coming out at a different frequency, so the search
would have to include that as a parameter.
The additional astronomical input that could make the
difference between detecting or not detecting GWs could
come from a variety of observations. Clearly, long-term X-
ray timing is highly desirable, and here it is disappointing
that the future of RXTE is limited and no X-ray mission
with comparable capabilities seems to be planned for the
period before Advanced LIGO comes online. Thorough ex-
ploitation of the RXTE data archive is therefore especially
important. On the other hand, sensitive radio observations
of X-ray systems in quiescence may provide unexpected in-
formation, and this will be easier with arrays like the Square
Kilometre Array and its pathfinders.
7 ACKNOWLEDGEMENTS
We would like to thank Alan Levine for advice on ASM
countrates, Jake Hartman for advice on the effects of po-
sitional uncertainty on pulsar timing, Reinhard Prix for
help with the computational cost estimates, Alberto Vecchio
and Chris Messenger for discussions regarding the parame-
ter space metric, and Maria Alessandra Papa, Mike Landry
and Graham Woan for useful comments on a draft of this
paper. We are grateful to Rana Adhikari, Peter Fritschel,
Gregg Harry, Bangalore Sathyaprakash, David Shoemaker,
Kentaro Somiya and Ken Strain for pointing us to Bench
and for valuable discussions related to the Advanced LIGO
noise spectrum. We thank Rana Adhikari also for providing
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
us with the Enhanced LIGO noise curve. ALW would like
to thank the Kavli Institute for Theoretical Physics for hos-
pitality during a KITP/MPA postdoctoral exchange visit,
when this work was started. BK is grateful to the LSC
continuous-waves data analysis group for valuable discus-
sions. LB acknowledges support from the National Science
Foundation via grants PHY 05-51164 and AST 02-05956.
BK and BFS acknowledge the support of the DFG's spe-
cial research centre SFB/Transregio 7 "Gravitational Wave
Astronomy".
REFERENCES
Abbott B. et al., 2004, Phys. Rev. D 69, 182004
Abbott B. et al., 2005b, Phys. Rev. Lett. 94, 181103
Abbott B. et al., 2007b, Phys. Rev. D 76, 042001
Abbott B. et al., 2005a, Phys. Rev. D 72, 102004
Abbott B. at al., 2008, Phys. Rev. D 77, 022001
Abbott B. et al., 2007a, Phys. Rev. D 76, 082001
Abbott B. et al., 2007c, Phys. Rev. D 76, 082003
Altamirano, D., van der Klis, M., M´endez, M., Migliari, S.,
Jonker, P.G., Tiengo, A., Zhang, W. 2005, ApJ, 633, 358
Altamirano D., Casella P., Patruno A., Wijnands R., van
der Klis M., 2008, ApJ, 674, L45
Anderson, S.F., Margon, B., Deutsch, E.W., Downes, R.A.,
Allen, R.G., 1997, ApJ, 482, L69
Andersson N., Kokkotas K.D., Stergioulas N. 1999, ApJ,
516, 307
Andersson N., Jones D.I., Kokkotas K.D., Stergioulas N.
2000, ApJ, 534, L75
Andersson N., Kokkotas K.D., 2001, Int. J. Mod. Phys.,
10, 381
Andersson N., Jones D.I., Kokkotas K.D. 2002, MNRAS,
337, 1224
Andersson N., Glampedakis K., Haskell B., Watts A.L.
2005, MNRAS, 361, 1153
Arras P., Flanagan E.E., Morsink S.M., Schenk A.K.,
Teukolsky S.A., Wasserman I., 2003, ApJ, 591, 1129
Astone P. et al. 2005, Class. Quant. Grav., 22, S1243-S1254
Astone P. et al. 2002, Phys.Rev.D, 65, 042003
Augusteijn T., van der Hooft F., de Jong J.A., van Kerk-
wijk M.H., van Paradijs J. 1998, A&A, 332, 561
Ballantyne D.R., Strohmayer T.E., 2004, ApJ, 602, L105
Baraffe I., Kolb U, 2000, MNRAS, 318, 354
Barnes A.D., Casares J., Cornelisse R., Charles P.A.,
Steeghs D., Hynes R.I., O'Brien K., 2007, MNRAS, 380,
1182
Barret D., Olive J.-F., Miller M.C. 2005, MNRAS, 361, 855
Barret D., Boutelier M., Miller M.C., 2008, MNRAS, 384,
1519
Barziv O., Kuulkers E., M´endez M., van der Hooft F.,
Groot P.J., van der Klis M., Kemper C., van Paradijs
J. 1997, A&A, 325, 1035
Berendsen S.G.H., Fender R., Kuulkers E., Heise K., van
der Klis M. 2000, MNRAS, 318, 599
Bhattacharyya S., Strohmayer T.E., Markwardt C.B.,
Swank J.H. 2006, ApJ, 639, L31
Bhattacharyya S. 2007, MNRAS, 377, 198
Bildsten L., 1998, ApJ, 501, L89
Bildsten L. & Chakrabarty, D., 2001, ApJ, 557, 292
Bodghee A., Mowlavi N., Kuulkers E., Wijnands R., Shaw
30 Watts et al.
S.E., Courvoisier T., Markwardt C., Oosterbroek T., Orr
A., Paizis A., Ebisawa K., Kretschmar P. 2005, Atel 592
Boirin L., Barret D., Olive J., Bloser P., Grindlay J., 2000,
A&A, 361, 121
122001
Dhurandhar S.V., Krishnan B., Mukhopadhyay H., and
Whelan J., Phys. Rev. D 77, 082001
Dib, R., Ransom, S.M., Ray, P.S., Kaspi, V.M., Archibald,
Bondarescu R., Teukolsky S.A., Wasserman I., 2007, Phys.
A.M. 2005, ApJ, 626, 333.
Rev. D 76, 64019
di Salvo T., M´endez M., van der Klis M., Ford E., Robba
Boutloukos S., van der Klis M., Altamirano D., Klein-Wolt
M., Wijnands R., Jonker P.G., Fender R.P. 2006, ApJ,
653, 1435
N.R. ApJ, 546, 1107
di Salvo T., M´endez M., van der Klis M. 2003, A&A, 406,
177
Bradshaw C.F., Fomalont E.B., Geldzahler B.J. 1999, ApJ,
Dubus G., Lasota J.-P., Hameury J.-M., Charles P., 1999,
512, 121
Bradt H.V.D., McClintock J.E. 1983, Ann. Rev. Astron.
Astrophys. 21, 13
Brady P.R., Creighton T., Cutler C., Schutz B. 1998, Phys.
Rev. D. 57, 1
Brady P.R., Creighton T., 2000, Phys. Rev. D, 61, 082001
Brandt S., Castro-Tirado A.J., Lund N., Dremin V., Lap-
shov I., Sunyaev R.A. 1992, A&A, 262, L15
Buonanno, A & Chen, Y 2002, Phys.Rev.D, 65, 042001
Burderi L., di Salvo T., Lavagetto G., Menna M.T., Papitto
A., Riggio A., Iaria R., d'Antona F., Robba N.R., Stella
L. 2007, ApJ, 657, 961
Callanan P.J., Curran P., Filippenko A.V., Garcia A.V.,
Margon B., Deutsch E., Anderson S., Homer L., Fender
R.P. 2002, ApJ, 574, L143
Casares J., Charles P., Kuulkers E. 1998, ApJ, 493, L39
Casares J., Dubus G., Shahbaz T., Zurita C., Charles P.A.
2002, MNRAS, 329, 29
Casares J., Cornelisse R., Steeghs D., Charles P.A., Hynes
R.I., O'Brien K., Strohmayer T.E. 2006, MNRAS, 373,
1235
Casella P., Altamirano D., Patruno A., Wijnands R., van
der Klis M., 2008, ApJ, 674, L41
Chabrier G., Baraffe I., 2000, ARA&A, 38, 337
Chabrier G., Baraffe I., Allard F., Hauschildt P., 2000, ApJ,
542, 464
Chakrabarty D., Morgan E., Muno M., Galloway D., Wij-
nands R., van der Klis M., Markwardt C., 2003, Nature,
424, 42
Chevalier C., Ilovaisky S.A. 1998, IAUC 6806.
Chou, Y., Grindlay, J.E. 2001, ApJ, 563, 934
Chou Y., Grindlay J.E., Bloser P.F. 2001, ApJ, 549, 1144
Chou Y., Chung Y., Hu C.P., Yang T.C., 2008, ApJ, 678,
1316
Christian D.J., Swank J.H. 1997, ApJS, 109, 177
Cook, G. B., Shapiro, S. L. & Teukolsky, S. A. 1994, ApJ,
423, L117
Cooke B.A., Ponman T.J. 1991, A&A, 244, 358
Cominsky L.R., Wood K.S., 1984, ApJ, 283, 765
Cornelisse R., Casares J., Steeghs D., Barnes A.D., Charles
P.A., Hynes R.I., O'Brien K., 2007, MNRAS, 375, 1463
Courvoisier T.J.-L., Parmar A.N., Peacock A., Pakull M.
1986, ApJ, 309, 265
Cutler C. 2002, Phys. Rev. D., 66, 084025
Cutler C., Gholami I., Krishnan B. 2005, Phys. Rev. D, 72,
042004
Cutler C., Schutz B.F. 2005, Phys.Rev.D, 72, 063006
Degenaar N. et al., 2007, Atel 1106
Deloye C.J., Bildsten L., 2003, ApJ, 598, 1217
Deutsch E.W., Margon B., Anderson S.F., Wachter S.,
Goss W.M. 1999, ApJ, 524, 406
Dhurandhar S.V., Vecchio A., 2001, Phys. Rev. D 63,
MNRAS, 303, 139
Falanga M., Kuiper L., Poutanen J., Bonning E.W.,
Hermsen W., di Salvo T., Goldoni P., Goldwurm A., Shaw
S.E., Stella L. 2005, A&A, 444, 15
Fender R.P., Hendry M.A., 2000, MNRAS, 317, 1
Ford E., Kaaret P., Tavani M., Barret D., Bloser P., Grind-
lay J., Harmon B.A., Paciesas W.S., Zhang S.N. 1997,
ApJ, 475, L123
Ford E.C., van der Klis M., van Paradijs J., M´endez M.,
Wijnands R., Kaaret P., 1998, ApJ, 508, L155
Ford E.C., van der Klis M., M´endez M., Wijnands R.,
Homan J., Jonker P.G., van Paradijs J., 2000, ApJ, 537,
368
Fox D.W., Lewin W.H.G., Rutledge R.E., Morgan E.H.,
Guerriero R., Bildsten L., van der Klis M., van Paradijs
J., Moore C.B., Dotani T., Asai K. 2001, MNRAS, 321,
776
Fox D.B., Kulkarni S.R., 2004, Atel 354
Fox D.B. 2005, Atel 526, 1
Freire, P. C. C., Ransom, S. M., Begin, S., Stairs, I. H.,
Hessels, J. W. T., Frey, L. H., Camilo, F. 2008, ApJ, 675,
670
Galloway D., Chakrabarty D., Muno M., Savov P., 2001,
ApJ, 549, L85
Galloway D., Chakrabary D., Morgan E.H., Remillard R.A.
2002, ApJ, 576, L137
Galloway D., Cumming A., Chakrabarty D., 2004,
AAS/High Energy Astrophysics Division, 8,
Galloway D.K., Markwardt C.B., Morgan, E.H.,
Chakrabarty, D., Strohmayer, T.E., 2005, ApJ, 622,
L45
Galloway D.K., Accretion powered millisecond pulsar out-
bursts, 2006, in 'The transient Milky Way: a perspective
for MIRAX', eds F. D'Amico, J. Braga & R. Rothschild,
AIP Conf. Proc, 840, 50
Galloway D.K., Cumming A. 2006, ApJ, 652, 559
Galloway D.K., Psaltis D., Muno M.P., Chakrabarty D.
2006, ApJ, 639, 1033
Galloway D.K., Morgan E.H. Krauss M.I., Kaaret P.,
Chakrabarty D. 2007, ApJ, 654, L73
Galloway D.K., Muno M.P., Hartman J.M., Savov P.,
Psaltis D., Chakrabarty D. 2008, ApJS submitted,
astro-ph/0608259.
Galloway D.K., Ozel F., Psaltis D., MNRAS, 387, 268
Garcia M.R., Callanan P.J., McCarthy J., Eriksen K.,
Hjellming R.M. 1999, ApJ, 518, 422
Gavriil F.P., Strohmayer T.E., Swank J.H., Markwardt
C.B., 2007, ApJ, 669, L29
Ghosh P., Lamb F. 1978, ApJ, 223, L83
Giacconi R., Gorenstein P., Gursky H., Usher P.D., Waters
J.R., Sandage A., Osmer P., Peach J.V. 1967, ApJ, 148,
L129
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
31
Gierli´nski M., Done C., 2002, MNRAS, 331, L47
Gierli´nski M., Done C., Barret D., 2002, MNRAS, 331, 141
Gierli´nski M., Poutanen J., 2005, MNRAS, 343, 1301
Giles A., Hill K., Strohmayer T., Cummings N., 2002, ApJ,
Paradijs J., M´endez M., Ford E., Kuulkers E., Lamb F.,
2000, ApJ, 537, 374
Jonker P., M´endez M., van der Klis M., 2002, MNRAS,
336, L1
568, 279
Jonker P., van der Klis M., Homan J., M´endez M., Lewin
Giles A.B., Greenhill J.G., Hill K.M., Sanders E. 2005, MN-
W., Wijnands R., Zhang W., 2002, MNRAS, 333, 665
RAS, 361, 1180
Jonker P.G., Nelemans G., Bassa C.G. 2007, MNRAS, 374,
Gottlieb E.W., Wright E.L., Liller W., 1975, ApJ, 195, L33
Gottwald M., Steinle H., Pietsch W., Graswer U. 1991, As-
999
Juett A.M., Psaltis D., Chakrabarty D. 2001, ApJ, 560,
tronomy & Astrophysics Supplement Series, 89,
L59
Greenhill J.G., Giles A.B., Hill K.M. 2002, IAUC 7889, 1
Grindlay, J.G., Liller, W. 1978, ApJ, 220, L127
Hartman J., Chakrabarty D., Galloway D., Muno M.,
Savov P., M´endez M., van Straaten S., di Salvo T., 2003,
AAS/High Energy Astrophysics Division, 7,
Hartman J., Patruno A., Chakrabarty D., Kaplan D.L.,
Markwardt C.B., Morgan E.H., Ray P.S., van der Klis
M., Wijnands R. 2008, ApJ, 675, 1468
Hasinger, G., van der Klis M., 1989, A&A, 225, 79
Haskell B., Jones D.I., Andersson N. 2006, MNRAS, 373,
1423
Kaaret P., in't Zand J., Heise J., Tomsick J., 2002, ApJ,
575, 1018
Kaaret P., Morgan E.H., Vanderspek R., Tomsick J. 2006,
ApJ, 638, 963
Kaaret P., Prieskorn Z., in't Zand J.J.M., Brandt S., Lund
N., Mereghetti S., Gotz D., Kuulkers E., Tomsick J.A.
2007, ApJ, 657, L97
Kawai N., Suzuki M. 2005, Atel, 534
King A.R., Kolb U., Burderi L., 1996, ApJ, 464, L127
Kinney J.B., Mendell G. 2003, Phys. Rev. D, 67. 024032
Klein-Wolt M., Wijnands R., Swank J.H., Markwardt C.B.,
Haskell B., Samuelsson L., Glampedakis K., Andersson N.,
2007, Atel 1065
2008, MNRAS, 385, 531
Klein-Wolt M., Maitra D., Wijnands R., Swank J.H., Mark-
Heasley, J.N., Janes, K.A., Zinn, R., Demarque, P., da
wardt C.B., Bailyn C., 2007, Atel 1070
Costa, G.S., Christian, C.A. 2000, AJ, 120, 879
Klein-Wolt M., Wijnands R., Swank J.H., Markwardt C.B.,
Hertz P., Vaughan B., Wood K.S., Norris J.P., Mitsuda K.,
2007, Atel 1075
Michelson P.F., Dotani T. 1992, ApJ, 396, 201
Hessels J.W.T., Ransom S.M., Stairs I.H., Freire P.C.C.,
Kaspi V.M., Camilo F. 2006, Science, 311, 1901
Hessels J.W.T., Ransom S.M., Stairs I.H., Kaspi, V. M. &
Freire P.C.C. 2007, ApJ, 670, 363
Heyl J. 2002, ApJ, 574, L57
Heyl J.S. 2004, ApJ, 600, 939
Homan J., M´endez M., Wijnands R., van der Klis M., van
Paradijs J., 1999, ApJ, 513, L119
Kolb U., King A.R., Baraffe I., 2001, MNRAS, 321, 544
Krauss M.I., Wang Z., Dullighan A., Juett A.M., Ka-
plan D., Chakrabarty D., van Kerkwijk M.H., Steeghs D.,
Jonker P.G., Markwardt C.B. 2005, ApJ, 627, 910
Krauss M.I., Juett A.M., Chakrabarty D., Jonker P.G.,
Markwardt C.B. 2006, Atel 777
Krimm H.A. et al., 2007, ApJ, 668, L147
Krishnan B. et al. 2004, Phys. Rev. D, 70, 082001
Kuulkers E., van der Klis M., van Paradijs J. 1995, ApJ,
Homan J., van der Klis M., Jonker P., Wijnands R., Ku-
450, 748
ulkers E., M´endez M., Lewin W., 2002, ApJ, 568, 878
Kuulkers E., van der Klis, Oosterbroek T., van Paradijs J.,
Homer L., Deutsch E.W., Anderson S.F., Margon B. 2001,
Lewin W.H.G. 1997, MNRAS, 287, 495
AJ, 122, 2627
Horne K., Wade R.A., Szkody P., 1986, MNRAS, 219, 791
Hynes R.I., Horne K., O'Brien K., Haswell C.A., Robinson
E.L., King A.R., Charles P.A., Pearson K.J., 2006, ApJ,
648, 1156
Iaria R., di Salvo T., Robba N.R., Burderi L., Stella L.,
Frontera F., van der Klis M. 2004, ApJ, 600, 358
Iaria R., di Salvo T., Lavagetto G., Robba N.R., Burderi
L. 2006, ApJ, 647, 1341
Iaria R., di Salvo T., Lavagetto G., d'A´ı A., Robba N.R.
2007, A&A, 464, 291
in't Zand J.J.M., Kuulkers E., Verbunt F., Heise J., Cor-
Kuulkers E., Homan J., van der Klis M., Lewin W.H.G.,
M´endez M. 2002, A&A, 382, 947
Kuulkers E., den Hartog P.R., in't Zand J.J.M., Verbunt
F.W.M., Harris W.E., Cocchi M. 2003, A&A, 399, 663
Kuznetsov S.I. 2002, Astronomy Letters, 28, 73
Lattimer J.M., Prakash M. 2007, Physics Reports, 442, 109
Levin Y. 1999, ApJ, 517, 328
Levine A.M., Bradt H., Cui W., Jernigan J.G., Morgan
E.H., Remillard R., Shirey R.E., Smith D.A., 1996, ApJ,
469, L33
Levine A.M., Corbet R. 2006, Atel 940
Linares M., van der Klis M., Altamirano D., Markwardt
nelisse R. 2003, A&A, 411, L487
C.B. 2005, ApJ, 634, 1250
in't Zand J.J.M., Jonker P.G., Markwardt C.B., 2007,
A&A, 465, 953
Lochner J.C., Roussel-Dupre D. 1994 ApJ, 435, 840
Makino J. and Taiji M., 1995, Proc. 1995 ACM/IEEE con-
Jaranowski P., Kr´olak A, Schutz B.F. 1998, Phys. Rev. D
ference on supercomputing, 63.
58, 063001
Markwardt C., Strohmayer T., Swank J., 1999, ApJ, 512,
Johnston H.M., Fender R., Wu K. 1999, MNRAS, 308, 415
Johnston H.M., Wu K., Fender R., Cullen J.G. 2001, MN-
L125
Markwardt C.B., Marshall F.E., Swank J.H., Cui W. 1999,
RAS, 328, 1193
IAUC 7300, 1
Jongert H.C., van der Klis M. 1996, A&A, 310, 474
Jonker P.G., Wijnands R., van der Klis M., Psaltis D., Ku-
Markwardt C., Swank J., Strohmayer T., in't Zand M.,
Marshall F., 2002, ApJ, 575, L21
ulkers E., Lamb F.K. 1998, ApJ, 499, L191
Markwardt C., Juda, M., Swank J., 2003, IAU Circ., 8095,
Jonker P., van der Klis M., Wijnands R., Homan J., van
2
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
32 Watts et al.
Markwardt C., Swank J., 2003, IAU Circ., 8144
Markwardt C.B., Klein-Wolt M., Swank J.H., Wijnands R.
Owen, B.J. 1996, Phys. Rev. D, 53, 6749
Owen B.J., Lindblom L., Cutler C., Schutz B.F., Vecchio
2007, Atel 1068
Mart´ı J., Mirabel I.F., Rodr´ıguez L.F., Chaty S. 1998,
A&A, 332, L45
Meers B.J. 1988, Phys. Rev. D, 38, 2317
Meinshausen N., Bickel P. & Rice J. 2007, arXiv:0712.1663
Melatos A., Payne D.J.B. 2005, ApJ, 623, 1044
M´endez M., van der Klis M., van Paradijs J., Lewin,
W.H.G., Vaughan B., Kuulkers, E., Zhang, W., Lamb,
F.K., Psaltis, D. 1998, ApJ, 494, L65
A., Andersson N., 1998, Phys. Rev. D, 58, 084020
Paizis A., Nowak M.A., Wilms J., Courvoisier T.J.-L., Ebi-
sawa K., Rodriguez J., Ubertini P. 2005, A&A, 444, 357
Papaloizou J., Pringle J.E. 1978, MNRAS, 184, 501
Papitto A., di Salvo T., Burderi L., Menna M.T., Lavagetto
G., Riggio A. 2007, MNRAS, 375, 971
Papitto A., Menna M.T., Burderi L., di Salvo T., Riggio,
A., 2008, MNRAS, 383, 411
Parmar A.N., White N.E., Giommi P., Gottwald M. 1986,
M´endez M., van der Klis M., Wijnands R., Ford E., van
ApJ, 308, 199
Paradijs J., Vaughan B., 1998, ApJ, 505, L23
Patruno A., Altamirano D., Hessels J.W.T., Casella P., Wi-
M´endez M., van der Klis M., van Paradijs J. 1998, ApJ,
jnands R., van der Klis M., 2008, arXiv:0801.1031
506, L117
M´endez M., van der Klis M., 1999, ApJ, 517, L51
M´endez M., van der Klis M., 2000, MNRAS, 318, 938
M´endez M., Belloni T., 2007, MNRAS, 381, 790
Messenger C., Woan G., 2007, Class. Quant. Grav., 24, 469
Middleditch J., Priedhorsky W., 1986, ApJ, 306, 230
Migliari S., van der Klis M., Fender R., 2003, MNRAS, 345,
L35
Migliari, S., Fender, R.P., Rupen, M., Wachter, S., Jonker,
P.G., Homan, J., van der Klis, M. 2004, MNRAS, 351, 186
Migliari S., Fender R.P., 2006, MNRAS, 366, 79
Migliari S., Tomsick J.A., Maccarone T.J., Gallo E., Fender
R.P., Nelemans G., Russell D.M. 2006, 643, L41
Mignani R.P., Chaty S., Mirabel I.F., Mereghetti S., 2002,
A&A, 389, L11
Moore C.B., Rutledge R.E., Fox D.W., Guerriero R.A.,
Lewin W.H.G., Fender R., van Paradijs J. 2000, ApJ, 532,
1181
Payne D.J.B., Melatos A. 2006, ApJ, 641, 471
Penninx W., 1989, in: The 23rd ESLAB Symposium on
Two Topics in X-ray Astronomy. Volume 1: X-ray Bina-
ries, p. 185-196, ESA.
Penninx W., Zwarthoed G.A.A., van Paradijs J., van der
Klis M., Lewin W.H.G., Dotani T. 1993, A&A, 267, 92
Piro A.L., Bildsten L. 2005, ApJ, 629, 438
Pooley D., Lewin W.H.G., Verbunt F., Homer L., Margon
B., Gaensler B.M., Kaspi V.M., Miller J.M., Fox D.W.,
van der Klis M. 2002, ApJ, 573, 184.
Podsiadlowski P., Rappaport S., Pfahl E.D. 2002, ApJ, 565,
1107.
Prix R., Itoh Y. 2005, Class. Quant. Grav., 22, S1003
Prix R., 2007, Class. Quant. Grav., 24, S481
Prix R., 2007, Phys. Rev. D 75, 023004
Rappaport S., Joss P.C., 1984, ApJ, 283, 232
Rappaport S., Ma C.P., Joss P.C., Nelson L.A. 1987, ApJ,
322, 842
Motch C., Pedersen H., Beuermann K., Pakull M.W., Cour-
Ransom S.M., Cordes J.M., Eikenberry S.S., 2003, ApJ,
voisier T.J.-L. 1987, ApJ, 313, 792.
589, 911
Muno M.P., Fox D.W., Morgan E.H., Bildsten L. 2000,
ApJ, 542, 1016
Muno M., Chakrabarty D., Galloway D., Savov P., 2001,
ApJ, 553, L157
Muno M.P., Ozel F., Chakrabarty D., 2002, ApJ, 581, 550
Muno M.P., Remillard R.A., Chakrbarty D., 2002, ApJ,
Ransom S.M., Hessels, J. W. T., Stairs, I. H., Friere, P. C.
C., Camilo, F., Kaspi, V. M., Kaplan, D. L. 2005, Science,
307, 892
Remillard R. 2004, Atel 357.
Retter A., Chou Y., Bedding T.R., Naylor T. 2002, MN-
RAS, 330, L37
568, L35
Revnivtsev M.G., Sunyaev R. 2002, Astronomy Letters, 28,
Muno M.P., Baganoff F.K., Arabadijs J.S. 2003, ApJ, 598,
19
474
Murdin P., Jauncey D.L, Lerche I., Nicolson G.D.,
Kaluzienski L.J., Holt S.S., Haynes R.F. 1980, A&A, 87,
292
Natalucci L., Cornelisse R., Bazzano A., Cocchi M., Uber-
tini P., Heise J., in't Zand J., Kuulkers E., 1999, ApJ, 523,
L45
Nayyar M., Owen B.J. 2006, Phys. Rev. D, 73, 084001
Nelemans G., Jonker P.G., Marsh T.R., van der Klis M.
Riggio A., di Salvo T., Burderi L., Iaria R., Papitto A.,
Menna M.T., Lavagetto G., 2007, MNRAS, 382, 1751
Riggio A., di Salvo T., Burderi L., Menna M.T., Papitto
A., Iaria R., Lavagetto G., 2008, ApJ, 678, 1273
Rupen M.P., Dhawan V., Mioduszewski A.J., 2003, ATel
210
Rupen M. P., Dhawan V., Mioduszewski A.J., 2004, ATel
364
Rupen M. P., Mioduszewski A.J., Dhawan V. 2004, ATel
2004, MNRAS, 348, L7
286
Nelemans G., Jonker P.G., Steeghs D., 2006, MNRAS, 370,
Sathyaprakash B.S., Dhurandhar S.V., Phys. Rev. D, 44,
255
3819
O'Neill P.M., Kuulkers E., Sood R.K., van der Klis M. 2002,
Sathyaprakash B.S., Dhurandhar S.V., Phys. Rev. D, 49,
MNRAS, 336, 217
1707
Oosterbroek T., van der Klis M., Kuulkers E., van Paradijs
Shabaz T., Zurita C., Casares J., Dubus G., Charles P.A.,
J., Lewin W.H.G. 1995, A&A, 297, 141
Wagner R.M., Ryan E. 2003, ApJ, 585, 443
Ooesterbroek T., Parmar A.N., Sidoli L., in't Zand J.J.M.,
Smale A., Mason K., White N., Gottwald M., 1988, MN-
Heise J. 2001, A&A, 376, 532
Orosz J.A., Kuulkers E. 1999, MNRAS, 305, 132
Ortolani S., Barbuy B., Bica E., 1994, A&A, 108, 653
RAS, 232, 647
Smale A.P., Mukai K., Rees Williams O., Jones M.H., Cor-
bet R.H.D., 1992, ApJ, 400, 330
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
33
Smale A.P., Zhang W., White N.E. 1997, ApJ, 483, L119
Smale A.P., 1998, ApJ, 498, L141
Smale A.P. & Wachter S. 1999, ApJ, 527, 341
Smith D., Morgan E., Bradt H., 1997, ApJ, 479, L137
Sosin C., King, I.R. 1995, AJ, 109, 639
Southwell K.A., Casares J., Charles P.A. 1996, in: Cat-
aclysmic Variables and related objects, IAU Coll. 158,
A.Evans and J.H. Wood (eds), Kluwer Academic Pub-
lishers P. 365.
Steeghs D., Casares J. 2002, ApJ, 568, 273
Strohmayer T., Zhang W., Swank J., Smale A., Titarchuk
L., Day C., Lee U., 1996, ApJ, 469, L9
Strohmayer T., Jahoda K., Giles A., Lee U., 1997, ApJ,
486, 355
J. van Paradijs & E.P.J. van den Heuvel (London, Cam-
bridge Univ. Press), 536
van Straaten S., Ford E.C., van der Klis M., M´endez M.,
Kaaret P. 2000, ApJ, 540, 1049
van Straaten S., van der Klis M., di Salvo T., Belloni, T.
ApJ, 568, 912
van Straaten S., van der Klis M., M´endez M., 2003, ApJ,
596, 1155
van Straaten S., van der Klis M., Wijnands R., 2005, ApJ,
619, 455
Vaughan B.A., van der Klis M., Wood K.S., Norris J.P.,
Hertz P., Michelson P.F., van Paradijs J., Lewin W.H.G.,
Mitsuda K., Penninx W. 1994, ApJ, 435, 362.
Verbunt F., van Kerkwijk M.H., in't Zand J.J.M., Heise J.
Strohmayer T., Zhang W., Swank J.H., White N.E.,
2000, A&A, 359, 960
Lapidus I. 1998, ApJ, 498, L135
Strohmayer T.E., Zhang W., Swank J.H., Lapidus I. 1998,
ApJ, 503, L147
Strohmayer T.E., Markwardt C.B. 2002, ApJ, 577, 345
Strohmayer T.E., Markwardt C., Swank J., in't Zand J.,
2003, ApJ, 596, L67
Strohmayer T.E. & Bildsten L. 2006, in Compact stellar
X-ray sources, eds Lewin & van der Klis,Cambridge As-
trophysics Series No. 39 (Cambridge University Press),
p.113-156
Villarreal A. & Strohmayer T.E. 2004, ApJ, 614, L121
Wachter S., Margon B. 1996, AJ, 112, 2684
Wachter S. 1997, ApJ, 490, 401
Wachter S., Smale A.P. 1998, ApJ, 496, L21
Wachter S., Smale A.P., Bailyn C. 2000, ApJ, 534, 367
Wachter S., Hoard D.W., Bailyn C.D., Corbel S., Kaaret
P. 2002, ApJ, 568, 901
Wachter S., Wellhouse J.W., Patel S.K., Smale A.P., Alves
J.F., Bouchet P. 2005, ApJ, 621, 393
Wade R.A., Quintana H., Horne K., Marsh T.R. 1985,
Strohmayer T.E., Markwardt C.B., Kuulkers E., 2008,
PASP, 97, 1092
ApJ, 672, L37
Taylor J.H., Weisberg J.M. 1989, ApJ, 345, 434
Tomsick J., Halpern J., Kemp J., Kaaret P., 1999, ApJ,
521, 341
Tomsick J.A., Heindl W.A., Chakrabarty D., Halpern J.P.,
Wagoner R.V., 1984, ApJ, 278, 345
Wagoner R.V., 2002, ApJ, 578, L63
Watts A.L., Strohmayer T.E. 2006, MNRAS, 373, 769
Welsh W.F., Robinson E.L., Young P. 2000, AJ, 120, 943
Wen L., Levine A.M., Corbet R.H.D., Bradt H.V. 2006
Kaaret P. 2001, ApJ, 559, L123
ApJs, 163, 372
Tomsick J.A., Heindl W.A., Chakrabarty D., Kaaret P.
2002, ApJ, 581, 570
Tomsick J.A., Gelino D.M., Halpern J.P., Kaaret P. 2004,
ApJ, 610, 933
Torres M.A.P., McClinktock J.E., Garcia M.R., Murray
S.S. 2004, Atel 238
Torres M.A.P., Steeghs D., Garcia M.R., McClintock J.E.,
Jonker P.G., Kaplan D., Casares J., Greimel R., Au-
gusteijn T. 2006, Atel 784.
Torres M.A.P., Jonker P.G., Steeghs D., Roelofs G.H.A.,
Bloom J.S., Casares J., Falco E.E., Garcia M.R., Marsh
T.R., M´endez M., Miller J.M., Nelemans G., Rodr´ıguez-
Gil P. 2008, ApJ, 672, 1079
White N.E., Swank J.H., 1982, ApJ, 253, L61
White N., Zhang W. 1997, ApJ, 490, L87
Wijnands R., van der Klis M., 1997, ApJ, 482, L65
Wijnands R., van der Klis M., van Paradijs J., Lewin
W.H.G., Lamb F.K., Vaughan B., Kuulkers E. 1997, ApJ,
479, L141
Wijnands R., Homan J., van der Klis M., M´endez M., Ku-
ulkers E., van Paradijs J., Lewin W.H.G., Lamb F.K.,
Psaltis D.., Vaughan B. 1997, ApJ, 490, L157
Wijnands R., van der Klis M., 1998, Nature, 394, 344
Wijnands R., Homan J., van der Klis M., Kuulkers E., van
Paradijs J., Lewin W., Lamb F., Psaltis D., Vaughan B.,
1998, ApJ, 493, L87
Tout C.A., Pols O.R., Eggleton P.P., Han Z., 1996, MN-
Wijnands R., M´endez M., van der Klis M., Psaltis D., Ku-
RAS, 281, 257
ulkers E. Lamb F.K. 1998, ApJ, 504, L35
Ushomirsky G., Cutler C., Bildsten L. 2000, MNRAS, 319,
Wijnands R., Strohmayer T., Franco L., 2001, ApJ, 549,
902
L71
Vacca, W.D., Lewin, W.H.G., van Paradijs, J. 1986, MN-
Wijnands R., Groot P.J., Miller J.J., Markwardt C., Lewin
RAS, 220, 339
W.H.G., van der Klis M. 2001, Atel 72
van der Klis M., Swank J.H., Zhang W., Jahoda K., Morgan
E.H., Lewin W.H.G., Vaughan B., van Paradijs J. 1996,
ApJ, 469, L1
Wijnands R., van der Klis M., Homan J., Chakrabarty D.,
Markwardt C., Morgan E., 2003, Nature, 424, 44
Wijnands R., Nowak M., Miller J., Homan J., Wachter S.,
van der Klis M., Wijnands R.A.D., Horne K., Chen W.
Lewin W., 2003, ApJ, 594, 952
1997, ApJ, 481, L97
van der Klis M., 2006, in Compact stellar X-ray sources,
eds Lewin & van der Klis,Cambridge Astrophysics Series
No. 39 (Cambridge University Press), p. 39 - 112
Wijnands R., in't Zand J.J.M., Rupen M., Maccarone T.,
Homan J., Cornelisse R., Fender R., Grindlay J., van
der Klis M., Kuulkers E., Markwardt C.B., Miller-Jones
J.C.A., Wang Q.D. A&A, 449, 1117
Vanderlinde K.W., Levine A.M., Rappaport S.A. 2003,
Wolff M.T., Hertz P., Wood K.S., Ray P.S., Bandyopad-
PASP, 115, 739.
hyay R.M. 2002, ApJ, 575, 384
van Paradijs J., 1995, in X-ray binaries, eds W.H.G. Lewin,
Wood K.S., Norris J.P., Hertz P., Vaughan B.A., Michelson
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
34 Watts et al.
P.F., Mitsuda K., Lewin W.H.G., van Paradijs J., Penninx
W., van der Klis M. 1991, ApJ, 379, 295
Yin H.X., Zhang C.M., Zhao Y.H., Lei Y.J., Qu J.L., Song
L.M., Zhang F., 2007, A&A, 471, 381
Zdziarski A.A., Wen L., Gierli´nski M. 2007, MNRAS, 377,
1006
Zhang W., Smale P., Strohmayer T., Swank J., 1998, ApJ,
500, L171
Zhang W., Strohmayer T., Swank J., 1998, ApJ, 500, L167
Zurita C., Casares J., Shabaz T., Charles P.A., Hynes R.I.,
Shugarov S., Goransky V., Pavlenko E.P., Kuznetsova Y.
2000, MNRAS, 316, 137
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
Gravitational waves from accreting neutron stars
35
Table 1. Fluxes, distances and outburst properties for transients
Source
νs
(Hz)
Distance
(kpc)
Long-term flux Fav
(×10−8 ergs cm−2 s−1)
Outburst duration
(days)
Outburst flux Fob
(×10−8 ergs cm−2 s−1)
Accreting millisecond pulsars
IGR J00291+5934
Aql X-1
SAX J1748.9-2021
XTE J1751-305
SAX J1808.4-3658
HETE J1900.1-2455
XTE J1814-338
XTE J1807-294
XTE J0929-314
SWIFT J1756.9-2508
598.892 ± 2e-08
550.274 ± 9e-04
442.361 ± 5e-08
435.318 ± 4e-08
400.975 ± 6e-09
377.296 ± 5e-09
314.357 ± 1e-09
190.624 ± 8e-08
185.105 ± 9e-09
182.066 ± 7e-08
5 ± 1
4.55 ± 1.35
8.1 ± 1.3
9 ± 3a
3.5 ± 0.1
4.7 ± 0.6
6.7 ± 2.9
8.35 ± 3.65a
7.8 ± 4.2a
8 ± 4a
Burst oscillation sources
XTE J1739-285
4U 1608-522
SAX J1750.8-2900
GRS 1741.9-2853
4U 1636-536
X 1658-298
A 1744-361
KS 1731-260
4U 0614+09
4U 1728-34
4U 1702-429
MXB 1730-335
IGR J17191-2821
4U 1916-053
XB 1254-690
EXO 0748-676
kHz QPO sources
Cyg X-2
GX 340+0
4U 1735-44
GX 5-1
4U 1820-30
Sco X-1
GX 17+2
XTE J2123-058
GX 349+2
1122b
620
601
589
581
567
530b
524
415b
363
329
306b
294b
270b
95b
45
351 ± 34
343 ± 92
316 ± 32
288 ± 69
285 ± 65
272 ± 40
272 ± 50
272 ± 50
266 ± 13
7.3 ± 3.3a
4.1 ± 0.4
6.79 ± 0.14
7.2 ± 2.8
6 ± 0.5
12 ± 3
6 ± 3a
7.2 ± 1.0
2.5 ± 0.5a
5.2 ± 0.5
5.5 ± 0.2
9.25 ± 2.85
7.5 ± 3.5a
8.0 ± 2.2
13 ± 3
7.4 ± 0.9
10.55 ± 4.45
9.15 ± 5.15a
8.5 ± 1.3
7.85 ± 3.85a
7.4 ± 0.6
2.8 ± 0.3
11.4 ± 2
9.6 ± 1.3
4 ± 0.4
1.8e-03
1.2e-01
9.2e-03
2.0e-03
8.6e-03
1.8e-02
1.3e-03
1.4e-03
2.7e-03
1.4e-04
5.9e-03
2.5e-01
3.5e-03
1e-02c
0.47
8.0e-02
5.3e-03
2.2e-01
1.2e-01
0.23
0.13
6.4e-02
6.6e-04
6.9e-02
9.0e-02
4.6e-02
1.1
2.8
8.4e-01
5.3
1.4
39
1.8
1.1e-03
2.2
13
45
61
10
20
730
50
50
60
13
196
100
108
30c
· · ·
872
97
1886
· · ·
· · ·
· · ·
24
11
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
53
· · ·
1.6e-01
1.3
4.0e-01
2.9e-01
3.5e-01
9.0e-02
6.9e-02
7.2e-02
1.0e-01
4.0e-02
8.2e-02
2.0
0.12
1e-01c
· · ·
0.77
0.2
0.49
· · ·
· · ·
· · ·
0.59
0.26
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
· · ·
8.7e-02
· · ·
aDistance poorly constrained: upper and/or lower limit set arbitrarily.
bBurst oscillation frequency requires confirmation: tentative detection, or only seen in one burst from this source.
cNo reliable measurement of flux possible for this Galactic Centre source due to field crowding. The values given here are indicative
estimates only.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
3
6
W
a
t
t
s
e
t
a
l
.
Table 2. Parameter space and template requirements for the accreting millisecond pulsars.
Source
νs ± ∆νs
(Hz)
Nν
Porb ± ∆Porb
NPorb
ax sin i ± ∆ax sin i
Nap
(hours)
(lt-s)
∆Tasc
(days)
NTasc
log10(Ntemp)
Fstat
IGR J00291+5934
598.89213053
Aql X-1
± 2e-08
550.2745
± 9e-04
SAX J1748.9-2021
442.36108118
XTE J1751-305
± 5e-08
435.31799357
± 4e-08
SAX J1808.4-3658
400.975210221
± 6e-09
HETE J1900.1-2455
377.296171971
XTE J1814-338
XTE J1807-294
± 5e-09
314.35610879
± 1e-08
190.62350694
± 8e-08
XTE J0929-314
185.105254297
± 9e-09
SWIFT J1756.9-2508
182.065804253
± 7.2e-08
18
(12)
2.4566922
± 1.7e-06
23
(15)
6.4993e-02
± 2e-06
7.6e+05
(5.1e+05)
18.9479
± 2.0e-04
1557
(1038)
43
(29)
34
(23)
6
(4)
5
(4)
9
(7)
68
(46)
9
(6)
61
(41)
8.76525
± 3e-05
139
(93)
0.7070394
± 6e-07
2.013654711
± 4e-09
1.3875717
± 1.4e-06
4.274645250
± 5.6e-08
0.6678935
± 1e-07
0.7263183
± 8e-07
0.911696
± 2.3e-05
11
(7)
1
(1)
11
(7)
1
(1)
1
(1)
4
(3)
62
(42)
2.5
± 0.5
0.38760
± 4e-05
1.0125e-02
± 5e-06
6.028132e-02
± 2.4e-07
1.841e-02
± 1e-05
0.390633
± 9e-06
4.819e-03
± 4e-06
6.290e-03
± 9e-06
5.942e-03
± 2.7e-05
1
(1)
9692
(6462)
1
(1)
1
(1)
1
(1)
1
(1)
1
(1)
1
(1)
1
(1)
1
(1)
4e-07
3e-03
4e-06
4e-7
1.0e-06
7e-06
9e-07
6e-06
1e-05
6e-05
1
(1)
1049
(700)
1
(1)
1
(1)
1
(1)
1
(1)
1
(1)
1
(1)
1
(1)
1
(1)
2.6
(2.2)
15.6
(14.9)
3.8
(3.4)
2.5
(2.2)
0.8
(0.6)
1.7
(1.4)
1.0
(0.8)
1.8
(1.7)
1.5
(1.2)
3.6
(3.2)
15.9
(14.7)
26.6
(26.0)
17.1
(15.9)
15.9
(14.7)
13.2
(13.2)
14.7
(13.2)
13.2
(13.2)
14.7
(14.7)
14.7
(13.2)
17.1
(15.9)
Note. -- Parameter ranges searched for the pulsars, and number of templates resulting for the mountain (r-mode) scenario assuming a coherent fold with
Tobs = 2 years. For all parameters apart from Tasc we give the absolute value as well as the range searched since both quantities affect the number of templates.
Note that ∆Tasc is the uncertainty on the measurement, so the searched range is twice this value.
c(cid:13)
0
0
0
0
R
A
S
,
M
N
R
A
S
0
0
0
,
0
0
0
--
0
0
0
Table 3. Parameter space and template requirements for the burst oscillation sources (details as for Table 2).
Source
νs ± ∆νs
Nν
Porb ± ∆Porb
NPorb
ax sin i ± ∆ax sin i
Nap
∆Tasc
(days)
NTasc
log10(Ntemp)
Fstat
c(cid:13)
0
0
0
0
R
A
S
,
M
N
R
A
S
0
0
0
,
0
0
0
--
0
0
0
XTE J1739-285
4U 1608-522
SAX J1750.8-2900
GRS 1741.9-2853
4U 1636-536
X 1658-298
A 1744-361
KS 1731-260
4U 0614+09
4U 1728-34
4U 1702-429
MXB 1730-335
IGR J17191-2821
4U 1916-053
XB 1254-690
EXO 0748-676
(Hz)
1122
± 5
620
± 5
601
± 5
589
± 5
581
± 5
567
± 5
530
± 5
524
± 5
415
± 5
363
± 5
329
± 5
306
± 5
294
± 5
270
± 5
95
± 5
45
± 5
(hours)
120.1
± 119.9
67.5
± 57.5
120.1
± 119.9
120.1
± 119.9
4.2e+09
(2.8e+09)
4.2e+09
(2.8e+09)
4.2e+09
(2.8e+09)
4.2e+09
(2.8e+09)
4.2e+09
(2.8e+09)
3.7931263
± 3.8e-06
4.2e+09
(2.8e+09)
7.11610979
± 8e-08
4.2e+09
(2.8e+09)
4.2e+09
(2.8e+09)
4.2e+09
(2.8e+09)
4.2e+09
(2.8e+09)
4.2e+09
(2.8e+09)
4.2e+09
(2.8e+09)
4.2e+09
(2.8e+09)
1.62
± 0.37
121
± 119
0.29
± 0.04
5.1
± 4.9
120.1
± 119.9
120.1
± 119.9
120.1
± 119.9
4.2e+09
(2.8e+09)
0.83351411
± 2.5e-07
4.2e+09
(2.8e+09)
3.9334
± 2e-04
4.2e+09
(2.8e+09)
3.8241072
± 2.4e-06
2.2e+08
(1.5e+08)
1.0e+08
(6.7e+07)
1.2e+08
(7.8e+07)
1.2e+08
(7.7e+07)
238
(159)
3
(2)
1.9e+07
(1.3e+07)
1.0e+08
(6.7e+07)
3.0e+06
(2.0e+06)
1.5e+07
(9.7e+06)
6.5e+07
(4.3e+07)
6.0e+07
(4.0e+07)
5.8e+07
(3.9e+07)
3
(2)
2606
(1737)
9
(6)
(lt-s)
11.3
± 11.3
6.17
± 6.17
11.3
± 11.3
11.3
± 11.3
1.1
± 0.9
0.125
± 0.075
11.3
± 11.3
7e-03
± 7e-03
0.1
± 0.1
11.3
± 11.3
11.3
± 11.3
11.3
± 11.3
1.45e-02
± 1.05e-02
0.98
± 0.12
0.475
± 0.365
4.1e+05
(2.8e+05)
1.3e+05
(8.3e+04)
2.2e+05
(1.5e+05)
2.2e+05
(1.5e+05)
2.5
1.4
2.5
2.5
0.735
± 8.3e-02
1585
(1057)
2e-3
1.3e+06
(8.7e+05)
3.9e+05
(2.6+05)
7.0e+05
(4.7e+05)
6.8e+05
(4.6e+05)
1112
(742 )
41
(27)
6817
(4545)
1.7e+04
(1.1e+04)
9.3e-05
1302
(868)
3.4e-02
1.9e+05
(1.3e+05)
2.5
6.1e+05
(4.1e+05)
96
(64)
1200
(800)
1.2e+05
(8.2e+04)
1.1e+05
(7.6e+04)
1.1e+05
(7.3e+04)
94
(63)
390
(260)
593
(395)
6.1e-03
0.1
2.5
2.5
2.5
1.4e-04
3e-03
1.3e-03
300
(200)
3770
(2513)
3.8e+05
(2.6e+05)
3.6e+05
(2.4e+05)
3.4e+05
(2.3e+05)
4
(3)
365
(243)
40
(27)
34.4
(33.7)
29.1
(28.4)
33.6
(32.9)
33.6
(32.9)
17.8
(17.1)
15.4
(14.7)
23.5
(22.8)
31.3
(30.6)
20.2
(19.5)
25.5
(24.8)
32.9
(32.2)
32.8
(32.1)
32.7
(32.0)
12.2
(11.5)
17.7
(17.0)
14.5
(13.8)
36.0
(36.0)
33.7
(33.2)
36.0
(35.5)
36.0
(35.5)
27.9
(27.3)
26.0
(26.0)
30.7
(30.7 )
34.6
(34.6)
29.0
(28.4)
31.7
(31.7)
35.5
(35.1)
35.5
(35.1)
35.5
(35.1)
24.0
(23.2)
27.9
(27.3)
25.3
(25.3)
G
r
a
v
i
t
a
t
i
o
n
a
l
w
a
v
e
s
f
r
o
m
a
c
c
r
e
t
i
n
g
n
e
u
t
r
o
n
s
t
a
r
s
3
7
Table 4. Parameter space and template requirements for the kHz QPO sources (details as for Table 2).
Source
νs ± ∆νs
(Hz)
Nν
Porb ± ∆Porb
(hours)
NPorb
ax sin i ± ∆ax sin i
Nap
NTasc
log10(Ntemp)
Fstat
∆Tasc
(days)
3.0e-02
3
8
W
a
t
t
s
e
t
a
l
.
Cyg X-2
GX 340+0
4U 1735-44
GX 5-1
4U 1820-30
Sco X-1
GX 17+2
XTE J2123-058
GX 349+2
351
± 34
343
± 92
316
± 32
288
± 69
285
± 65
272
± 40
272
± 50
272
± 50
266
± 13
2.8e+10
(1.9e+10)
236.26560000
± 8.3e-08
1
(1)
7.6e+10
(5.1e+10)
120.1
± 119.9
8.4e+07
(5.6e+07)
2.7e+10
(1.8e+10)
4.6520042
± 7.7e-06
177
(118)
5.7e+10
(3.8e+10)
120.1
± 119.9
6.9e+07
(4.6e+07)
5.4e+10
(3.6e+10)
0.182781083
± 2.8e-08
3.3e+10
(2.2e+10)
18.89551
± 2e-05
9
(6)
58
(39)
4.2e+10
(2.8e+10)
120.1
± 119.9
6.2e+07
(4.2e+07)
4.2e+10
(2.8e+10)
5.956833
± 5.6e-05
1657
(1105)
1.1e+10
(7.2e+09)
22.5
± 0.1
4.0e+05
(2.7e+05)
c(cid:13)
0
0
0
0
R
A
S
,
M
N
R
A
S
0
0
0
,
0
0
0
--
0
0
0
(lt-s)
13.5
± 1.6
11.3
± 11.3
0.693
± 0.693
11.3
± 11.3
1.35e-02
± 6.5e-03
1.44
± 0.18
11.3
± 11.3
1.6
± 0.3
3.5
± 3.5
2.0e+04
(1.4e+04)
1.6e+05
(1.1e+05)
3228
(2152)
2.5
5.0e+05
(3.3e+05)
7825
(5217)
3.0e-3
761
(508)
1.3e+05
(8.7e+04)
2.5
4.1e+05
(2.7e+05)
74
(50)
1827
(1218)
2.2e-04
3.0e-03
28
(19)
350
(234)
1.2e+05
(7.9e+04)
2.5
3.7e+05
(2.5e+05)
3210
(2140)
1.0e-03
422
(281)
3.2e+04
(2.1e+04)
0.47
1.0e+05
(6.6e+04)
18.0
(17.4)
34.5
(33.8)
19.1
(18.4)
34.1
(33.4)
14.5
(13.8)
17.6
(16.9)
33.8
(33.1)
19.5
(18.8)
24.8
(24.1)
27.9
(27.3)
36.0
(36.0)
28.4
(27.9)
36.0
(35.5)
26.0
(25.3)
27.9
(27.3)
36.0
(35.5)
28.4
(28.4)
31.7
(31.2)
Gravitational waves from accreting neutron stars
39
Table 5. Computational cost effects for future performance scenarios.
Source
ν
(Hz)
× 50 improvement
Nstacks
Tobs (days)
× 100 improvement
Nstacks
Tobs (days)
Tdecoh
(years)
Accreting millisecond pulsars
599
550
442
435
401
377
314
191
185
182
IGR J00291+5934
Aql X-1
SAX J1748.9-2021
XTE J1751-305
SAX J1808.4-3658
HETE J1900.1-2455
XTE J1814-338
XTE J1807-294
XTE J0929-314
SWIFT J1756.9-2508
Burst oscillation sources
1 (1)
1 (1)
1 (1)
1 (1)
1 (1)
1 (1)
1 (1)
1 (1)
1 (1)
1 (1)
XTE J1739-285
4U 1608-522
SAX J1750.8-2900
GRS 1741.9-2853
4U 1636-536
X 1658-298
A 1744-361
KS 1731-260
4U 0614+09
4U 1728-34
4U 1702-429
MXB 1730-335
IGR J17191-2821
4U 1916-053
XB 1254-690
EXO 0748-676
kHz QPO sources
Cyg X-2
GX 340+0
4U 1735-44
GX 5-1
4U 1820-30
Sco X-1
GX 17+2
XTE J2123-058
GX 349+2
1122
620
601
589
581
567
530
524
415
363
329
306
294
270
95
45
351
343
316
288
285
272
272
272
266
730 (730)
512 (730)
730 (730)
730 (730)
730 (730)
730 (730)
730 (730)
730 (730)
730 (730)
730 (730)
0.1 (0.1)
3.1 (3.6)
0.1 (0.1)
0.1 (0.1)
98 (167)
611 (730)
1.2 (2.1)
0.6 (0.7)
1 (1)
1 (1)
1 (1)
1 (1)
1 (1)
1 (1)
1 (1)
1 (1)
1 (1)
1 (1)
1.5e+05 (1.5e+05)
7865 (6655)
1.5e+05 (7.3e+04)
1.5e+05 (7.3e+04)
1 (1)
1 (1)
1 (1)
2.4e+04 (2.1e+04)
730 (730)
645 (730)
730 (730)
730 (730)
730 (730)
730 (730)
730 (730)
730 (730)
730 (730)
730 (730)
0.1 (0.1)
3.3 (3.9)
0.1 (0.2)
0.1 (0.2)
123 (211)
730 (730)
1.5 (2.6)
0.6 (0.7)
1.5e+05 (1.5e+05)
8372 (7209)
1.5e+05 (1.5e+05)
1.5e+05 (1.5e+05)
1 (1)
1 (1)
1 (1)
2.4e+04 (2.1e+04)
1 (1)
14.9 (25.6)
1 (1)
18.8 (32.3)
9.6e+04 (8.2e+04)
1.5e+05 (7.3e+04)
7.3e+04 (7.3e+04)
7.3e+04 (7.3e+04)
1 (1)
1 (1)
1 (1)
123 (1)
1.5e+05 (7.3e+04)
1 (1)
1.5e+05 (7.3e+04)
1 (1)
1 (1)
7.3e+04 (7.3e+04)
1 (1)
1908 (1625)
3.6 (4.2)
0.1 (0.2)
0.2 (0.2)
0.2 (0.2)
730 (730)
102 (175)
730 (730)
60 (51)
0.1 (0.2)
35 (61)
0.1 (0.2)
730 (730)
110 (188)
0.2 (0.2)
26 (44)
41 (48)
9.1e+04 (7.7e+04)
7.3e+04 (7.3e+04)
7.3e+04 (7.3e+04)
7.3e+04 (7.3e+04)
1 (1)
1 (1)
1 (1)
1 (1)
1.5e+05 (7.3e+04)
1 (1)
1.5e+05 (7.3e+04)
1 (1)
1 (1)
7.3e+04 (7.3e+04)
1 (1)
1781 (1514)
3.8 (4.5)
0.2 (0.2)
0.2 (0.2)
0.2 (0.2)
730 (730)
128 (220)
730 (730)
40 (73)
0.1 (0.2)
44 (76)
0.1 (0.2)
730 (730)
138 (237)
0.2 (0.2)
33 (56)
44 (51)
1.26
0.17
0.35
0.67
0.82
0.43
1.12
0.86
0.66
2.81
0.48
0.13
0.67
0.37
0.07
0.08
0.61
0.08
0.31
0.11
0.14
0.11
1.39
0.13
0.07
0.17
0.02
0.02
0.03
0.01
0.03
0.02
0.02
0.85
0.05
Note. -- Computational cost effects on data analysis, assuming a 50 or 100 times improvement in computational
power over current capabilities, for the mountain (r-mode) scenario. Where Nstacks = 1, the fold is coherent and
Tobs is the maximum feasible quantity of data that can be folded. Where Nstacks > 1, the fold is semi-coherent and
Tobs is the length of each individual data segment. The assumed spin frequency is given for each source to assist in
cross-referencing this Table with the Figures. We also give the decoherence time Tdecoh, as defined in Section 6.
c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
|
astro-ph/0508406 | 1 | 0508 | 2005-08-18T20:57:59 | Cataclysmic and Close Binaries in Star Clusters. V. Erupting Dwarf Novae, Faint Blue Stars, X-ray sources, and the Classical Nova in the Core of M80 | [
"astro-ph"
] | Large populations of cataclysmic variables (CVs) in globular clusters have long been predicted, but the number of absolutely certain cluster CVs known in globulars is still less than 10. HST and Chandra observers have recently found dozens of cataclysmic variable candidates in several populous globular clusters. Confirmation and characterization of these candidates are extremely difficult, thus identification of unambiguous CVs remains important. We have searched all archival HST images of the dense globular cluster M80 for erupting dwarf novae (DN), and to check the outburst behaviors of two very blue objects first identified a decade ago. Two new erupting dwarf novae were found in 8 searched epochs, making M80 a record holder for erupting DN. The quiescent classical nova in M80 varies by no more than a few tenths of a magnitude on timescales of minutes to years, and a similar faint, blue object varies by a similar amount. Simulations and completeness tests indicate that there are at most 3 erupting DN like SS Cyg and at most 9 U Gem-like DN in M80. Either this very dense cluster contains about an order of magnitude fewer CVs than theory predicts, or most M80 CVs are extremely faint and/or erupt very infrequently like WZ Sge. We have detected a sequence of 54 objects running parallel to the main sequence and several tenths of a magnitude bluewards of it. These blue objects are significantly more centrally concentrated than the main sequence stars, but not as centrally concentrated as the blue stragglers. We suggest that these objects are white dwarf--red dwarf binaries, and that some are the faint CV population of M80. | astro-ph | astro-ph |
Accepted to ApJ
Cataclysmic and Close Binaries in Star Clusters. V. Erupting
Dwarf Novae, Faint Blue Stars, X-ray sources, and the Classical
Nova in the Core of M801
Michael M. Shara
Department of Astrophysics, American Museum of Natural History, 79th St. and Central
Park West, New York, NY, 10024
[email protected]
Sasha Hinkley
Department of Astrophysics, American Museum of Natural History, 79th St. and Central
Park West, New York, NY, 10024
[email protected]
David R. Zurek
Department of Astrophysics, American Museum of Natural History, 79th St. and Central
Park West, New York, NY, 10024
[email protected]
ABSTRACT
Large populations of cataclysmic variables (CVs) in globular clusters have
long been predicted, but the number of absolutely certain cluster CVs known in
globulars is still less than 10. HST and Chandra observers have recently found
dozens of cataclysmic variable candidates in several populous globular clusters.
Confirmation and characterization of these candidates are extremely difficult,
thus identification of unambiguous CVs remains important. We have searched
1Based on observations with the NASA/ESA Hubble Space Telescope, obtained from the Data Archive
at the Space Telescope Science Institute,which is operated by the Association of Universities for Research
in Astronomy, Inc., under NASA contract NAS5-26555. These observations are associated with program
#6460.
-- 2 --
all archival HST images of the dense globular cluster M80 for erupting dwarf
novae (DN), and to check the outburst behaviors of two very blue objects first
identified a decade ago. Two new erupting dwarf novae were found in 8 searched
epochs, making M80 a record holder for erupting DN. The quiescent classical
nova in M80 varies by no more than a few tenths of a magnitude on timescales
of minutes to years, and a similar faint, blue object varies by a similar amount.
Simulations and completeness tests indicate that there are at most 3 erupting DN
like SS Cyg and at most 9 U Gem-like DN in M80. Either this very dense cluster
contains about an order of magnitude fewer CVs than theory predicts, or most
M80 CVs are extremely faint and/or erupt very infrequently like WZ Sge. We
have detected a sequence of 54 objects running parallel to the main sequence and
several tenths of a magnitude bluewards of it. These blue objects are significantly
more centrally concentrated than the main sequence stars, but not as centrally
concentrated as the blue stragglers. We suggest that these objects are white
dwarf -- red dwarf binaries, and that some are the faint CV population of M80.
Subject headings: Stars: Cataclysmic Variables -- dwarf novae, galaxies: individ-
ual (LMC)
1.
Introduction
The discovery of unexpectedly rich populations of luminous X-ray sources in Galactic
globular clusters (Clark 1975; Katz 1975) was a sharp challenge to theorists. Efficient mech-
anisms for generating large numbers of mass-transferring binaries, dominated by a compact
primary, were required to explain the observations. Tidal capture scenarios involving two
stars (Fabian, Pringle, & Rees 1975) and/or exchange reactions involving three stars (Hills
1976) were rapidly developed. These processes are now believed to be the sources of the
neutron star-dominated X-ray binaries in globular clusters. Strong observational evidence
in favor of these scenarios is the remarkable correlation between stellar encounter rate and
number of X-ray sources in globular cluster cores (Pooley et al. 2003).
These two and three-body mechanisms also predict large populations (Di Stefano & Rap-
paport 1994) of accreting white dwarf-main sequence star binaries -- the cataclysmic variable
(CV) stars. Unfortunately only one globular cluster dwarf nova (DN) is easily resolved and
studied from the ground: V101 in M5 (Margon, Downes & Gunn 1981), (Shara, Potter, &
Moffat 1990), (Neill, Shara, Caulet, & Buckley 2002). The large amplitude variations of
many CVs (Warner 1995) suggest that Hubble Space Telescope (HST) observations of glob-
ular clusters would easily locate dozens in the core of each cluster. With expected apparent
-- 3 --
magnitudes in the range 17-23, prototypical CVs like SS Cyg and U Gem should be trivial
to find. It would clearly be of enormous benefit to CV science if dozens of these objects,
all at the same distance and with the same parent metallicity, could be located. Accurate
luminosity functions, and bias-free period distributions and eruption frequencies could be
derived to confront theoretical models. Tidal capture CVs could be contrasted with those
produced by ordinary binary evolution. Systematic variations in outburst properties and
orbital distributions might be uncovered.
Remarkably, only a handful of absolutely certain CVs in globular clusters are known
today. These include a few dwarf novae seen in eruption (e.g. (Paresce & De Marchi 1994),
(Charles, Clarkson, & van Zyl 2002), (Anderson, Cool, & King 2003)), one classical nova
(Auwers 1886), (Shara & Drissen 1995), and a few spectrographically confirmed, very blue
faint stars (e.g. (Grindlay et al. 1995), (Deutsch, Margon, Anderson, & Downes 1999; Knigge
et al. 2003)). Recently, dozens of CV candidates have been identified in 47 Tuc (Grindlay,
Heinke, Edmonds, & Murray 2001), (Knigge, Zurek, Shara, & Long 2002), M80 (Heinke et al.
2003) and in NGC 6397 (Grindlay et al. 2001) through deep Chandra X-ray Observatory and
HST observations. Unfortunately, large amounts of HST and Chandra time are essential for
follow-up studies, so unambiguous characterization of all candidates may take many years.
Thus any new, unambiguous CVs that can be found in archival data are well worth searching
for.
About half of all known field CVs (Downes et al. 2001) are dwarf novae (DN). Most
known DN reveal themselves through 2-5 magnitude outbursts every few weeks to months
(Warner 1995); similar DN in clusters should be identifiable in multi-epoch images. Since
large archival HST datasets of globulars are available we have been systematically looking
for erupting dwarf novae in the cores of all such clusters. Here we report the results of an
extensive and successful search of M80.
M80 is one of the densest globular clusters in the Milky Way, with a spectacular blue
straggler sequence (Ferraro et al. 1999), a quiescent old nova (Shara & Drissen 1995), and
19 X-ray sources (Heinke et al. 2003). A detailed comparison of M80 with 47 Tuc (Heinke et
al. 2003) suggests that the encounter rate in M80 is about a factor of two lower; roughly 50
CVs should exist in M80 as of order 100 are predicted in 47 Tuc (Di Stefano & Rappaport
1994).
In section 2 we present the observational database, search strategy, and the photometry
of variables. The erupting DN, non-eruptive CV candidates and their light curves are shown
in section 3. We discuss whether the candidates could be other kinds of variables in section
4. In section 5 we report simulations to test our completeness, and to place limits on the
numbers of DN of various luminosities in M80. We report on a remarkable sequence of very
-- 4 --
blue, centrally located objects in section 6, and our attempts to match these with the M80
X-ray sources in section 7. We briefly summarize our results in section 7.
2. Observations
The Hubble Space Telescope has imaged M80 during 8 separate epochs from 1994 to
2000. All epochs except the last included F336W frames, particularly useful in detecting
blue objects. The pass-band of the F450W frames of the last epoch is close to the F439W
passband of epochs 1 and 2, so that essentially constant limiting magnitude is attained for
our 8 epoch survey (see section 5). The dates of observation, PI of program and program
number, filters used, number of frames and total exposure time in each filter are given in the
observing log which is Table 1.
The individual WFPC2 images were combined using the "Montage2" routine contained
within the stand-alone DAOphot package. These frames were run through DAOphot's
matching program "DAOmaster" to derive the subpixel frame-to-frame shifts and thereby
register all images to the first image of the first epoch. These shifts were passed to Montage2,
which produced a single, sky subtracted, high signal-to-noise image for each epoch and filter.
HST was not perfectly aligned at all epochs in all filters. Figure 1 is an overview of the fields
of view of the 8 epochs, with four of the most interesting objects that we found (section 3)
indicated.
The dwarf-nova candidates (Figure 2) were found by image differencing and by blinking
rapidly through groups of the F336W frames to look for any subtle changes in the appearance
of the stellar field. Dwarf novae should rise from invisibility to easily detectable in the
timeframe defined by the observations. Brightening from previously empty regions of the
sky is easily seen.
In order to detect extremely faint stars, a master list of stars was derived using the
technique of Richer et al. (2004). This technique rejects the brigher pixels from a stack, giving
emphasis to fainter pixels, and thus probing deeper.
In addition, point spread functions
were created for the F336W, F450W, F656N, and F675W (U, B, Hα and R) data and then
provided to the allframe software (Stetson 1994) for multi-image simultaneous PSF fitting
to derive positions and magnitudes for as many stars as possible. The photometry was then
converted to the ST magnitude system as described in Shara & Drissen (1995). The resulting
color-magnitude diagram (CMD) of M80, the deepest ever produced, is shown in Figures 3
and 4. The two new dwarf nova candidates are evident, as are the two very blue stars first
noted by Shara & Drissen (1995). In addition, many more faint blue objects -- about 50 in
-- 5 --
all -- are now visible in this deeper CMD. Some of these very blue objects are also indicated
in the finder chart that is Figure 5, and discussed in sections 6 and 7.
3. Dwarf Novae and the Classical Nova
To be recognized as a candidate, a variable had to be visible on all frames in at least one
of the eight epochs, and to have varied by more than 1.0 magnitudes between its faintest and
brightest state. This selection method revealed two dwarf nova candidates from the entire
dataset (and no other obvious variables). These two strongly varying, very blue candidates
which we label DN1 and DN2 are erupting in epochs 3 and 6, respectively. Their positions
are: for DN1 α(2000) = 16:17:02.2, δ(2000) = -22:58:37.9; and for DN2, α(2000) =16:16:59.8,
δ(2000) = -22:58:18.0. The nightly images of these two objects are shown in Figures 6 and
7, and their photometry is presented in Figures 8 and 9. We defer a discussion of these stars
and their implications until after the next section, where we simulate expected DN images
and light curves.
We show the images and photometry of the likely quiescent old nova (T Sco 1860 A.D.)
in Figures 10 and 11, respectively. Brightness fluctuations do not exceed a few tenths of
a magnitude. The mean apparent F336W magnitude is 21.5, corresponding to an absolute
U-band magnitude of 6.5, well within the range of old nova luminosities (Warner 1995).
Finally, we show in Figures 12 and 13 the images and lightcurve of the faint blue object
noted by Shara & Drissen (1995). Located close to T Sco in the M80 color-magnitude dia-
gram, it too varies by at most a few tenths of a magnitude. Whereas the T Sco lightcurve
shows several points in a given epoch, the faint blue object's lightcurve shows only one
brightness value for each epoch. The object lies rather close to the HST Planetary Cam-
era (PC) chip's edge, and photometry was performed on the individual-epoch, CR-rejected
images to boost the signal-to-noise. While its brightness and color are very suggestive, this
object remains a CV candidate, and not a confirmed CV, until it is eventually shown to
erupt as a DN or found to have a spectrum consistent with cataclysmic classification.
4. Simulations of Known Dwarf Novae
To estimate our detection completeness for erupting dwarf novae to various magnitude
limits we carried out simulations of eruptions for two of the best known and characterized
Galactic dwarf novae: SS Cyg, and U Gem. (These objects have accurately determined par-
allaxes (Harrison et al. 1999) and, using their well tabulated apparent magnitudes (Warner
-- 6 --
1995), the absolute magnitudes are easily calculated. These two dwarf novae were chosen to
represent, respectively, the most luminous and more typical values of absolute magnitudes
seen for erupting and quiescent DN). To simulate SS Cyg, 100 artificial PSFs were placed in
random positions across an M80 HST F336W image of a single epoch. For these 100 PSFs,
brightnesses were determined by picking 100 random points from the SS Cyg American As-
sociation of Variable Star Observers (AAVSO) lightcurve stretching over 1000 days. These
100 artificial dwarf novae, (now mimicking SS Cyg at 100 random places in its lightcurve),
were shifted to the apparent magnitudes they would have at the 10.3 kpc distance of M80
(Brocato et al. 1998). This image was then blinked against an image (at a different epoch)
with these same artificial PSFs corresponding to SS Cyg in quiescence. The same proce-
dure was performed for U Gem, using 100 randomly sampled brightnesses from 1000 days of
the U Gem lightcurve. Typical images of these simulated dwarf novae, each at their maxi-
mum, intermediate and minimum brightness, are shown in Figure 14. Our simulations show
that, by comparing two randomly selected epochs, one detects an SS Cygni-like dwarf nova
25.0 ± 3.3% of the time, while U Gem is detected 6.0 ± 1.7% of the time. The detection
probability is thus PSS = 0.125 per epoch for SS Cyg and PU G = 0.030 for U Gem. In each
independent epoch, we therefore expect that the probability of not seeing a given SS Cyg in
eruption is 1 − 0.125 = 0.875, and the probability of not seeing a given U Gem in eruption is
1 − 0.030 = 0.970. Thus in eight independent epochs (a more pessimistic case than our own
8 epochs, where 5 epochs are precisely at weekly intervals) we might expect to not see SS
Cyg (0.875)8 = 34% of the time, and we expect to miss U Gem-like objects (0.97)8 = 78%
of the time. Equivalently, we expect to see at least 1 − 0.34 = 66% of SS Cyg-like eruptions
and at least 1 − 0.78 = 22% of all U Gem-like eruptions.
5.
Interpretation
5.1. Have We Found Erupting M80 Dwarf Novae?
Comparison of the two erupting DN candidates shown in Figures 6 and 7 with the
M80-DN simulation images shown in Figure 14 demonstrates that the brightness and vari-
ability behaviors of our candidates are consistent with those expected of moderately luminous
erupting dwarf novae in M80. All erupting dwarf novae achieve outburst U band absolute
magnitudes . 5, and the simulations of Figure 15 show that we easily reached fainter than
that limit in all epochs. The very blue outburst colors, outburst brightnesses, presence and
blue colors in quiescence, our completeness of detection of stars at different magnitudes (Fig-
ure 15), and strong concentration to the center of M80 all argue very strongly in favor of
DN1 and DN2 being dwarf novae. However, in the absence of spectra or a second recorded
-- 7 --
outburst, one could always argue that one or both of these are variables of some other type,
perhaps not even associated with M80.
What are possible variables that might mimic M80 DN behavior? Amongst these are:
chance superpositions of background supernovae or classical novae, gamma ray bursts (GRB),
microlensing events or Milky Way variables along the line of sight to M80. Large area,
multi-epoch surveys for faint variables (e.g. Hawkins (1984)) show that the strong central
concentration of our variables to the core of M80 (and the moderately high galactic latitude
of the cluster, 19.5◦) is far too high for them to be field RR Lyrae stars or supernovae.
The brightness and very blue colors of the candidates rule out other types of Galactic
variables. RR Lyrae and flare stars don't match the observed brightness and/or blue colors
of the two DN candidates. The rarity of GRB (about 1 over the entire sky per day) and
microlensing events, and the very blue colors of our variables almost certainly rules out these
possibilities. But while the magnitudes and colors both during outburst and quiescence of our
DN candidates argue strongly for that classification, an absolutely certain characterization
will require challenging but important follow-up observations. These include
1) Spectra near quiescence (to demonstrate the presence of Balmer emission lines).
2) Imagery every day or two for several months with HST to reveal repeated eruptions
separated by weeks to months.
3) Several hours of HST time-resolved UV or optical photometry to reveal the orbital
modulation characteristic of CVs.
5.2. Detections versus Expected Detections
The size of our field of view was 1574 arcsec2, extending out to 1.9 core radii from the
cluster center. The average time between eruptions for 21 well studied Galactic DN is 29
days (Szkody & Mattei 1984). The length and depth of our observing run (5 epochs spaced 1
week apart, and three other random epochs) taken by themselves suggest that most erupting
SS Cyg-like or U Gem-like dwarf nova would have been detected. In fact, our simulations
of SS Cyg and especially of U Gem (in section 4) demonstrate that outburst frequency is
the key parameter in determining whether a given dwarf nova will be detected. SS Cyg is
intrinsically more luminous in outburst than U Gem, but both are straightforward to detect,
in eruption, in our dataset. It is the relative infrequency of U Gem outbursts that makes
it three times less likely (22%) to be detected than SS Cyg-like outbursts. Given these
probabilities of detecting SS Cyg and U Gem-like dwarf nova eruptions in M80, what limits
-- 8 --
can we place on the total populations of similar objects in the cluster? Since we have only
seen 1 eruption of each of the M80 DN we don't know their eruption frequency or maximum
luminosity. Either could be U Gem-like or SS Cyg-like. Our conservative estimate is that
there are . 2/0.66 = 3 SS Cyg-like DN in M80, and . 2/0.22 ∼ 9 U Gem-like DN in the
cluster.
The fact that we detected only two erupting DN, and that there are almost certainly
fewer than ∼ 9 DN in M80, when (the admittedly simple) tidal capture model predicts an
order of magnitude more CVs suggests that
i) SS Cyg and U Gem-like CVs in M80 are ∼ 50/9 ∼ 6× fewer in number than theory
predicts, and/or
ii) most CVs are more like the rarely erupting WZ Sge, and much fainter than the
prototypical DN SS Cyg and U Gem, and/or
iii) globular CVs may be mostly magnetic (Grindlay et al. 1995)and non-eruptive.
What is the true, average inter-outburst period for all dwarf novae? Unfortunately, the
answer is unknown. CV catalogs are dominated by the easy-to-find, frequently outbursting
DN. In addition, the distances to all but a dozen or so Galactic DN are too uncertain to
yield an accurate luminosity distribution. (We note that the canonical literature distance to
the prototypical DN, SS Cyg, was in error by a factor of two until HST parallaxes became
available in 2000). A warning that CV populations may be dominated by extremely faint
and/or infrequently erupting DN is personified in the nearest known DN, WZ Sge. This
object is a mere 43.5 pc away (Thorstensen 2003), (Harrison et al. 2004) with absolute
visual magnitudes at minimum and maximum of 11.8 and 3.9, and 1 -- 2 month-long eruptions
separated by about 25 years. Our chances of detecting a WZ Sge-like outburst are two orders
of magnitude smaller than the likelihood of detecting a U Gem-like eruption (again, because
of the relative outburst frequency). There could easily be 100 WZ Sge stars in M80 -- and
our chances of detecting even one in our dataset would only be of order 20%. We conclude
that: there are ∼ 3 or fewer SS Cyg-like, or 9 or fewer U Gem-like DN in M80. However, 100
WZ Sges could exist in the cluster. If most CVs are similar to WZ Sge (often referred to as
CV "graveyard" objects) then the small number of detected M80 DN is naturally explained.
6. The Faint Blue Objects in M80
The deep U vs. (U − B) CMD of Figure 5 contains a remarkable sequence of 54 objects
running parallel to the main sequence and between 0.3 and 1.4 magnitudes blueward of it.
-- 9 --
We have verified that these faint blue objects are not due to e.g. "hot" pixels by comparing
their images on frames from different epochs with significant spatial offsets. While it is cer-
tainly possible that some of these objects' blue colors are due to photometric errors, at least
a dozen objects with (U − B) < 0.1 are (on visual inspection) very blue and unblended. This
is precisely the part of the CMD that would be occupied by CVs and non-interacting white
dwarf-red dwarf binaries. If these 54 objects are, indeed CVs and/or WD-RD binaries then
they must possess masses larger than main sequence stars of similar magnitude. Equiparti-
tion of energy should then have concentrated these blue objects towards the cluster center.
This, in fact, is an acid test that must be passed if we are to give credence to the suggestion
of the blue M80 stars as a WD-RD binary sequence. Figure 16 strongly suggests that these
blue stars are more centrally concentrated than main sequence stars, at least at r > 8 arcsec
from the cluster center. The K-S test determines that the blue objects and main sequence
stars of Figure 16 do not share the same radial distribution with > 95% confidence. The
blue stragglers and main sequence stars are even more dramatically separated in Figure 16.
The K-S test states that the BS and MS stars are drawn from different radial distributions
with > 99.5% likelihood.
Artificial star tests were performed to determine the the recovery rate for artificial
stars placed randomly on the chip. Figure 17 shows the fraction of stars recovered in 8
different annuli centered on the center of the cluster, for several different brightesses. These
simulations show that we approach maximum completeness (∼ 80%) for the fainter blue
stars with U > 22 only when r > 8 arcsec from the core of M80. Thus the apparent weak
central concentration of the blue stars in the inner 8 arcsec is probably due to significant
incompleteness of the sample of these very faint objects in the core of the cluster, rather than
their physical absence.
The central concentration of blue stragglers in globular clusters is ubiquitous and well
established for many clusters e.g. (Paresce et al. 1991; Ferraro et al. 1999), and indicative
of objects with measured masses 2-3 times the turnoff mass (Shara, Saffer, & Livio 1997;
Saffer et al. 2002). We conclude from Figure 16 that the blue sequence stars of Figure 13
are intermediate in mass between the main sequence and blue straggler stars -- exactly as
one expects for WD-RD binaries.
7. X-ray sources and Faint Blue Objects
We have compared the positions of our four objects of interest (DN1, DN2, T Sco and
the Shara-Drissen object) as well as the 54 faint, blue objects with the 19 X-ray sources noted
by Heinke et al. (2003). The Right Ascension and Declinations of our full sample of 58 CVs
-- 10 --
and CV candidates were offset in an iterative manner until we found the optimal minimum
separation between the 19 X-ray sources and the CV candidates. The best overall match
(in the least-squares sense) occurs with offsets of ∆α ≃ −0.5′′ and ∆δ ≃ +4.7′′ applied to
the CV candidates, with or without the two significant outliers CX10 and CX19 removed.
The resulting source distributions are shown in Figure 18. The 19 X-ray source positions
together with the adjusted positions of the closest faint blue stars are listed in Table 2.
The nearly 5" offset between the 19 X-ray sources and the 58 CVs and CV candidates
is much larger than the uncertainties in HST and Chandra pointings. We also find no
correlation between the 19 X-ray sources' luminosities and the F336W brightnesses of the
19 "matched" faint blue stars in Table 2. We interpret this to mean that, while the two
populations are centrally concentrated, many (and probably most or all) of the apparent
matches of Table 2 are spurious. Deep HST far-UV observations of M80 are currently being
scheduled, and these will be important for providing optical-UV counterparts to this cluster's
X-ray sources.
Finally, we have also checked the Hα luminosities of the 54 faint blue sources. An R
versus Hα − R color-magnitude diagram is shown in Figure 19. Of the full sample of 54
blue objects, 51 had measurable Hα fluxes. None of these faint blue sources appear to have
an Hα brightness more significant than that of the overall cluster population. If these are
degenerate star -- red dwarf binaries then their mass transfer rates are very low or zero.
8. Conclusions
We have observed M80 during 8 separate epochs with sufficient resolution and sensitivity
to detect any erupting dwarf novae within the inner 1.9 core radii of that cluster. Two very
strong candidates, with outburst colors, brightnesses, central concentration, and presence in
quiescence, all consistent with DN classification, were detected. The old nova candidate T
Sco (nova 1860) and another very blue, faint star (Shara & Drissen 1995) are remarkably
constant in brightness. Simulations and completeness tests indicate that ≤ 3 regularly
outbursting dwarf novae similar to SS Cyg, and . 9 U Gem-like dwarf novae exist today in
M80. The presence of several dozen very faint blue, centrally concentrated stars in the core
of M80 hints at a large red dwarf -- degenerate star binary population; some of these may be
a very rarely-erupting CV population.
We acknowledge with thanks the variable star observations from the AAVSO Interna-
tional Database contributed by observers worldwide and used in this research.
-- 11 --
Support for program #6460 was provided by NASA through a grant from the Space
Telescope Science Institute, which is operated by the Association of Universities for Research
in Astronomy, Inc., under NASA contract NAS 5-26555."
REFERENCES
Anderson, J., Cool, A. M., & King, I. R. 2003, ApJ, 597, L137
Auwers, A. 1886, Astronomische Nachrichten, 114, 47
Bailey, J. 1980, MNRAS, 190, 119
Brocato, E., Castellani, V., Scotti, G. A., Saviane, I., Piotto, G., & Ferraro, F. R. 1998,
A&A, 335, 929
Bruch, A. & Engel, A. 1994, A&AS, 104, 79
Charles, P. A., Clarkson, W. I., & van Zyl, L. 2002, New Astronomy, 7, 21
Clark, G. W. 1975, ApJ, 199, L143
Deutsch, E. W., Margon, B., Anderson, S. F., & Downes, R. A. 1999, AJ, 118, 2888
Di Stefano, R. & Rappaport, S. 1994, ApJ, 423, 274
Downes, R. A., Webbink, R. F., Shara, M. M., Ritter, H., Kolb, U., & Duerbeck, H. W.
2001, PASP, 113, 764
Fabian, A. C., Pringle, J. E., & Rees, M. J. 1975, MNRAS, 172, 15P
Ferraro, F. R., Paltrinieri, B., Rood, R. T., & Dorman, B. 1999, ApJ, 522, 983
Grindlay, J. E., Cool, A. M., Callanan, P. J., Bailyn, C. D., Cohn, H. N., & Lugger, P. M.
1995, ApJ, 455, L47
Grindlay, J. E., Heinke, C., Edmonds, P. D., & Murray, S. S. 2001, Science, 292, 2290
Grindlay, J. E., Heinke, C. O., Edmonds, P. D., Murray, S. S., & Cool, A. M. 2001, ApJ,
563, L53
Harrison, T. E., McNamara, B. J., Szkody, P., McArthur, B. E., Benedict, G. F., Klemola,
A. R., & Gilliland, R. L. 1999, ApJ, 515, L93
-- 12 --
Harrison, T. E., Johnson, J. J., McArthur, B. E., Benedict, G. F., Szkody, P., Howell, S. B.,
& Gelino, D. M. 2004, AJ, 127, 460
Hawkins, M. R. S. 1984, MNRAS, 206, 433
Heinke, C. O., Grindlay, J. E., Edmonds, P. D., Lloyd, D. A., Murray, S. S., Cohn, H. N., &
Lugger, P. M. 2003, ApJ, 598, 516
Hills, J. G. 1976, MNRAS, 175, 1P
Hills, J. G. & Day, C. A. 1976, Astrophys. Lett., 17, 87
Katz, J. I. 1975, Nature, 253, 698
Knigge, C., Zurek, D. R., Shara, M. M., & Long, K. S. 2002, ApJ, 579, 752
Knigge, C., Zurek, D. R., Shara, M. M., Long, K. S., & Gilliland, R. L. 2003, ApJ, 599, 1320
Margon, B., Downes, R.A., & Gunn, J. E. 1981, ApJ, 247, L89
Neill, J. D., Shara, M. M., Caulet, A., & Buckley, D. A. H. 2002, AJ, 123, 3298
Paresce, F., Meylan, G., Shara, M., Baxter, D., & Greenfield, P. 1991, Nature, 352, 297
Paresce, F. & de Marchi, G. 1994, ApJ, 427, L33
Pogson, N. 1860, MNRAS, 21, 32
Pooley, D., et al. 2003, ApJ, 591, L131
Pryor, C. & Meylan, G. 1993, ASP Conf. Ser. 50: Structure and Dynamics of Globular
Clusters, 357
Richer, H. B., et al. 2004, AJ, 127, 2904
Saffer, R. A., Sepinski, J. F., Demarchi, G., Livio, M., Paresce, F., Shara, M. M., & Zurek,
D. 2002, ASP Conf. Ser. 263: Stellar Collisions, Mergers and their Consequences, 157
Shara, M. M., Potter, M., & Moffat, A. F. J. 1990, AJ, 100, 540
Shara, M. M. & Drissen, L. 1995, ApJ, 448, 203
Shara, M. M., Saffer, R. A., & Livio, M. 1997, ApJ, 489, L59
Shara, M. M., Zurek, D. R., Baltz, E. A., Lauer, T. R., & Silk, J. 2004, ApJ, 605, L117
-- 13 --
Stetson, P. B. 1994, PASP, 106, 250
Stetson, P. B. 2000, PASP, 112, 925
Szkody, P. & Mattei, J. A. 1984, PASP, 96, 988
Thorstensen, J. R. 2003, AJ, 126, 3017
Tody, D. 1986, Proc. SPIE, 627, 733
Verbunt, F. & Hut, P. 1987, IAU Symp. 125: The Origin and Evolution of Neutron Stars,
125, 187
Warner, B. 1995, Cambridge Astrophysics Series, Cambridge, New York: Cambridge Uni-
versity Press, -- c1995,
This preprint was prepared with the AAS LATEX macros v5.2.
-- 14 --
Epoch 2
Epochs 1,3-8
T Sco
Blue
Object
DN2
DN1
N
E
Fig. 1. -- A mosaic of HST images of the field of M80, indicating the orientations of our 8
epochs of observation.
-- 15 --
Dwarf Nova 1 Dwarf Nova 2
T Sco
Blue Object
E
N
Fig. 2. -- A closer view of each of the four objects of interest. Each image is 2.37 arcseconds
on a side. The lower resolution in the image of Dwarf Nova 2 reflects the fact that this
object was only imaged on the HST WF4 chip, while all the others were imaged on the
higher resolution Planetary Camera CCD.
-- 16 --
Fig. 3. -- CMD for M80. In the plot, the four objects of Figure 1 are marked. Since there
was only R imaging for one epoch (epoch 3), obtaining both eruption and quiescent colors
was impossible. The range of U magnitudes for the two dwarf novae are shown to give a
sense of the overall range during outburst. In addition, since the Blue Object was essentially
invisible in the R-band, an R value near the plate limit was assumed, hence the upper limit
arrow indicated.
-- 17 --
Fig. 4. -- The U versus (U −B) CMD for M80. The 54 triangles that lie blueward of the main
sequence are newly detected white dwarf-red dwarf binary and/or CV candidates. The four
objects discussed in this paper are also marked. The faint blue object of Shara & Drissen
(1995) is marked with a circle, while T Sco (nova 1860) is starred. Note that the location of
T Sco is nearly coincident with the first dwarf nova (DN1) in its quiescent state. As before,
the brightness ranges of the two dwarf novae DN1 and DN2 are marked.
-- 18 --
F336W
F450W
E
N
Fig. 5. -- Two subregions of the F336W and F450W PC images showing a handful of those
objects significantly more blue than the main sequence. North and East are as indicated.
Each image is 8.38 arcseconds on a side.
-- 19 --
M80 Dwarf Nova 1: All images on PC chip
F675W
F336W F439W F450W F555W F656N
E
N
epoch 1
epoch 2
epoch 3
epoch 4
epoch 5
epoch 6
epoch 7
epoch 8
Fig. 6. -- Thumbnail images for the first Dwarf Nova (DN1) in M80. It can clearly be seen
in eruption in epoch 3. Each thumbnail is 1.04 arcseconds on a side, and North and East
are indicated at the upper right.
-- 20 --
M80 Dwarf Nova 2: All images on WF4 chip
F336W F439W F450W F555W F656N
F675W
E
N
epoch 1
epoch 2
epoch 3
epoch 4
epoch 5
epoch 6
epoch 7
epoch 8
Fig. 7. -- Thumbnail images for the second Dwarf Nova (DN2) in M80. It is in eruption in
epoch 6 and just off the PC chip in epoch 2. Each thumbnail is 2.1 arcseconds on a side,
and North and East are indicated at the upper right.
-- 21 --
Fig. 8. -- The lightcurve for the first Dwarf Nova (DN1) in F336W. The object is in eruption
in epoch 3.
-- 22 --
Fig. 9. -- The lightcurve for the second Dwarf Nova (DN2) in F336W. The object is in
eruption in epoch 6, and just outside the field-of-view in epoch 2.
-- 23 --
T Sco in M80. All images on PC chip.
F336W F439W F450W F555W F656N
F675W
E
N
epoch 1
epoch 2
epoch 3
epoch 4
epoch 5
epoch 6
epoch 7
epoch 8
Fig. 10. -- Thumbnail images of the classical nova T Sco in M80. These can be compared
to Shara and Drissen (1995). Each thumbnail is 1.04 arcseconds on a side.
-- 24 --
Fig. 11. -- F336W Photometry for the old nova T Sco (nova 1860). The object is remarkably
constant in brightness.
-- 25 --
U-bright object. All images on the PC chip.
F336W F439W F450W F555W F656N
F675W
E
N
epoch 1
epoch 2
epoch 3
epoch 4
epoch 5
epoch 6
epoch 7
epoch 8
Fig. 12. -- Images of the extremely Blue object of Shara & Drissen (1995). Note that for
epochs 3-7, the object is near the edge of the PC chip. The object was just off the chip in
epoch 2. Each thumbnail is 1.54 arcseconds on a side.
-- 26 --
Fig. 13. -- An F336W lightcurve for the faint blue object. The large error bars in epochs
3-7 are due to the relatively low S/N caused by the object's proximity to the chip edge.
-- 27 --
F336W/PC
Quiescence Intermediate
Eruption
SS Cyg
U Gem
Fig. 14. -- Examples of artificial Dwarf Novae in M80. Typical images of the two prototypical
Dwarf Novae SS Cyg and U Gem, placed in M80 F336W PC images, are shown in three
different phases, from quiescence to eruption.
-- 28 --
Fig. 15. -- Completeness curves for a single-epoch coadded F336W image of M80. The figure
assumes an M80 distance modulus of (m − M) = 15.06. Several hundred artificial stars were
added to the F336W image in each magnitude bin. Only those stars recovered at the same
position AND within 0.5 mag of the artificial star were considered recovered. Crowding
accounts for the ∼ 20% incompleteness for magnitudes between 0 and ∼ 6.
-- 29 --
Fig. 16. -- The cumulative radial distribution of three different groups of stars as indicated.
The open triangles correspond to the faint blue stars running nearly parallel to the main
sequence indicated in Figure 5. Their radial distribution differs from that of the main
sequence stars (from the K-S test) with > 95% likelihood
-- 30 --
Fig. 17. -- A simulation showing the recovery fraction of a sample of artificial stars randomly
placed on an F336W image as a function of distance from the cluster center. We are sig-
nificantly incomplete at r < 8 arcsec for stars fainter than ≃ 22.0 magnitude. This further
strengthens the suggestion (From Figure 16) that the faint blue stars in M80 are centrally
concentrated and hence good binary candidates.
-- 31 --
Fig. 18. -- A plot showing the positions (circles) of the 19 X-ray bright sources noted by
Heinke et al. (2003). The positions (triangles) of the closest faint blue stars are also shown
(see text). The locations of Dwarf Nova 1 and 2 are marked.
-- 32 --
Fig. 19. -- An R versus Hα − R color-magnitude diagram with 51 of the faint blue objects
(see Fig. 5) marked with triangles (see text). None of these faint blue objects appear to
have an Hα luminosity significantly greater than the overall cluster population. The three
circled objects are heavily crowded stars near the cluster core, and their photometry is thus
highly suspect.
-- 33 --
Table 1. Table of M80 observations: Listed are the number of useable observations at
each date, as well as the total exposure time.
Epoch
PI/Prog. ID
F336W
F439W
F450W
F555W
F656N
F675W
1: 10/5/94
2: 4/5/96
3: 8/29/97
4: 9/5/97
5: 9/12/97
6: 9/21/97
7: 9/29/97
Shara
#5677
Ferarro
#5903
Shara
#6460
Shara
#6460
Shara
#6460
Shara
#6460
Shara
#6460
8: 6/21/00 King
#8655
4×300s
⇒1200s
2×30s
⇒60s
4×900s
⇒3600s
4×600s
⇒2400s
2×700, 800s
⇒2200s
2×700, 800s
⇒2200s
2×700, 800s
⇒2200s
2×700, 800s
⇒2200s
2×700, 800s
⇒2200s
5×160s, 7× 200s
⇒2200s
2×2s, 4×23s
⇒96s
3×1300s
3×260s
⇒3900s ⇒780s
-- 34 --
Table 2. List of the 19 bright X-ray sources from Heinke et al. (2003) with the closest blue
objects, and their separation in arcseconds, after applying the offsets discussed in the text.
Source
R.A. (X-ray)
Decl. (X-ray)
R.A. (blue)
Decl. (blue)
∆
Notes
CX1
CX2
CX3
CX4
CX5
CX6
CX7
CX8
CX9
CX10
CX11
CX12
CX13
CX14
CX15
CX16
CX17
CX18
CX19
16 17 02.814
16 17 02.576
16 17 01.597
16 17 02.005
16 17 01.708
16 17 03.569
16 17 02.164
16 17 01.114
16 17 02.401
16 17 00.407
16 17 02.472
16 17 02.565
16 17 01.755
16 17 02.553
16 17 02.100
16 17 02.119
16 17 02.220
16 17 02.820
16 17 03.850
-22 58 32.67
-22 58 36.48
-22 58 27.95
-22 58 33.03
-22 58 15.34
-22 58 25.30
-22 58 37.27
-22 58 29.33
-22 58 32.60
-22 58 28.87
-22 58 37.86
-22 58 45.00
-22 58 29.29
-22 58 30.50
-22 58 31.80
-22 58 19.80
-22 58 33.70
-22 58 36.00
-22 58 47.10
16 17 2.813
16 17 2.573
16 17 1.474
16 17 1.928
16 17 1.612
16 17 3.569
16 17 2.189
16 17 1.255
16 17 2.346
16 17 1.255
16 17 2.532
16 17 2.539
16 17 1.880
16 17 2.601
16 17 2.141
16 17 2.079
16 17 2.141
16 17 2.916
16 17 3.081
-22 58 31.51
-22 58 36.55
-22 58 24.38
-22 58 32.80
-22 58 15.13
-22 58 25.61
-22 58 37.95
-22 58 27.44
-22 58 33.13
-22 58 27.44
-22 58 36.66
-22 58 45.44
-22 58 26.22
-22 58 29.53
-22 58 32.19
-22 58 22.40
-22 58 33.20
-22 58 37.38
-22 58 36.26
1.160
0.085
3.969
1.086
1.282
0.302
0.766
2.731
0.967
11.820
1.483
0.564
3.524
1.176
0.680
2.666
1.174
1.901
15.172
DN2
DN1
|
0810.1760 | 1 | 0810 | 2008-10-09T20:53:28 | Core instability models of giant planet accretion II: forming planetary systems | [
"astro-ph"
] | We develop a simple model for computing planetary formation based on the core instability model for the gas accretion and the oligarchic growth regime for the accretion of the solid core. In this model several planets can form simultaneously in the disc, a fact that has important implications specially for the changes in the dynamic of the planetesimals and the growth of the cores since we consider the collision between them as a source of potential growth. The type I and II migration of the embryos and the migration of the planetesimals due to the interaction with the disc of gas are also taken into account. With this model we consider different initial conditions to generate a variety of planetary systems and analyse them statistically. We explore the effects of using different type I migration rates on the final number of planets formed per planetary system such as on the distribution of masses and semimajor axis of extrasolar planets, where we also analyse the implications of considering different gas accretion rates. A particularly interesting result is the generation of a larger population of habitable planets when the gas accretion rate and type I migration are slower. | astro-ph | astro-ph |
Mon. Not. R. Astron. Soc. 000, 1 -- ?? (0000)
Printed 11 December 2018
(MN LATEX style file v2.2)
Core instability models of giant planet accretion II:
forming planetary systems
Y. Miguel1,2⋆ and A. Brunini1,2†
1Facultad de Ciencias Astron´omicas y Geof´ısicas. Universidad Nacional de La Plata. 1900 La Plata, ARGENTINA.
2Instituto de Astrof´ısica de La Plata (CCT La Plata-CONICET, UNLP), Paseo del Bosque S/N, La Plata, Argentina
11 December 2018
ABSTRACT
We develop a simple model for computing planetary formation based on the core
instability model for the gas accretion and the oligarchic growth regime for the ac-
cretion of the solid core. In this model several planets can form simultaneously in the
disc, a fact that has important implications specially for the changes in the dynamic of
the planetesimals and the growth of the cores since we consider the collision between
them as a source of potential growth. The type I and II migration of the embryos
and the migration of the planetesimals due to the interaction with the disc of gas
are also taken into account. With this model we consider different initial conditions
to generate a variety of planetary systems and analyse them statistically. We explore
the effects of using different type I migration rates on the final number of planets
formed per planetary system such as on the distribution of masses and semimajor
axis of extrasolar planets, where we also analyse the implications of considering differ-
ent gas accretion rates. A particularly interesting result is the generation of a larger
population of habitable planets when the gas accretion rate and type I migration are
slower.
Key words: Planets and satellites: formation - Solar System: formation
1
INTRODUCTION
Planetary astronomy is a young science, and until recently,
was essentially devoted to the study of planetary bodies in
our own Solar System. The extrasolar planets found since
1995 have vastly expanded our database by increasing the
number of known planets by more than 200. Although the
distribution of masses and semi-major axis of observed ex-
trasolar planets is highly biased towards those planets that
are detectable using Doppler radial velocity technique, the
increase in precision and continuity of the surveys has given
the possibility of unveiling a large variety of planets and re-
formulate the theories of planetary formation. As the num-
ber of detections grows, statistical studies of the properties
of exoplanets and their host stars can be conducted to un-
ravel some of the processes leading to the formation of plan-
etary systems.
In this frame several models of planetary formation
have been presented in the last years. Chambers (2006) and
⋆ E-mail: [email protected]
† Member of the Carrera del Investigador Cient´ıfico. Consejo Na-
cional de Investigaciones Cient´ıficas y T´ecnicas (CONICET).E-
mail: [email protected]
Thommes et al. (2003) have developed semi analytical mod-
els based on the oligarchic growth regime, that retain much
of the simplicity of the analytic models, but incorporate
more physics from the detailed ones. Ida & Lin (2004a,b,
2008) have studied the mass and semimajor axis distribu-
tion of extrasolar planets through a very simple model but
considering all the relevant physics involved in the process
of planetary formation.
our
studies
in
a
We
begin
previous
paper
(Miguel & Brunini 2008), where we developed a very
simple model based on the core instability model. The
accretion of solids was based in the regime where the
largest embryos dominate the dynamics of the planetesimal
disc: the oligarchic growth regime (Kokubo & Ida 1998;
Ida & Makino 1993). When the cores are large enough to
have an associate envelope which can not be sustained by
the hydrostatic equilibrium any more, the gas accretion
process begins. In our previous work we explored the effects
of different gas accretion rates on the final mass distribution
of extrasolar planets, and found a strong dependence. But
our simple model allowed us to form only one core per disc
and did not allow us to consider the dynamical evolution of
the cores or the planetesimal disc.
In this paper, we have improved the model introduced
2
Y. Miguel and A. Brunini
in Paper I considering other important effects. The mainly
improvements are the formation and evolution of several
cores simultaneously in the protoplanetary nebula, a fact
that also influences the dynamic of the planetesimal disc.
The embryos can grow by the accretion of planetesimals or
by accreting other embryos, in this work we consider the
collision among them as a source of growth.
We also incorporate the orbital evolution of the pro-
toplanets due to its tidal interaction with the gas in the
surrounding protoplanetary nebula, considering two regimes
of planetary migration: type I and II regimes, which result
mostly in a radial motion towards the host star. Type I
migration involves low mass protoplanets who are rapidly
moved towards the star (Goldreich & Tremaine 1980; Ward
1986) and type II is the regime of migration for those pro-
toplanets whose mass is large enough to allow them to open
up a gap in its orbit and migrate on a rate controlled by the
disc viscous time-scale (Lin & Papaloizou 1985). This pro-
cesses could have important consequences on the final mass
and semimajor axis distribution, such as on the final number
of planets formed per protoplanetary disc.
The disc of planetesimals also interacts with the nebu-
lar gas. This gas drag effect causes the migration of plan-
etesimals towards the host star before they become large
enough to decouple from the disc gas. This effect, which
was included in our model, affects directly the distribution
of solids in the disc and the growth of the cores.
We show that a variety of planetary systems can be
constructed using these principles, varying the initial condi-
tions, and that the mass and semimajor distribution of ex-
trasolar planets is strongly dependent on the gas accretion
model and type I migration rate considered, being larger the
population of habitable planets when the rate of gas accre-
tion and type I migration are slower.
2 DESCRIPTION OF THE MODEL
In our previous work (Miguel & Brunini 2008), Paper I, we
developed a simple model for computing planetary forma-
tion based in a work of Ida & Lin (2004a)(hereafter IL04),
but incorporating the accretion rates given by Fortier et al.
(2007). This lead us to a very different results regarding
the final mass distribution of extrasolar planets, which was
found strongly dependent on the gas accretion model con-
sidered.
With the aim of improve the model introduced in Paper
I, we have extended it by considering other important effects.
In this section we explain briefly our previous model and
show the improvements made.
2.1 The previous model
For the sake of completeness, we will explain our previous
model. Although very simple, it incorporates all the impor-
tant aspects of planetary formation.
We considered one initial core per protoplanetary disc,
with an initial mass of 10−5M⊕. The initial disc was based
on the minimum mass solar nebula of Hayashi (1981)(for
more details see section 2.2), and it was defined between
0.01 and 100au. The population of the nebular mass scal-
ing parameter, fd, which states the solid mass in the disc
in terms of the minimum mass solar nebula model, was dis-
tributed with a Gaussian distribution in terms of log10(fd)
with dispersion 1 and centred at 0.25, in order to charac-
terise different discs. We supposed that fg, which is a scaling
parameter for the gas disc, was equal to fd, this correspond
to discs with solar metallicity.
The solid and gaseous disc were not time-invariant. The
gaseous disc changed globally, decaying exponentially with
a characteristic time-scale of 4x106years, in accordance to
current estimates of disc lifetimes around young solar type
stars (Beckwith & Sargent 1996) and the solid disc changed
locally, suffering the depletion of planetesimals produced by
the effect of core's accretion.
The 10−5M⊕ initial core was located with equal prob-
ability per interval of log(a), and it grew with the fol-
lowing solids accretion rate based on the particle-in-a-box-
approximation (Safronov 1969),
dMc
dt
= 10.53ΣdΩK R2(cid:16)1 +
2GMt
Rσ (cid:17)
(1)
where R is the physical radius of the core, ΩK is the Kepler
frequency, Mt is the total mass of the core, Σd is the solids
disc surface density and σ is the relative velocity between
the embryo and the disc of planetesimals.
The factor 10.53 was introduced for considering the F
factor introduced by Greenzweig & Lissauer (1992), and the
approximations made for the eccentricity e, the inclination
i and the disc scale of high h(a) in the high-σ equilibrium
regime, 2i ≃ e and h(a) ≃ ai.
Finally, a factor of 4 was introduced in the above equa-
tion, which is an approximate value taken in order to fit
the solid accretion rate used by Fortier et al. (2007), which
includes the evolution of the planetesimals r.m.s. e and i
and the drag effect caused by the gaseous envelope on the
incoming planetesimals. When the gravitational perturba-
tion due to the protoplanets is balanced by dissipation due
to the gas drag, the planetesimals r.m.s. eccentricity attains
an equilibrium value, which was obtained by Thommes et al.
(2003). This effect was considered in the simulations per-
formed by Fortier et al. (2007), where the solid accretion
rate was prescribed as that obtained by Thommes et al.
(2003), and where the enhancement of the protoplanet's
cross section due to the drag effect caused by the proto-
planet's gaseous envelope on the entering planetesimals was
also considered. We compared the core accretion rates found
without considering these effects with all the cases consid-
ered by Fortier et al. (2007). The results showed that the
core accretion rate is ∼ 4 times larger when the evolution of
the e and i are considered (Miguel & Brunini 2008).
When a core of mass Mc reached a certain critical mass,
it started the gas accretion on to the core. The critical
mass at which the gas accretion process began was obtained
analytically by Stevenson (1982) and it was improved by
Ikoma et al. (2000) through numerical simulations. We con-
sidered a simplified version,
Mcrit ∼ (cid:18)
Mc
10−6M⊕yr−1(cid:19)
1
4
(2)
The gas accretion rate was obtained by fitting the re-
sults of the self-consistent code developed by Fortier et al.
(2007)
dMg
dt
=
Mt
τg
Core instability models of giant planet accretion II
3
(3)
L⋆
L⊙
≃ (cid:18) M⋆
M⊙(cid:19)4
(11)
where Mg is the mass of the surrounding envelope and τg is
its characteristic growth time,
τg = 1.64x109(cid:18) Mt
M⊕(cid:19)−1.91
yrs
(4)
This process stopped when the protoplanet consumed
all the gas available on its feeding zone, or what is the same,
when it reached its isolated gas mass, given by,
Mg,iso = 2πabrHΣg
(5)
where Σg is the gas surface density and brH is the typical
distance between two adjacent embryos, which according to
Kokubo & Ida (1998) is ∼ 10rH , being rH the Hill radius.
2.2 The protoplanetary nebula and the initial
cores
As in our previous paper (Miguel & Brunini 2008), the
structure of the protoplanetary nebula is based on the min-
imum mass solar nebula (MMSN) of Hayashi (1981), where
the surface density of solids is defined as follows:
Σd = 7fdηice(cid:18) a
2
1au(cid:19)− 3
gcm−2
(6)
where ηice is 1 inside the ice condensation radius and 4 out-
side it, expressing the effect of water ice formation across the
M⊙(cid:1)2
ice boundary. The "snow line" is located at aice = 2.7(cid:0) M⋆
au from the central star of mass M⋆.
Similarly, the volume density of gas follows an expo-
nential profile, as explained Thommes et al. (2003), which
is
where ρg,0 is the mid-plane value density given by,
ρg,0(a) = 1.4x10−9fg(cid:18) a
4
1au(cid:19)− 11
gcm−3
(8)
Unlike the previous calculations, we assume that the
surface density of gas everywhere varies with time as:
ρg = ρg,0e
− t
τdisc
(9)
with τdisc the gas depletion time-scale.
We consider a disc between Rin and 30 au, where the
inner disc radius is estimated by,
Rin = 0.0344Ψ(cid:18) 1500◦K
Tsub (cid:19)2(cid:18) L⋆
L⊙(cid:19)
1
2
au
(10)
as explained Vinkovic (2006). In the above equation Ψ is
a factor of ≈ 2 which depends on the disc structure and
radiative transfer, the dust sublimation temperature, Tsub,
is taken as 1500◦K and L⋆ and L⊙ are the stellar and Sun
luminosity respectively.
In this work we consider stars with masses between 0.7
and 1.4M⊙, the stellar luminosity to this range of masses is
given by
ρ(a, z) = ρg,0(a)e
z2
h(a)2 gcm−3
(7)
Moli ≃
In our previous work we considered one core per disc,
here we form several cores in the same disc. Initially we start
the simulation with a number Nplanets,0 of cores through the
disc, separated by 10rH . This could have important conse-
quences on the final distribution of masses and semimajor
axis of extrasolar planets, specially for the changes in the
dynamic of the planetesimals, and also considering the colli-
sions between embryos as a very important source of growth,
as was shown by Brunini & Benvenuto (2008).
In the core instability model the first step in the plane-
tary formation is the coagulation of dust particles. When
particles of many meters in size have been formed, the
gravitational forces became more important than gas drag,
and the larger planetesimals grow faster than the smaller
ones. This stage of increasingly rapid growth is the run-
away growth (Greenberg et al. 1978; Kokubo & Ida 1996).
In this stage, the planetesimal dynamic is dominated by the
interaction with other planetesimals, until the larger ones
grows so much and start to dominate the velocity distribu-
tion of the planetesimals, (Kokubo & Ida 1998). This is the
oligarchic growth stage, and is the stage where the embryos
form. In this high-σ regime, collisions between planetesimals
tend to be destructive leaving fragments of planetesimals as
a result of the impacts. This fragments are small enough to
be rapidly damped by the gas drag and as a consequence,
gravitational focusing favours the embryo, increasing its ac-
cretion rate (Chambers 2006). This could be an important
factor in the growth of the embryos, but is not taken into
account in this work.
Ida & Makino (1993) found through N-body simula-
tions, the minimum mass necessary for starting the oli-
garchic growth stage, which is given by
1.6a
6
5 b
3
5 m
3
3
5
5 Σ
d
1
5
M
⋆
(12)
where m is the effective planetesimal mass, which was found
adopting the next a cumulative power law mass distribution
of planetesimals,
N (m)dm ∝ m−pdm
(13)
where p is 1.8 − 3. We consider p = 2.5,consistent with the
results of Kokubo & Ida (1998, 2000) for the spectrum of
masses of a population that has relaxed to isolated runaway
bodies. Then the typical mass we use is found through
(14)
mmin
m = R mmax
R mmax
mmin
m′N (m′)dm′
N (m′)dm′
here we suppose mmin = 6.3x1012g and mmax = 6.3x1021g
are the minimum and maximum planetesimal mass, equiv-
alent to 0.1 − 100Km radius (e.g. the same values adopted
by Brunini & Benvenuto (2008)), here we note that the disc
mass is contained in the small bodies. The typical planetes-
imal mass found with equation (14) is the one used and is a
constant in all the simulations performed.
One initial core is located at a = Rin, the rest of the
cores are separated 10 rH each other until the end of the
disc is reached. Their initial masses are given by equation
4
Y. Miguel and A. Brunini
(12), where we see that different location in the disc leads
to a different initial oligarchic mass.
2.3 The growth of the cores
The solid accretion rate for a core in the oligarchic growth
regime, considering the particle-in-a-box approximation
(Safronov 1969) is given by equation (1). Thommes et al.
(2003) found a expression considering the evolution of the
planetesimal rms e and i. They found that when the em-
bryos' gravitational perturbation are balanced by dissipa-
tion due to the gas drag, the planetesimal rms e attains an
equilibrium value given by,
eeq(≃ 2ieq ) ≃
1.7 · m
1
5 C
b
1
5
D ρ
1
15 M
1
5
gasM
1
3
t ρ
2
15
m
1
3
⋆ a
1
5
(15)
where CD is a dimensionless drag coefficient which is of order
unity, and ρm is the planetesimal bulk density, which is ≃
1.5gcm−3. The equation (1) could be rewritten using this
expression,
3.9b
≃
dMc
dt
2
5
gasΣd
DG
2
5
2
5 C
4
15
m ρ
ρ
1
3
M a
1
2 M
1
6
⋆ ρ
2
15
1
10 m
2
3
M
t
(16)
where ρM is the embryo bulk density (ρM ≃ 1.5gcm−3).
The main difference between equation (16) and the one
considered in Paper I is that in the above equation we take
into account the evolution of rms e and i through the equa-
tion (15), and in Paper I we approximated this evolution
considering the cases analysed by Fortier et al. (2007).
Equilibrium rms velocities are a good approximation as
long as the embryos reach roughly one Earth mass, a state
that is quickly attained in the regions a < 10au (Chambers
2006). Nevertheless, our model intent to remain as simple
as possible, in such a way that we can perform statistics of
a large number of simulated result. Within this in mind, a
more appropriate velocity evolution model is for the moment
out of our possibilities.
The growth of the cores terminate when they consume
all the planetesimals on their feeding zone (∆ac = brH),
or equivalently when the solid superficial density is Σd = 0
(see section 2.4). But they also can stop the growth when the
density of planetesimals is substantially depleted by ejection
(Thommes et al. 2003; Ida & Lin 2004a), this is, when the
ratio of collision to ejection probabilities is
≪ 1
vs(cid:17)4
(cid:16) ve
where ve = p2GM⋆/a is the escape velocity from the pri-
mary and vs = pGMt/R is their characteristic surface ve-
locity.
(17)
Once the core became massive enough to retain a gas
envelope, the effect of this atmospheric gas drag on the plan-
etesimals increases the collision cross section of the proto-
planet. Considering the model for a purely radiative atmo-
sphere (Stevenson 1982; Inaba & Ikoma 2003), Chambers
(2006) found an approximate expression for the enhanced
collision radius (Rc) of the embryo,
where c is the velocity of light, P is the orbital period of
the protoplanet, κ is the opacity of the atmosphere which is
considered as ≃ 4cm2g−1, rm is the planetesimal's typical
radius and the equilibrium eccentricity is considered as ≈ 2
in this expression.
Once the core reaches the critical mass given by equa-
tion (2), the gas accretion process begins. The model for the
gas accretion on to the core is the same used in Paper I, and
it was already explain in section 2.1.
We also perform some simulations considering the gas
accretion rate given by IL04,
τg,(IL04) = 1.x109(cid:18) Mt
M⊕(cid:19)−3
yrs
(19)
which differs mainly in the exponent, taken as 3 by
them. It leads as to a smaller gas accretion rate, which has
an enormous influence on the distribution of masses, as was
explained on Paper I.
Until this point, we have assumed that embryos grow by
accreting gas and planetesimals only. However, embryos can
also grow by accreting other embryos. When two protoplan-
ets are too close to each other, their mutual gravitational
perturbation induce high eccentricities, which enable their
orbits to cross. This process may lead to close gravitational
encounters and violent collisions between the embryos.
We suppose that collision between protoplanets will oc-
cur if their orbital spacing is less than 3.5 Hill radius. Con-
sidering that the evolution of planetary atmosphere after a
merger is a complex and poorly understood process, we anal-
yse two different and extreme scenarios. On the one hand we
assume that all embryo-embryo collisions will lead to coales-
cence to form a single body where the result is simply the
sum of the masses (gas and solids) of both embryos, no mat-
ter the composition of the protoplanets. On the other hand
impacts so energetic can cause a net loss of atmosphere, be-
cause the kinetic energy of the impact is partly transferred
to the associate envelope accelerating its molecules to veloc-
ities greater than the escape velocity from the protoplanet.
So we want to explore this possibility and in the other sce-
nario, if both colliding planets have an associate envelope,
after the impact their atmospheres would be lost and the
result would be a merger of their cores. The mass and semi-
major axis distribution were similar in both cases, for this
reason we stayed with scenario 1) to not add greater com-
plexity to the model.
2.4 Evolution of the solids surface density
Protoplanetary discs are not quite steady, so the surface
density of solids is not a time-invariant. It will change over
time due to different effects, one of them is the depletion
of planetesimals produced by the effect of cores' accretion
(this effect was also considered in Paper I). If, Σd,0 is the
initial surface density, we have to remove what already ate
the embryo, so the evolution of Σd is given by,
Σd = Σd,0 −
Mc
2πabrH
(20)
R (cid:19)4
(cid:18) Rc
=
0.000344µ4 cP
κrmΣd
M⊕(cid:19)2(cid:18) 24e2
(cid:18) Mt
eq
24 + 5e2
eq(cid:19)
(18)
On the other hand, the disc of planetesimals also inter-
acts with the nebular gas. This gas drag effect will cause
a radial motion of planetesimals before they become large
Core instability models of giant planet accretion II
5
enough to decouple from the disc gas. The orbital decay
occur on a rate (Adachi et al. 1976; Thommes et al. 2003),
da
dt
≃ −2
a
Tgas(cid:18) 5
8
e2 +
1
2
i2 + η(cid:19)
1
2(cid:18)η +
49
16
e2 +
1
8
i2(cid:19) (21)
where Tgas is the characteristic time-scale of the gas drag
and η is the fractional deviation of the gas velocity vgas
from Keplerian velocity vK due to the pressure gradient,
η =
vK − vgas
vK
=
13π
64 (cid:16) cs
vK(cid:17)2
(22)
with cs being the sound speed and cs/vK ≃ h(a)/a.
The characteristic time-scale of the gas drag is given by,
Tgas =
m
(CD/2)πr2
mρgaΩK
(23)
Finally, applying continuity to the planetesimal disc,
the gas drag effect acting on the planetesimals leads to a
rate of change of the surface density given by,
∂Σd
∂t
= −
1
a
∂
∂a(cid:16)aΣd
da
dt(cid:17)
(24)
Equation (20) is added to the equation (24) as a sink
term, so we obtain the total change of the solid surface den-
sity in the disc.
with q = 1 − 1.5, we use 1.5.
The time-scale of this inward migration is smaller
than the disc lifetime, since it seems very unlikely, re-
cent works dealing with planet-disc interactions have pro-
posed mechanisms that could slow down or stop type I
migration. Several mechanisms have been proposed like
the study of the magneto-hydrodynamic turbulence in the
disc (Nelson & Papaloizou 2004), the effects of a disc con-
taining a toroidal magnetic field (Terquem 2003), the tri-
dimensional effects studied by Masset et al. (2006), varia-
tion in the temperature gradient and surface density in the
disc (Masset et al. 2006), the effect of including a proper
energy balance on the interaction of a low-mass planet with
a protoplanetary disc (Paardekooper & Mellema 2006), and
so on.
1
Halting or slowing down the inward migration requires
careful computation of tidal effects between the core and the
star, this is much beyond the capabilities of our model. For
this reason a factor of
is introduced in equation (25),
to consider non linear effects without introduce a mayor de-
gree of complexity to our model. If we want to slow down
migration rates, CmigI must be 6 1. We perform simula-
tions with CmigI = 1, 0.1, 0.01. We also consider that the
migration mechanism stops when the core reach the inner
edge of the disc given by equation (10).
CmigI
2.5 Model for planetary migration
2.5.2 type II migration
An important contribution to our understanding about plan-
etary formation and evolution was the discovery of the firsts
Jupiter-like planets orbiting very close to its central star
(hot-Jupiters). The difficulties associated with the forma-
tion of this objects are most pronounced as we are closer to
their host star, and they have awaken an interest in theories
for the migration of protoplanetary embryos due to grav-
itational interaction with the disc. In this section we will
consider two regimes of planetary migration: the type I and
II regimes, which result mostly in a radial motion towards
the star and concern planets of low and high mass respec-
tively.
2.5.1 type I migration
Type I migration acts on low-mass protoplanets, that is, for
those embryos who are unable to open a gap on its orbit.
Our model for this regime of planetary migration is simi-
lar to that of Ida & Lin (2008), who describe in detail the
assumptions made.
This process is caused by an imbalance in the tidal
torques from inner and outer disc leading to angular mo-
mentum exchange and making the planet drifts relative to
the disc material. The time-scales involved in this process
are smaller than the discs lifetimes, and were calculated by
Tanaka et al. (2002) through 3D linear simulations,
τmigI = −
a
a
=
According to the observed range of orbital semimajor axes,
many giant planets are found to a very close proximity to
their parent stars. At this distance the equilibrium tempera-
ture is above the 170◦K, the temperature at which ice (and
the icy cores of giant planets) exists. Actually, giant plan-
ets would form, preferably, at or outside of the ice line, this
implies that they could have suffered great orbital changes
after formation.
The model of giant planet type II migration presented
here is a very simple one, and is essentially the same use by
Ida & Lin (2004a, 2008), so we will describe it briefly.
The process starts when the giant planet form a gap in
the disc. According to Lin & Papaloizou (1993), a necessary
condition to open a gap is
rH > h
(26)
The above equation leads to a mass condition, when the
mass of the protoplanet is larger than
Mgap = 3.75 · 10−4
3
4
M⋆
·(cid:16) a
1au(cid:17)
(27)
a gap is opened in the disc. When this happens, there are
tidal forces acting between the protoplanet and the inner
and outer edges of the gap, whose balance (or imbalance)
depends on their difference in density. The imbalance can
result on a migration of the protoplanet 's orbit (Lin et al.
1996) inward or outward. The migration time-scale is
τmigII = 0.8x106f −1
g (cid:16) Mp
MJ(cid:17)(cid:16) M⊙
M⋆ (cid:17)(cid:16) α
10−4(cid:17)−1(cid:16) a
1au(cid:17)
1
2
yrs(28)
=
1
2.7 + 1.1q(cid:16) cs
≃
5.5x105
2.7 + 1.1q
aΩK(cid:17)2 M⋆
M⊕(cid:17)−1
fg(cid:16) Mp
Mp(cid:16) M⋆
2hρga2(cid:17)Ω−1
(cid:0) a
2(cid:16) M⋆
M⊕(cid:17)
1au(cid:1)
1
K
3
3
2
yrs
(25)
where α is a dimensionless parameter which charac-
terises the viscosity and is taken as 10−3 through this work.
The protoplanet will migrate towards the star if a < Rm
and away if a > Rm, where Rm is the radius of maximum
6
Y. Miguel and A. Brunini
viscous couple. This radius depends on the distribution of
gas surface density, so it changes because Σg is not a con-
stant. From the conservation of angular momentum, and
considering the total disc mass as a constant, it can be found
that
Rm = 10e
2t
5τdisc
au
(29)
Various stopping mechanisms have been proposed, but
none of them seem to be effective in halting migration, for
this reason we stop migration arbitrarily when the planet
reach the inner radius of the disc, which is given by equation
(10).
When a giant planet is migrating inwards, it will per-
turbate the cores located on its path towards the host star.
This encounter can cause the ejection or the accretion of the
core by the giant planet, or neither, which means that the
core will survive the giant planet path.
The effect of the migrating giant planets on the for-
mation of terrestrial planets was studied by Fogg & Nelson
(2007), who present
their N-body simula-
tions of giant planet migration through an inner proto-
planet/planetesimal disc, where the migration of the low-
mass embryos (type I) was also considered.
results of
At the end of the simulations, they found that ∼ 71% of
the initial solid disc survive after the migration of the giant
planet, a very small percentage is ejected into a hyperbolic
orbit (less that 0.2%), and the rest is accreted.
We introduce this results in our simulations, with a very
simplified model where in a close encounter between a giant
migrating inwards and another core, the last one has the 29%
of possibilities of being accreted by the giant and the rest
71% of surviving the giant passage with its orbit unchanged.
We note that this is a very simple model, but in reality this
fraction will change depending on the distance of embryos
from the star.
This effect does not significantly change the results,
since we have more than one migrating giant planet in the
disc. The passage of many giant planets will reduce the sur-
viving probabilities of the cores, until they were finally ac-
creted.
3 RESULTS
In this section we first analyse the characteristics of the
planetary systems formed, and then discuss the effects of
different prescriptions for the gas accretion rate and differ-
ent retardation constants for type I migration on the mass
and semimajor axis distribution of extrasolar planets.
3.1 Characteristics of the planetary systems
formed
In each simulation we generate 1000 discs. Each system
evolves for 107 years, in a rich-metal disc, where the time-
scale for the depletion of the gas, has a uniform log distribu-
tion between 106 and 107 years, and the stellar mass has a
uniform distribution in log scale in the range of 0.7 −1.4M⊙.
The initial number of planets per disc, Nplanets,0, depends
on the initial cores' mass (equation 12), the inner disc radius
(equation 10) and the mass of the host star. The final num-
ber of planets, Nplanets, is a result and shows the evolution
of the planetary system.
Each disc of gas is defined by fg,0, which has a log
normal distribution with a dispersion of 1 and centred on
fg,0 = 1, and the disc of solids is taken as fd = 100.1fg,0
(Ida & Lin 2004b). This is slightly different from what we
had considered before, because in Paper I we used discs with
solar metallicity and here we suppose more metallic discs.
The final number of planets per disc also depends on fg,0
and fd, as seen in Fig. 1, where fd is plot against Nplanets,
considering different values for CmigI .
Figure 1(a) was obtained with CmigI = 1, Fig. 1(b)
considers CmigI = 0.1, the value taken in Fig. 1(c) was 0.01
and the type I migration was not considered in Fig. 1(d).
As seen in the Figures, planetary systems with a large
number of planets (Nplanets > 100) correspond to small val-
ues of fd (< 3), and this fact does not change consider-
ing different rates of type I migration. These discs have low
mass and as a consequence the initial cores will not grow
too much. When the mass of the cores is less than ≈ 0.1M⊕,
they are not so affected by migration (which would favor the
merging) and therefore the discs remain with many small
planets, no matter the rate of migration considered.
On the other hand the values of fd required to form
planetary systems with few planets change when different
migration rates are considered. When CmigI = 1, discs with
fd > 0.5 can form planetary systems with few planets. This
can be explained with the rapid type I migration rate, which
moves the biggest cores inwards accreting the other cores to
its path, forming planetary systems with few planets near
to the host star. Small values of fd implies discs less mas-
sive, which will form planetary systems with few (terrestrial)
planets.
As migration slows down, the range of values of fd re-
quired to obtain planetary systems with few planets becomes
smaller, as seen in figures 1(b), 1(c) and 1(d). In the last fig-
ure, when CmigI = 0, we note that those planetary systems
with few planets correspond to fd > 10, which leads to high
mass discs which will generate giant planets who accrete all
the other cores to its path towards the central star.
We also analyse the kind of planets formed per disc. We
consider that terrestrial planets are those with Mt < 7M⊕,
giants with low percentage of gas are suppose to be the
planets with Mt > 7M⊕ and a percentage of gas mass less
than < 15%, and the others, which have a larger percentage
of gas are gas giants. The histogram presented in Fig. 2
shows the number of terrestrial planets (dotted line with
filled circles), giants with low percentage of gas (dashed line
with filled triangles) and gas giants (solid line with squares)
formed per planetary system.
Figure 2(a) is the one obtained when CmigI = 1. We
can see that planetary systems with few planets have mostly
terrestrial ones and there are few with giant planets. As we
explained, for this rapid migration rate, the range of discs
which form planetary systems with low Nplanets are those
with 0.5 < fd < 30. In this range most of the discs have
low mass and cores have less time to grow because they are
rapidly moved to the central star. As a result most of the
planetary systems with few planets will form only terrestrial
planets. On the other hand those discs with a large value of
fd are able to form planetary systems with giant planets, but
Core instability models of giant planet accretion II
7
100
10
P
S
N
1
1
100
10
P
S
N
1
1
100
10
P
S
N
1
1
10
Nplanets
(a)
10
Nplanets
(b)
10
Nplanets
(c)
100
100
100
Figure 2. Histogram of the final number of terrestrial (dotted
line with filled circles), giant with low percentage of gas (dashed
line with filled triangles) and gas giant planets (solid line with
squares) per planetary disc. This histograms were obtained with
different values of CmigI . In Fig. 2(a) CmigI = 1, figure 2(b) was
obtained when CmigI = 0.1 and in figure 2(c) the value of CmigI
is 0.01. The number of planetary systems with few terrestrial
planets are larger when Cmig is 1, and it decreases when the
type I migration rate slows down. On the contrary, the number of
planetary systems with few gas giants are smaller with CmigI = 1,
and it increases with small values of CmigI .
are the less ones. We also note a big absence of planetary
systems with an intermediate number of planets (Nplanets >
30), this is because of the rapid migration rate, which moves
quickly the cores to the inner disc limit, so they accrete the
other cores to its path, forming planetary systems with few
planets.
(a)
(b)
(c)
(d)
Figure 1. The final number of planets is plot against the disc
mass scaling parameter, fd. Fig. 1(a) shows the results when
CmigI = 1 is considered, fig. 1(b) and 1(c) present the results
when CmigI is equal to 0.1 and 0.01 respectively and in Fig. 1(d)
CmigI is 0 which means that type I migration is not considered.
As seen in the figures, planetary systems with a large number of
planets are those with small values of fd and a large disc mass
is necessary to form planetary systems with few planets, but the
range depends on the type I migration rate considered.
8
Y. Miguel and A. Brunini
When CmigI is smaller (Fig. 2(b) with CmigI = 0.1 and
2(c) with CmigI = 0.01), the population of planetary sys-
tems with few gas giant planets increases, meaning that the
formation of gas giants is favoured with slower type I migra-
tion rates. On the other hand, the number of giant planets
with low percentage of gas remains low, this population is
not affected by the rate of migration. We also note that the
number of planetary systems with an intermediate number
of planets increases, so when migration slows down the fi-
nal amount of planets per disc tends to follow a uniform
distribution.
3.2 Mass and semimajor axis distribution
To investigate the mass and semimajor axis of extrasolar
planets we perform a series of numerical simulations consid-
ering different prescriptions for the gas accretion rate and
different retardation constants for type I migration.
Results are displayed in Fig. 3(a,b,c,d) and 4(a,b,c,d).
The first ones were found by considering the gas accretion
rates obtained by fitting the results of Fortier et al. (2007)
and the second ones were obtained using the gas accretion
rates given by IL04. In the figures 3(a) and 4(a) the factor
considered for delaying the type I migration rate is CmigI =
1, in Fig.3(b) and 4(b) CmigI = 0.1 Fig. 3(c) and 4(c) are
those obtained with CmigI = 0.01 and finally Fig. 3(d) and
4(d) where found without considering the effects of type I
migration (CmigI = 0).
As Fig. 3 and 4 show, the planetary distribution of mass
and semimajor axis is strongly dependent on the gas ac-
cretion model considered (Paper I), as well as the type I
migration rate used.
When CmigI = 1 most of the planets migrate to the
inner disc radius. The mainly difference between the dis-
tribution found with our model and the one obtained with
IL04's gas accretion rate (Fig. 3(a)and 4(a),respectively), is
the population of planets with masses between 1 and 10M⊕,
which is larger on the first one.
In Fig.3(b) and 4(b) (CmigI = 0.1), the type I migration
is slower and there are less planets who reach the inner edge
of the disc. We can see a major population of planets with
1 − 10M⊕ and a > 1au. The population of giant planets are
also increased, but this is larger in Fig.4(b) because with
IL04's gas accretion rate the run away of gas is reached
sooner, and also the Mgap. This allows the planets to start
the type II migration, which is slower than the type I.
Type I migration is the slowest when CmigI = 0.01,
and as seen in Fig.3(c),4(c),3(d) and 4(d), the distribution
found when CmigI = 0.01 is very similar to those found with
type II migration only. When CmigI = 0.01 the population
of planets with masses less than 10 M⊕ is larger than in the
other cases and there are more giant planets, specially when
a > 1au. The differences between 3(c) and 4(c) are similar
to those found when CmigI = 0.1, which is in agreement
with the results found in Paper I.
The "planetary desert", which is a region with a deficit
of planets, was found to be between 100 − 1000M⊕ (Paper
I) with our model. Here we found that the effect of the plan-
etary migration and specially type I migration, is to enlarge
the desert, allowing to the planets with masses between 10
and 100M⊕ to reach the inner limit of the disc and empty
the area with semimajor axis between ≈ 0.2 and ≈ 3au.
]
s
s
a
M
h
t
r
a
E
[
t
M
]
s
s
a
M
h
t
r
a
E
[
t
M
]
s
s
a
M
h
t
r
a
E
[
t
M
]
s
s
a
M
h
t
r
a
E
[
t
M
10000
1000
100
10
1
10000
1000
100
10
1
10000
1000
100
10
1
10000
1000
100
10
1
0.1
1
a [AU]
(a)
10
0.1
1
10
a [AU]
(b)
0.1
1
a [AU]
(c)
10
0.1
1
10
a [AU]
(d)
Figure 3. Distribution of masses and semimajor axis obtained
with our model. Figure 3(a) shows the data obtained with C1 = 1,
in 3(b) C1 = 0.1, in 3(c) C1 = 0.01 and the effect of type I
migration is not considered in 3(d). We find a larger population
of terrestrial planets and a lower population of gas giant beyond
the snow line than those observed on Fig. 4. We also observed a
Core instability models of giant planet accretion II
9
]
s
s
a
M
h
t
r
a
E
[
t
M
]
s
s
a
M
h
t
r
a
E
[
t
M
]
s
s
a
M
h
t
r
a
E
[
t
M
]
s
s
a
M
h
t
r
a
E
[
t
M
10000
1000
100
10
1
10000
1000
100
10
1
10000
1000
100
10
1
10000
1000
100
10
1
0.1
1
10
a [AU]
(a)
0.1
1
10
a [AU]
(b)
0.1
1
a [AU]
(c)
10
0.1
1
10
a [AU]
(d)
Figure 4. Similar figures as 3, but the gas accretion rates of
IL04 are considered. A comparison with Fig. 3, shows a smaller
population of terrestrial planets and a larger of giant planets with
Mt > 100M⊕, especially when the migration slows down,4(b),
4(c) and 4(d).
Table 1. Percentage of final planets that are habitable (0.3 6
Mt 6 7M⊕ according to Ikoma & Genda (2006)'s criteria) found
with our model (Model I), compared with the results obtained
with IL04's gas accretion rate (Model II). Different delaying type
I migration constants are considered.
Scenario
Habitable
(0.3 6 Mt 6 7M⊕)
Planets
Model I, CmigI = 1
Model II, CmigI = 1
Model I, CmigI = 0.1
Model II, CmigI = 0.1
Model I, CmigI = 0.01
Model II, CmigI = 0.01
87.02 (%)
82.2 (%)
90.5 (%)
86.9 (%)
91.57 (%)
87.18 (%)
Another important result is that we found a major pop-
ulation of habitable planets with our model than the one
found with IL04's gas accretion rates. In order to compare
the different population of terrestrial planets found with dif-
ferent gas accretion rates, we adopt a criteria where a hab-
itable planet is a planet with masses between ≈ 0.3 and
≈ 7M⊕ (Ikoma & Genda 2006). The upper limit to the
mass was obtained by Ikoma & Genda (2006) with a dif-
ferent time-scale for gas accretion, with our time-scale this
value is probably larger, but we consider 7M⊕ as a nominal
value for comparison.
The Table 1 shows the percentage of final planets that
are habitable according to Ikoma & Genda (2006)'s defini-
tion. In the table, model I is the one considering our gas
accretion rate, and model II consider IL04's gas accretion
rate.
With our model the run away gas accretion process is
reached by the cores at a larger mass (Paper I) so there are
terrestrial planets that never reach this "crossover mass"
and for this reason we found more planets with masses less
than 7M⊕. So a slower gas accretion rate leads to larger
population of habitable planets.
4 CONCLUSIONS
In our previous work, we had developed a very simple model
for computing planetary formation, which had allowed us
to show the strong dependence between the gas accretion
model considered and the mass distribution of extrasolar
planets.
In order to get a better understanding of the processes
involved in planetary formation, we improve our previous
model, including significant effects that had not been taken
into account in Paper I while maintaining the computational
speed.
We have presented the results of the simulations per-
formed with our improved model, considering the formation
of several planets per disc and taking into account the colli-
sion among them as a source of potential growth. The forma-
tion of several cores simultaneously in the disc has a strong
influence on the dynamics of the planetesimal disc. The evo-
lution due to this effect and due to the gas drag effect was
also considered in this work.
to a larger population of habitable planets, and lower pop-
ulation of gas giants beyond the ice condensation line.
REFERENCES
Adachi I.,Hayashi C.& Nakazawa K., 1976, Prog. Theor.
Phys. 56, 1756
Beckwith S.V.W., Sargent A. I., 1996, Nat, 383, 139
Brunini A., Benvenuto O.,2008, Icarus, 194 ,800
Chambers J., 2006, Icarus, 180, 496
Fogg M., Nelson R., 2007, A&A, 472,1003
Fortier A., Benvenuto O., Brunini A., 2007, A&A, 473, 311
Greenberg R., Hartmann W.K., Chapman C.R., Wacker
J.F., 1978, Icarus, 35, 1
Greenzweig Y., Lissauer J.J., 1992, Icarus, 100, 440
Goldreich P., Tremaine S., 1980, ApJ, 241, 425
Hayashi C., 1981, Prog. Theor. Phis. Suppl., 70, 35
Ida S., Lin D.N.C., 2004, ApJ , 604, 388
Ida S., Lin D.N.C., 2004, ApJ , 616, 567
Ida S., Lin D.N.C., 2008, ApJ , 673, 487
Ida S., Makino J., 1993, Icarus, 106, 210
Ikoma M., Genda H., 2006, ApJ, 648, 696
Ikoma M., Nakazawa K., Emori E., 2000, ApJ, 537, 1013
Inaba S., Ikoma M., 2003, A&A, 410, 711
Kokubo E., Ida S., 1996, Icarus , 123 , 180
Kokubo E., Ida S., 1998, Icarus, 131, 171
Kokubo E., Ida S., 2000, Icarus, 143, 15
Lin D. N. C., Bodenheimer P., Richardson D., 1996, Na-
ture, 380, 606
Lin D. N. C., Papaloizou J. C. B., 1985, Protostars and
Planets II, ed. D. C. Black & M. S. Matthew (Tucson:
Univ. of Arizona Press), 981
Lin D. N. C., Papaloizou J. C. B., 1993, Protostars and
Planets III, ed. E. H. Levy and J. I. Lunine (Tucson: Univ.
of Arizona Press), 749
Masset F. S., D'Angelo G., Kley W., 2006, ApJ, 652, 730
Masset F. S., Morbidelli A., Crida A., Ferreira J., 2006,
ApJ, 642, 478
Miguel Y., Brunini A., 2008, MNRAS, 387, 463
Nelson R. P., Papaloizou J. C. B., 2004, MNRAS, 350, 849
Paardekooper S. J., Mellema G., 2006, A&A 459, L17
Safronov V., 1969, Evolution of the Protoplanetary Cloud
and Formation of the Earth and Planets (Moscow: Nauka
Press)
Stevenson D.J., 1982, Planet. Space Sci., 30, 755
Tanaka H., Takeuchi T., Ward W., 2002, ApJ, 565, 1257
Terquem C. E. J. M. L. J., 2003, MNRAS, 341, 1157
Thommes E.W., Duncan M.J., Levison H.F., 2003, Icarus,
161, 431.
Vinkovic D., 2006, ApJ, 651, 906
Ward W., 1986, Icarus, 67, 164
10
Y. Miguel and A. Brunini
Finally, we have analysed the interaction between the
protoplanets and the disc, which leads to a planetary migra-
tion (type I and II). The migration of a giant planet towards
the central star could perturbate the cores placed in its path,
causing the ejection or the accretion of the core by the giant
planet. The another possibility is the survival of the core to
the passage of the giant planet. We have performed numeri-
cal simulations taking into account that in a close encounter
a terrestrial planet has the 71 percent chance of surviving
the passage of a giant, but these simulations have not pro-
duced significant changes in the results due to the fact that
in a real system several protogiant planets cross the region
of terrestrial planets.
The final number of planets formed per disc is a direct
result of the planetary system evolution. We showed that
the distribution is strongly dependent on the type I migra-
tion rate considered, noting an absence of planetary systems
with an intermediate number of planets when the fastest mi-
gration rate is considered and seeing that it becomes more
uniform when the migration slowed down.
The final number of planets per planetary system also
depends on the initial mass of the disc, those discs with low
mass will form planetary systems with a large number of
planets and a higher mass disc is necessary to form plane-
tary systems with a few ones, but the range of disc masses
corresponding to the final number of planets depends on the
type I migration rate considered. For a rapid migration rate
low mass discs are able to form planetary systems with few
planets, but when the migration slows down only high mass
discs could form this kind of planetary systems.
However we note that many of the systems with a large
number of planets will be unstable and undergo further evo-
lution since the orbital spacing is small, and eccentricities
will increase after the gas is gone. The final number of plan-
ets will be substantially smaller in some systems, but we do
not consider this fact in this work.
We also have analysed the kind of planets formed per
disc, and found that planetary systems with a large number
of planets have only terrestrial ones, but when we looked at
those planetary systems with few planets, the kind of planets
formed varies according to the type I migration rate consid-
ered. When we used the fastest rate, most of the planets are
terrestrial planets, but when the rate is slower the amount
of terrestrial planets decreases and most are now giants. In
conclusion we found that the gas giant formation is favored
at slower type I migration rate. On the other hand the giant
planets with few gas reminds little in all the simulations,
this population is not affected by the rate of migration.
When we analyse the mass and semimajor axis distribu-
tion we consider different gas accretion rates, and the results
show that this distribution is strongly dependent on the gas
accretion model and also on the rate of migration, mainly
due to type I migration effects. The boundaries of the plan-
etary desert are enlarged due to the rapid migration of the
embryos that did not reach the necessary mass to open up
a gap. But we note that the planetary desert would be a bit
different, since the actual observational sample shows that
the number of observed planets at distances smaller than
0.1 AU from the central star is not as large as that found
through numerical simulations, which could mean that most
planets do not stop their migration at the inner edge of the
disc. Finally we found that a lower gas accretion rate leads
|
astro-ph/0310800 | 1 | 0310 | 2003-10-28T18:01:09 | Galactic Binaries as Sources of Gravitational Waves | [
"astro-ph"
] | I review the expected Galactic sources of gravitational waves, concentrating on the low-frequency domain and summarise the current observational and theoretical knowledge we have. A model for the Galactic population of close binaries, which is tested against the observations is used to predict the expected signal for the future space based gravitational wave detector LISA. With a simple model for the electro-magnetic emission from the same binaries I argue that a fair number of the LISA systems have electro-magnetic counterparts, which can be used to improve in particular the distance and mass measurements of these systems by LISA. Furthermore, LISA will enable us to test some aspects of the theory of binary evolution that are very difficult to asses in different ways. | astro-ph | astro-ph |
Galactic Binaries as Sources of Gravitational
Waves
G. Nelemans
Institute of Astronomy, Madingley Road, Cambridge CB3 0HA, UK
Abstract. I review the expected Galactic sources of gravitational waves, concentrating on the low-
frequency domain and summarise the current observational and theoretical knowledge we have. A
model for the Galactic population of close binaries, which is tested against the observations is used
to predict the expected signal for the future space based gravitational wave detector LISA. With a
simple model for the electro-magnetic emission from the same binaries I argue that a fair number of
the LISA systems have electro-magnetic counterparts, which can be used to improve in particular
the distance and mass measurements of these systems by LISA. Furthermore, LISA will enable us
to test some aspects of the theory of binary evolution that are very difficult to asses in different ways.
1. INTRODUCTION: GALACTIC SOURCES OF GWR
As for all astrophysical phenomena, if it is present in our own Galaxy it often can be
studied in most detail, simply because of the proximity of the sources. This also holds
in some sense for gravitational wave phenomena. Many Galactic sources are, or could
be (strong) gravitational wave radiation (GWR) sources. In particular binary stars are
obvious sources as has been realised long ago [e.g. 1].
The amplitude of the gravitational wave signal increases with the chirp mass of the
binary and its gravitational wave frequency [e.g. 2], which means that more massive,
equal mass, short period binaries are the most promising sources. The frequency range
covered by the current and planned detectors is limited to to frequencies between about
0.1 mHz and 1 Hz for the space based detector LISA and about 50 to 5000 Hz for ground
based detectors. For binary objects this translates to orbital periods between 5.5 hr and 2
seconds and 40 and 0.4 milliseconds for space and ground based detectors respectively.
These short periods imply small separations (or very high masses) through Kepler’s law,
thus for stellar mass objects this means that we have to concentrate on compact stars, in
particular helium stars, white dwarfs, neutron stars and black holes.
The most important classes of binary stars as sources of gravitational waves are:
helium star – white dwarf/neutron star binaries, double whi te dwarfs, white dwarf
– neutron star/black hole binaries, and double neutron star /black hole binaries (see
Fig. 1). If these objects form in binaries with orbital periods below ∼10 hr, the angular
momentum losses due to GWR will make them spiral together within a Hubble time,
until the separations are small enough that mass transfer starts. Helium stars start mass
transfer typically at periods of 30 – 60 min (GWR frequencies of 1 – 2 mHz), but evolve
to shorter periods, before they reach a period minimum around 10 min [e.g. 3, 4]. Double
white dwarfs and white dwarf – neutron star/black hole binar ies start mass transfer at
|
astro-ph/0105379 | 1 | 0105 | 2001-05-22T08:29:09 | Metal-rich and metal-poor globular clusters in ellipticals: Did we learn anything? or Constraints on galaxy formation and evolution from globular cluster sub-populations | [
"astro-ph"
] | A brief review on globular cluster sub-populations in galaxies, and their constraints on galaxy formation and evolution is given. The metal-poor and metal-rich sub-populations are put in a historical context, and their properties, as known to date, are summarized. We review why the study of these sub-populations is extremely useful for the study of galaxy formation and evolution, but highlight a few caveats with the current interpretations. We re-visit the current globular cluster system formation scenarios and show how they boil down to a single scenario for the metal-poor clusters (namely the formation in ``universal'', small fragments at high $z$) and that a hierarchical formation seems favored for the metal-rich clusters. | astro-ph | astro-ph |
Extragalactic Star Clusters
IAU Symposium Series, Vol. Vol. 207, 2001
Eva K. Grebel, Doug Geisler, and Dante Minniti, eds.
Metal-rich and metal-poor globular clusters in ellipticals:
Did we learn anything?
or Constraints on galaxy formation and evolution from
globular cluster sub-populations
Markus Kissler-Patig
ESO, Karl-Schwarzschild-Str. 2, 85748 Garching, Germany
Abstract. A brief review on globular cluster sub-populations in galax-
ies, and their constraints on galaxy formation and evolution is given. The
metal-poor and metal-rich sub-populations are put in a historical context,
and their properties, as known to date, are summarized. We review why
the study of these sub-populations is extremely useful for the study of
galaxy formation and evolution, but highlight a few caveats with the
current interpretations. We re-visit the current globular cluster system
formation scenarios and show how they boil down to a single scenario
for the metal-poor clusters (namely the formation in "universal", small
fragments at high z) and that a hierarchical formation seems favored for
the metal-rich clusters.
1. History of blue and red populations
1.1.
In the beginning...
Stellar sub-population in galaxies were introduced with the concept of popula-
tion I and II in the Milky Way (Baade 1944). Galactic globular clusters were
typically associated with the population II, although some were recognized as
belonging to the disk/bulge (Becker 1950 and others). The idea of globular
cluster sub-populations in the Galaxy was probably only brought to general ac-
ceptance with the work of Zinn (1985).
Ellipticals were not thought of as complex systems until the 80's and mostly
viewed as hosting an old, metal-rich stellar population. The lack of resolution
into single stars still prohibits us from clearly identifying stellar sub-populations
in early-type galaxies. Only as spiral–spiral merger became an alternative to
forming ellipticals, Ashman & Zepf (1992) came up with a simple model that
predicted globular cluster sub-populations in ellipticals if these formed in ma-
jor mergers. Shortly afterwards, Zepf & Ashman (1993) identified such sub-
populations in the color distributions of globular clusters in giant ellipticals.
1.2. Are mergers the answer?
For several years, the idea of sub-populations associated with mergers raged
within the community, still leading to hot debates today. The formation of new
globular(?) clusters was observed in merging galaxies (Holtzman et al. 1992
1
2
Markus Kissler-Patig
for the first compelling evidence), further supporting the merger scenario. The
debate shifted to whether these new clusters would actually evolve into globular
clusters (which now appears to be the case, e.g Schweizer's contribution in this
volume), and whether their general properties (spatial distributions, total num-
ber, specific frequency, ages and metallicities) are compatible with those of the
metal-rich populations in ellipticals (which is still an open question). Note the
caveat that we implicitly assume that what we learn from today's mergers ap-
plies to the mergers that potentially formed the giant ellipticals at much higher
redshift, which might not be the case.
In the mid-90's, several groups started to propose alternatives to major
mergers being the only explanation for multiple sub-populations (e.g. Kissler-
Patig 2000 and references therein). But the alternative "scenarios" remain fairly
vague in terms of predictions or details. Generally, all scenarios tend to mix to
some degree globular cluster formation, globular clusters system assembly, and
galaxy formation. Section 5.2. attempts to provide a critical summary.
2. Sub-population properties: room to improve
In order to understand the origin of the sub-populations (which is the first step
required in order to use them to constrain galaxy formation and evolution), we
need to understand their properties. Do the sub-populations differ in metallicity
only (metallicity differences being the reason why we detected them in the first
place)? Or do they resemble each other in many aspects? A brief summary of
our current knowledge is given below.
2.1. Our biases
In the mid-90's several giant ellipticals were known to host sub-populations.
Blue and red globular clusters were identified in the color distributions and sub-
samples created based on the colors. A few aspects/selections of these studies are
of interest and lead to biases that need to be understood as we try to understand
the sub-population properties in more detail.
i) No "clean" sub-sample can be obtained from the color distributions
alone; rather each sub-sample contains an unknown fraction of the other sub-
population. That fraction is not trivial to quantify since it depends both on
observational errors that smear each color peak, and on the necessary but too
simple assumption that colors are driven by metallicity alone. The fraction
of contamination should be estimated and the impact on the uncertainty of the
derived properties quantified.
ii) The term bi-modal was introduced early on and is still widely used,
driving our thoughts automatically to two sub-populations only. A third peak
in the color distributions was mentioned in some galaxies, but in order to study
the properties of individual sub-populations, only two groups were considered
(true for all studies to date). The latter assumption seem acceptable as a first
approximation since clear differences were noticed, but the question remains how
long this perspective will limit our ability to understand systems that are almost
certainly more complex. We should keep in mind that not two but multiple sub-
populations are most probably present, and should work towards a finer splitting
of the sub-populations.
Globular Cluster Sub-Populations
3
iii) Our samples of galaxies are still very much biased in favor of central
giant ellipticals, or at least very luminous giant ellipticals. The reason is of
course that these galaxies host the most clusters and are therefore the easiest
systems (with enough number statistics) to be studied. Unfortunately, these are
the galaxies that we then tend to call "typical" despite the fact that they are
the rarest and most extreme examples. A much more subtle bias is introduced
in sub-population studies by an implicit selection of galaxies with well separated
peaks in their globular cluster color distribution. For these cases the separation
of the two sub-population is of course the easiest. However, assuming that the
metal-poor population is rather constant in age and metallicity (see below) this
selection does bias us against galaxies with less metal-rich clusters and interme-
diate age metal-rich sub-populations. Future samples should include early-type
galaxies of all types and luminosities, and in all types of environments (espe-
cially intermediate-luminosity field galaxies), irrespective of the properties of the
globular cluster system.
2.2. Differences and Similarities
In the following we briefly list common properties and differences between the
metal-rich and metal-poor sub-populations, as studied to date. No long descrip-
tion is given and the list of references is restricted to a few studies that include
further references (see also Ashman & Zepf 1998, Kissler-Patig 2000, Harris 2001
for further references).
The spatial distributions The first property that was studied separately for
blue and red clusters is the radial surface density profile. Geisler, Kim & Lee
(1996) first noticed, in NGC 4472, that the red clusters were more concentrated
towards the center than the blue ones. The radial surface density profile of the
red clusters is steeper than the one of the blue ones leading to a color gradient
with radius for the whole system. This behavior was observed in the cluster
systems of several giant elliptical galaxies.
Further, Kissler-Patig et al. (1997) first noticed, in NGC 1380, that red and
blue clusters also differed in their 2-dimensional spatial distributions. The red
cluster system appears more elliptical than the blue one. The blue clusters are
not only more diffuse but also follow a rather spherical distribution, while the
red clusters follow the ellipticity and position angle of the galaxy. Again, this
behavior was confirmed in a few other galaxies.
This leads to an association of the blue clusters with the "halo" (although
ill defined for early-type galaxies) and an association of the red clusters with the
"bulge" or spheroid, which represents the majority of the stars in these galaxies
(see also Sect. 4.).
The kinematics Kinematics for the red and blue sub-populations were inves-
tigated in three early-type galaxies only (M 87, NGC 4472, NGC 1399), all
central giant ellipticals. It is thus unclear whether these can be regarded as typ-
ical. Nevertheless, it is clear that the kinematics (both the rotation and velocity
dispersion) of the red and blue clusters differ in these galaxies. A consistent
picture has not yet emerged (e.g. Kissler-Patig & Gebhardt 1998, Kissler-Patig
et al. 1999, Zepf et al. 2000). Notice also that no clear predictions exist for the
kinematics of the sub-populations as a function of assembly/formation scenario.
4
Markus Kissler-Patig
Ages and metallicities The difference in metallicity between red and blue sub-
populations is clear from the color distributions (metallicity being the main
contributor to color at old ages). However, it remains unclear how much red
and blue populations differ in age. Relative age differences can be measured
spectroscopically (e.g. Kissler-Patig et al. 1998, Cohen et al. 1998) or photomet-
rically (e.g. Puzia et al. 1999). The bottom line is that red and blue clusters
appear coeval to within the (large) measurement errors (2–4 Gyr), with some
disputed claims of the red clusters being younger by a couple of Gyr. The abso-
lute ages of the clusters in early-type galaxies appears similar to the one of the
old clusters in the Milky Way and M 31, as judged by spectroscopic line indices.
The sizes Recently (Kundu & Whitmore 1998, Puzia et al. 1999, Larsen et
al. 2000) size differences were discovered between red and blue clusters in early-
type galaxies. The blue clusters appear systematically more extended in all
galaxies, independently of radius. The currently favored explanation is that the
size difference is a relic of the formation process: the blue clusters could have
formed in a less dense environment than the red ones.
3. More than two sub-populations
The above differences in sub-population properties confirm that a splitting in
color leads to two sub-groups with physical differences. A question that was
seldom asked to date is: whether each sub-group really represents a single
sub-population, or whether one or both sub-groups actually host multiple sub-
populations.
The color (equivalent to [Fe/H]) histogram shows the distribution of a loga-
rithmic metallicity value. In figure 1, we illustrate for three cases (with increas-
ingly pronounced metal-rich population) how such a distribution would look
in linear metallicity Z. The two clear peaks in [Fe/H] (translated linearly from
V − I) disappear when plotted in Z. For this particular choice of zero point
(solar=1), the blue peak gets "compressed" into an even clearer peak between
0 and 0.1–0.2 Z⊙. The typical gap in color at [Fe/H]∼ −0.7 dex can still be
guessed at Z∼0.2 . However, the red peak gets spread over several tenths in Z
and is not recognizable as a single peak anymore. Simple simulations show that
the spread of the metal-rich population is not artificially induced by the fixed
errors in logarithmic space, but indeed due to a spread in metallicity that gets
"played down" in [Fe/H].
While the blue clusters still appear to be a single physical group (see also
the next sections), there is no compelling evidence that this is true for the red
clusters too. Clearly, this prompts the question whether the metal-rich sub-
group is, in many or most cases, just an amalgamation of multiple metal-rich
sub-populations that do not necessarily share the same origin. Alternatively, for
smaller red populations such as the one of the Milky Way, the spread could also
be explained by an extremely fast enrichment of the material out of which the
metal-rich clusters formed (contrary to the environment in which the metal-poor
clusters formed). This would imply that the metal-rich clusters show a small
spread in age too, and that they formed in a potential well deep enough to retain
the enriched gas. It remains to be shown whether the latter scenario could also
Globular Cluster Sub-Populations
5
Figure 1.
Comparing [Fe/H] and Z distributions of globular clusters
in three early-type galaxies with a sequence of pronounced metal-rich
sub-populations.
apply to the formation of several hundreds to thousands of globular clusters in
giant ellipticals.
Thus, the interpretation of the metal-rich peak appears more complicated
than originally assumed and the red population might well host multiple sub-
populations. This would imply a complex star/cluster formation history, i.e. not
a single collapse or single major merger event might be at the origin of the metal-
rich clusters, but rather several such events. Given the above: Does the result of
a KMM test (that imposes 2 Gaussians to the color distribution) make a physical
sense? Or does it only mislead us to consider two and not more sub-populations
in a system, and thus over-simplify the interpretation?
4. The globular cluster – star connection
The above problems lead us to consider the connection between globular clus-
ters and stars. Can we identify the multiple stellar populations that should be
associated with the multiple globular cluster sub-population?
4.1. The link between globular cluster and star formation
A shaky aspect of our current interpretation of globular cluster formation in
terms of star-formation history of the galaxy is that we assume for simplicity
that globular cluster formation traces star formation one-to-one. The fact that
6
Markus Kissler-Patig
a strong correlation exists between star and cluster formation is supported by
the fairly constant (within a factor of two) specific frequency in all galaxies,
the correlation between star and cluster formation, etc... (e.g. Introduction in
Puzia et al. 1999). However, a one-to-one correlation is certainly too simplistic.
In fact, there is growing evidence that the blue globular clusters are associated
with few stars, while the red clusters form together with the majority of stars.
The first piece of evidence comes from the fact that high specific frequencies
are systematically observed in "halos" and in dwarf galaxies, both dominated
by blue clusters.
The second hint comes from the fact that diffuse stellar population studies
in early-type galaxies (Maraston & Thomas 2000, Lotz et al. 2000) imply only a
small fraction (typically 10%) of old metal-poor stars as opposed to a majority
of old metal-rich stars. In contrast, old metal-poor and metal-rich clusters are
roughly present with a 50%/50% share in these galaxies, i.e. blue clusters have
fewer stars associated with them than red clusters do.
The third indication comes from direct number counts of stars and globular
clusters as a function of metallicity in the nearby elliptical NGC 5128 (see Harris,
Harris & Poole 1999). A comparison of the metallicity distributions for stars
and globular clusters shows the much higher ratio of stars to globular clusters
at the metal-rich end.
4.2. The color of specific frequency
The conditions for formation of the metal-poor globular clusters do not seem
favorable to the formation of a large fraction of stars, while the contrary seems
true for the metal-rich population. A possible explanation could be that metal-
poor clusters form in shallower potential wells and eject a lot of gas during their
formation that cannot be processed further into stars. While the metal-rich
clusters form in deeper potential wells retaining the gas and allowing an efficient
star formation. This assumes that clusters collapse at the very beginning of a
star formation burst.
The fact is that the specific frequency (Sn) of the blue sub-population is
much higher than the one of the red sub-population. The specific frequency
of a galaxy could thus be increased by forming (or accreting) a large quantity
of metal-poor clusters+stars. On the other hand, when comparing the specific
frequency of different galaxies, one should correct for the fraction of blue to red
clusters present. Also the (in my opinion wrong) argument that specific fre-
quency has anything to do with a major merger in the past history of the galaxy
needs to be revised. From the above, a major merger, producing metal-rich
clusters, could only decrease the specific frequency, and the specific frequency
of a spiral–spiral merger could be equal to or lower than for one of the progeni-
tor spirals alone. But such arguments imply that we know the Sn produced in
mergers, which we do not.
In any case, the "color of the specific frequency" needs to be taken into
account when using Sn to constrain any formation scenario, as well as the fact
that "Sn" is constant when normalized to total baryons instead of just stellar
light (McLaughlin 1999).
Globular Cluster Sub-Populations
7
5. Formation models for dummies (by a dummy)
Ideally, we would like to use globular clusters to constrain the star formation
history of their host galaxies. In practice, we need to know how/where globular
clusters form in the first place, and what the assembly histories of the galaxies
were. Information on all these points is contained in the globular cluster system
properties. However, all the aspects are entangled and when we try to set up a
"scenario" we often mix all these aspects in.
The currently discussed scenarios fall roughly into three categories: i) the
formation of globular clusters and of the host galaxy in a major merger event; ii)
the formation of all globular clusters within the host galaxy; iii) the formation
of the globular clusters in fragments of different sizes that assemble later into
a giant galaxy (e.g. Kissler-Patig 2000 for references to the different flavors of
each scenario). Some more scenarios combine all or some of the above.
We quickly summarize what we know about blue and red populations and
discuss these properties in the frame of the different scenarios.
5.1. The Truth about blue and red populations
Are the blue and/or red clusters closely connected with the final host galaxy?
There are still too many Truths on blue and red populations. One recent topic
to review critically is the correlations of sub-population color with host galaxy
properties (Ashman & Bird 1993, Forbes et al. 1997, Burgarella et al. 2001,
Forbes & Forte 2001, Larsen et al. 2001). Only a very weak correlation seems to
exist between mean metallicity of the blue sub-population and mean metallic-
ity/size of the final host galaxy. For the red sub-population a correlation appears
more likely, although different dataset give different results. As an example we
plot in Fig. 2 (for all studies reporting a clear or likely "bimodality" in the
globular cluster color distribution) the red peak color versus galaxy metallicity
(Mg2 index). All values in the literature were used, values for a same galaxy
from different studies were averaged. The clear correlation sometimes claimed
for "cleaned" samples becomes less clear and more complicated to interpret un-
ambiguously (see Kissler-Patig et al. 2001 for more details and further samples).
The essence Despite all the uncertainties, we can retain several properties for
each sub-population.
For the blue sub-populations, the following facts are reasonably secure:
• The mean metallicity correlates only weakly if at all with the host galaxy
metallicity, and if present it is much shallower than a one-to-one correspondence.
• Several properties (sizes, high Sn, ...) fit well with the idea of them having
formed in shallow potential wells (small fragments).
• Their spatial distribution and kinematics favor them as being "halo" objects.
These clusters most probably formed in small "universal" fragments (falling
back on the Searle & Zinn (1978) idea), some of them having formed with
the dark matter potential of the galaxy (explaining a weak correlation of the
metallicities) but some having formed in satellites and having been accreted
later.
Concerning the red population, we know that:
• The red color peak correlates more or less well with the galaxy size and metal-
8
Markus Kissler-Patig
Figure 2.
Correlation between red peak color of the globular clusters
and Mg2 index of the host galaxy, for a entire sample of clear and likely
"bi-modal" systems, all values in the literature being used.
licity (although the correlation could still be driven by biases in our samples)
• The red sub-population is likely to consist of several metal-rich sub-populations,
driven by the probably complex assembly history of the galaxy.
• These metal-rich clusters must have formed in denser environments (mergers?)
and deeper potential wells (larger fragments) than the metal-poor ones, given
their sizes and the many stars associated with them.
• The red globular clusters are as old (or almost) as the blue ones.
These properties are still compatible with all red clusters having formed in
one collapse or one major merger event, but it appears more likely that several
such "major" events contributed to build up the giant systems. The key to a
discrimination will be the study of red sub-populations in order to identify or
rule out sub-populations within the metal-rich clusters.
Globular Cluster Sub-Populations
9
5.2. And the winner is ...
Is one model currently favored over another? The main problem is that no model
makes a clear, unique prediction that would allow us to rule it out or confirm
it, and as stated earlier, these models often mix a number of aspects.
Clearly, the current idea of the formation of metal-poor clusters is com-
patible with all three flavors of scenarios. Thus, at early stages (assuming that
metal-poor material is related to an early epoch, which is probably true in most
cases) all scenarios burn down to one: namely small "universal" fragments col-
lapsing and assembling to "halos" with high Sn (e.g. see Burgarella et al. 2001
for a correspondence with the high-z universe). The main open question is:
what fraction formed in clear association with the final host galaxy and what
fraction formed independently and was accreted later?
Only for the metal-rich clusters, the scenarios might differ somewhat. We
need to remember that most clusters (and spheroids) assembled as redshifts
z > 1 (probably z > 2), at which stage the large collapse leading to the formation
of the bulge might well have been induced by a gas-rich merger. The question
thus is: did the red clusters form during the single collapse of the spheroid or did
they form in multiple-collapses that assemble subsequently to form the final host
galaxy? Probably both, but which mode dominated? If the red clusters really
show distinct sub-populations, the latter, hierarchical mode, would be favored.
Acknowledgments.
I wish to thank D.Burgarella, V.Buat, T.H.Puzia,
P.Goudfrooij, C.Maraston, D.Thomas for interesting discussions that inspired
part of the subjects discussed in this contributions.
References
Ashman, K.M., & Zepf, S.E. 1992, ApJ, 384, 50
Ashman, K.M., & Bird, C.M. 1993, AJ, 106, 2281
Ashman, K.M., & Zepf, S.E. 1998, Globular Cluster Systems, Cambridge Uni-
versity Press
Baade, W. 1944, ApJ, 100, 137
Becker, W. 1950, Sterne und Sternsysteme, 2nd edition, p.166
Burgarella, D., Kissler-Patig, M., & Buat, V. 2001, AJ, 121, 2647
Cohen, J.G., Blakeslee, J.P., & Ryzhov A. 1998, ApJ, 486, 808
Forbes, D.A., Brodie, J.P., & Grillmair, C.J. 1997, AJ, 113, 1652
Forbes, D.A., & Forte, J.C. 2001, MNRAS, 322, 257
Geisler, D., Lee, M.G., & Kim, E. 1996, AJ, 111, 1529
Harris, W.E. 2001, Star Cluster, ed. L.Labhardt and B.Binggeli (New York:
Springer), 223+
Harris, G.L.H, Harris, W.E., & Poole, G.B. 1999, Aj , 117, 855
Holtzman 1992
Kissler-Patig, M. 2000, Reviews in Modern Astronomy, Vol. 13, ed. Schielicke,
13
Kissler-Patig, M., Richter, T., Storm, J., Della Valle, M. 1997, A&A, 327, 503
10
Markus Kissler-Patig
Kissler-Patig, M., Brodie, J.P., Schroder, L.L. et al. 1998, AJ, 115, 105
Kissler-Patig, M., & Gebhardt, K. 1998, AJ, 116, 2237
Kissler-Patig, M., Grillmair, C.J., & Meylan, G. et al. 1999, AJ, 117, 1206
Kundu, A., & Whitmore, B.C. 1998, AJ, 116, 2841
Larsen, S.S., Brodie, J.P., Huchra, J.P. et al. 2001, AJ in press
Lotz, J.M., Ferguson, H.C., & Bohlin, R.C. 2000, ApJ, 532, 830
Maraston, C., & Thomas, D. 2000, ApJ, 541, 126
McLaughlin, D.E. 1999, AJ, 117, 2398
Puzia, T.H., Kissler-Patig, M., Brodie, J.P., et al. 1999, AJ, 118, 2734
Searle, L., & Zinn, R. 1978, ApJ 225, 357
Zepf, S.E., & Ashman, K.M. 1993, MNRAS, 264, 611
Zepf, S.E., Beasley, M.A., Bridges, T.J. et al. 2000, AJ, 120, 2928
Zinn, R. 1985, ApJ, 293, 424
Questions
Grillmair:
(on the color of Sn) If all stars formed in clusters could differences
between globular and stellar color (or [Fe/H]) distributions simply reflect differ-
ent destruction rates? Old, metal-poor clusters have higher angular momentum
and are more protected from bulge shocking etc, and contribute few field stars.
Red clusters orbit in more violent environments, are destroyed more often and
contribute lots of field stars. Comments? Do we need more spectroscopy of field
stars?
Kissler-Patig:
I like your idea, although I cannot comment on its likelihood.
I guess that simulations of destruction rates would be "easy" if we could feed
them with realistic initial conditions. Currently, not enough is known on red
and blue cluster kinematics to make a secure guess. Another way of testing this
hypothesis would be to compare in detail the abundances and abundance ratios
in stars and clusters. This will be a by-product of T.H.Puzia's thesis, aiming at
high S/N spectroscopy of clusters and stellar light in a sample of galaxies. So
we might know more in a couple of years.
(on a similar topic) Is enough known about the dispersed pop-
Elmegreen:
ulation of stars in clusters of galaxies (e.g. color) to identify them with one
population or the other of globular clusters?
Kissler-Patig:
I am afraid that our knowledge of the dispersed stellar popu-
lation in clusters is still extremely sparse. So the answer is no. But as wide
field cameras become available, the colors of the diffuse light is certainly an
interesting way to follow-up this question.
Zepf: The width of the peaks in [Fe/H] might be dominated by observational
uncertainties.
In this case, when you take the linear errors in [Fe/H] to an
exponential to get Z, you naturally produce a peak with a tail. So I think the
inferred distribution in Z is a little hard to interpret. A critical point in this
regard that you mentioned earlier are the kinematic difference seen between the
red and blue population.
Kissler-Patig:
I agree that the photometric errors are still significantly con-
tributing to the width of the peaks and produce part of the tail in the Z distribu-
Globular Cluster Sub-Populations
11
tion. However, we conducted a few simulations that showed that observational
errors cannot explain all the dispersion.
Ideally, one wants to test this on a
sample that is not dominated by errors anymore. Such samples could either
be very high quality HST photometry, or infrared color distributions, were the
metallicity sensitivity is largely increased with respect to the photometric errors.
Such tests are currently being carried out by our group.
Whitmore:
Zepf has already made most of the point I wanted to make, but
let me add that this also shows why we tend to get a bi-modal distribution
in most cases. The metallicity enhancement happens relatively rapidly, so any
distribution in time with an initial peak at ∼ 15 Gyr will end up with a roughly
bi-modal distribution in [Fe/H]. However, I again caution about amplified noise.
If you start with 2 delta functions in age at say 15 and 10 Gyr, and then predict
[Fe/H], it would be similar to the observations. I agree with the constancy of
the blue peak and with a larger spread in the red peak, which we also discussed
in AJ, 114, 1797 (1997).
Kissler-Patig:
I think that your argument boils down to whether all globu-
lar clusters form faster than any enrichment process (in which case you would
not automatically expect a spread in Z and a bimodal [Fe/H]) or whether the
metal-rich clusters form over "longer" periods of time and can profit from the
abundance enrichment of that star-burst. But note that this latter case would be
indistinguishable observationaly from multiple metal-rich sub-population, unless
these exhibit a large spread in age (more than several Gyr). An answer might
come from the study of α-elements in metal-rich globular clusters, which probe
the timescale of formation and will reveal whether in a star-burst the clusters
formed before or after the bulk of stars (another aspect of T.H.Puzia's thesis).
Forbes: CDM models fail in many ways to reproduce the real universe. Any
comment on how much we should trust them?
Kissler-Patig:
Hierarchical clustering models might fail to reproduce some
features of the real universe, but they get quite close in describing all essential
aspects. Also, they are the best models we have so far. Of course, we can de-
sign formation scenarios for globular cluster system around hypothetic, sketchy
models (nowhere described in details, not making any concrete predictions), but
would that make more sense? I think the important point for the above discus-
sion is that i) for the blue sub-population we do not need any galaxy formation
model. We actually know enough on their properties and formation that we are
in the ideal situation of being able to constrain any galaxy formation model,
namely it should include at an early stage "universal", metal-poor fragments of
106 to 1010 solar masses. ii) concerning the red sub-population, we need to stay
open minded, but from an observational side, large structures appear to be rare
at redshifts z > 1.5, which roughly corresponds to their redshift of formation.
Thus, red clusters have likely formed in larger structures than the metal-poor
ones, but not in structures as large as their final host galaxies.
|
astro-ph/0212063 | 1 | 0212 | 2002-12-03T13:06:24 | The linear MHD Taylor-Couette instability for liquid sodium | [
"astro-ph"
] | The linear stability of MHD Taylor-Couette flow of infinite vertical extension is considered for liquid sodium with its small magnetic Prandtl number Pm of order of 10^-5 and an uniform axial magnetic field. The sign of the constant a in the basic rotation law \Omega=a+b/R^2 strongly influences the presented results. It is negative for resting outer cylinder. The main point here is that the subcritical excitation which occurs for large Pm disappears for small Pm. This is the reason that the existence of the magnetorotational instability remained unknown over decades. For rotating outer cylinder the limiting case a=0 plays an exceptional role. The hydromagnetic instability exists with certain Reynolds numbers at certain Hartmann numbers of the magnetic field. These Reynolds numbers exactly scale with Pm^-1/2 resulting in moderate values of order 10^4 for Pm=10^-5. However, already for the smallest positive value of a the Reynolds numbers start to scale as 1/Pm leading to much higher values of order 10^6 for Pm=10^-5. Also nonaxisymmetric modes have been considered. With vacuum boundary conditions their excitation is always more difficult than the excitation of axisymmetric modes; we never observed a crossover of the lines of marginal stability. For conducting walls, however, such crossovers exist for both resting and rotating outer cylinders, and this might be essential for future dynamo experiments. | astro-ph | astro-ph |
The linear MHD Taylor-Couette instability for liquid sodium
Astrophysikalisches Institut Potsdam, An der Sternwarte 16, D-14482 Potsdam, Germany†
Gunther Rudiger, Manfred Schultz, Dima Shalybkov∗
(Dated: November 29, 2018)
The linear stability of MHD Taylor-Couette flow of infinite vertical extension is considered for
liquid sodium with its small magnetic Prandtl number Pm of order of 10−5. The calculations are
performed for a container with η = 0.5 with an axial uniform magnetic field excluding counter
rotating cylinders. The sign of the constant a in the basic rotation law Ω = a + b/R2 strongly
influences the presented results. It is negative for resting outer cylinder. The main point here is
that the subcritical excitation which occurs for large Pm disappears for small Pm (cf. Fig. 3).
This is the reason that the existence of the magnetorotational instability remained unknown over
decades.
For rotating outer cylinder the limiting case a = 0 (i.e. µ = η2) plays an exceptional role.
The hydrodynamic instability starts to disappear while the hydromagnetic instability exists with
minimal Reynolds numbers at certain Hartmann numbers of the magnetic field. These Reynolds
numbers exactly scale with Pm−1/2 resulting in moderate values of order 104 for Pm=10−5. However,
already for the smallest positive value of a the Reynolds numbers start to scale as 1/Pm leading
to much higher values of order 106 for Pm=10−5. Hence, for outer cylinders rotating faster than
the limit a = 0 it is exclusively the magnetic Reynolds number Rm which directs the excitation
of the instability. They are resulting as lower for insulating walls ('vacuum') than for conducting
walls. Generally, the magnetic Reynolds numbers for liquid sodium have to exceed values of order
10 leading to frequencies of about 20 Hz for the rotation of the inner cylinder if containers with
(say) 10 cm radius are considered. The required magnetic fields are about 1000 Gauss.
Also nonaxisymmetric modes have been considered. With vacuum boundary conditions their
excitation is always more difficult than the excitation of axisymmetric modes; we never observed a
crossover of the lines of marginal stability. For conducting walls, however, such crossovers exist for
both resting and rotating outer cylinders, and this might be essential for future dynamo experiments.
In this case, however, the instability also can onset in form of oscillating axisymmetric patterns of
flow and field and the Reynolds numbers of these solutions are lower than the Reynolds numbers
for the stationary solutions.
PACS numbers: 47.20.Ft, 47.20.-k, 47.65.+a
I.
INTRODUCTION
The longstanding problem of the generation of turbu-
lence in various hydrodynamically stable situations has
found a solution in recent years with the MHD shear
flow instability, also called magnetorotational instabil-
ity (MRI), in which the presence of a magnetic field
has a destabilizing effect on a differentially rotating flow
with the angular velocity decreasing outwards. The MRI
has been formulated decades ago [1, 2] for ideal Taylor-
Couette flow, but its importance as the source of tur-
bulence in accretion discs with differential (Keplerian)
rotation was first recognized by Balbus and Hawley, [3].
However, the MRI has never been observed in the lab-
oratory [4, 5, 6, 7]. After Goodman and Ji [8] was the
absence of MRI due to the small magnetic Prandtl num-
ber approximation used in [2]. The magnetic Prandtl
number Pm is really very small under laboratory condi-
tions (∼ 10−5 and smaller, see Table I).
A proper understanding of this phenomenon is very im-
TABLE I: Parameters of the fluids suitable for MHD experi-
ments taken from [2] and [9]
ρ [g/cm3] ν [cm2/s] η [cm2/s]
Pm
Mercury
Gallium
5.4
6.0
Sodium
0.92
1.1·10−3
3.2·10−3
7.1·10−3
7600
2060
810
1.4·10−7
1.5·10−6
0.88·10−5
portant for possible future experiments, including Taylor-
Couette flow dynamo experiments. The simple model of
uniform density fluid contained between two vertically-
infinite rotating cylinders is used with constant magnetic
field parallel to the rotation axis. For viscous flows the
most general form of the rotation law Ω(R) in the fluid
is
Ω(R) = a + b/R2,
(1)
∗permanent address: A.F. Ioffe Institute for Physics and Technol-
ogy, 194021, St. Petersburg, Russia
†Electronic address: [email protected]
where a and b are two constants related to the angular
velocities Ωin and Ωout with which the inner and the outer
cylinders are rotating and R is the distance from the
rotation axis. If Rin and Rout (Rout > Rin) are the radii
B
Ω
out
Ω
in
R
out
inR
FIG. 1: Cylinder geometry of the Taylor-Couette flow with
axial magnetic field.
of the two cylinders then
a =
µ − η2
1 − η2 Ωin
and
b = R2
in
1 − µ
1 − η2 Ωin
(2)
with
µ = Ωout/Ωin
and
η = Rin/Rout.
(3)
Following the Rayleigh stability criterion, d(R2Ω)2/dR >
0, rotation laws are hydrodynamically stable for a > 0,
i.e. µ > η2. They should in particular be stable for rest-
ing inner cylinder, i.e. µ → ∞. Richard and Zahn [10] fo-
cused attention to the experimental results of Wendt [11]
who found nonlinear instability for this case for Reynolds
numbers of order 105. The finite-amplitude instability of
hydrodynamically stable rotation laws must therefore re-
main in the astrophysical discussion. However, later ex-
periments with very similar Taylor-Couette flow experi-
ments for resting inner cylinder demonstrated the results
of Wendt as due to rather imperfect container construc-
tions and the flow remained laminar even for Reynolds
numbers up to 106, [12].
One of the targets in the present paper is the axisym-
metry of the excited modes. We have shown in [13] that
for containers with conducting boundaries it happens for
sufficiently strong magnetic fields that the mode with the
lowest eigenvalue (i.e. the lowest Reynolds number) is a
nonaxisymmetric mode. As an impressive example, in
Fig. 2 for Pm=0.01 the crossover of the instability lines
for axisymmetric and nonaxisymmetric modes is shown
for Hartmann numbers of about 400 (see [20]). Despite of
its general meaning this behavior is only known so far for
conducting walls and for magnetic Prandtl numbers not
2
FIG. 2: Instability lines for axisymmetric (solid) and nonax-
isymmetric modes (m=1, dashed line) for conducting walls
and Pm=0.01 (η = 0.5).
smaller than 10−2 (see [14]). For possible laboratory ex-
periments we have to extend, however, the computations
to insulating boundaries (vacuum) and to much smaller
magnetic Prandtl numbers Pm.
The equations, therefore, are solved here mainly for the
single small magnetic Prandtl number Pm = 10−5 very
close to the value for liquid sodium (see Table I). The
aspect ratio of the container walls radii in the present
paper is fixed to η = 0.5.
II. BASIC EQUATIONS
The MHD equations which have to be solved are
∂u
∂t
+ (u∇)u = −
1
ρ
∇p + ν∆u + J × B
(4)
and
∂B
∂t
= curl(u × B) + η∆B,
(5)
with the electric current J = curlB/µ0 and divu =
divB = 0. They are considered in cylindrical geometry
with R, φ, and z as the cylindrical coordinates. A viscous
electrically-conducting incompressible fluid between two
rotating infinite cylinders in the presence of a uniform
magnetic field parallel to the rotation axis leads to the
basic solution UR = Uz = BR = Bφ = 0, Bz = B0 =
const., and Uφ = aR + b/R, with U as the flow and B
as the magnetic field. We are interested in the stability
of this solution. The perturbed state of the flow may be
described by u′
z, p′ with p′ as the
pressure perturbation.
R, B ′
φ, B ′
R, u′
φ, u′
z, B ′
Here only the linear stability problem is considered.
By analyzing the disturbances into normal modes the
3
solutions of the linearized hydromagnetic equations are
of the form
B ′ = B(R) ei(mφ+kz−ωt),
u′ = u(R) ei(mφ+kz−ωt).
(6)
From hereon all dashes have been omitted from the no-
tations of fluctuating quantities. Only marginal stability
will be considered hence the imaginary part of ω, i.e.
I(ω), always vanishes. We use
∂2Bφ
∂R2 +
2im
+
1
R
∂Bφ
∂R
−
Bφ
R2
R2 BR − iPm(cid:18)mRe
−(cid:18) m2
R2 + k2(cid:19) Bφ+
− ω(cid:19) Bφ + ikuφ +
Ω
Ωin
+Pm Re R
BR = 0.
(13)
∂Ω/Ωin
∂R
Here the Reynolds number Re and the Hartmann number
Ha are defined as
H =pRin(Rout − Rin)
as the unit of length, the η/H as the unit of perturbed
velocity and B0 as the unit of perturbed magnetic field
and work with the magnetic Prandtl number
(7)
Re =
ΩinRin(Rout − Rin)
ν
and
Pm =
ν
η
,
(8)
Ha = B0s Rin(Rout − Rin)
µ0ρνη
.
(14)
(15)
with ν the kinematic viscosity and η the magnetic dif-
fusivity. Note H −1 also as the unit of wave numbers
and ν/H 2 as the unit of frequencies. After elimination
of both pressure fluctuations and the fluctuations of the
vertical magnetic field, B ′
z, the linearized equations are
im
R
uφ + ikuz = 0,
∂uR
∂R
uR
R
(9)
+
+
R2 + k2(cid:19) uφ−
2im
R2 uR − Re
−(cid:18) m2
1
R
−
uφ
R2
−(cid:18) m2
−
m
1
R
∂uφ
∂R
Ω
Ωin
∂2uφ
∂R2 +
−i(cid:18)mRe
k (cid:20) 1
−i(cid:18)mRe
Ha2(cid:18) m2
i
k
+
R
− ω(cid:19) uφ +
∂2uz
∂R2 +
Ω
Ωin
1
R2
∂uz
∂R
m
k
− ω(cid:19) uz
R(cid:21) +
R2 + k2(cid:19) Bφ = 0,
R2 + k2(cid:19) uz
Ha2(cid:20) 1
R
∂BR
∂R
R
∂
∂R(cid:18)R2 Ω
Ωin(cid:19) uR
−
+
BR
R2(cid:21) +
(10)
+
−
1
R
∂2uz
∂R2
∂3uz
∂R3 +
2m2
R3 uz − i(cid:18)mRe
−Ha2(cid:20) ∂2BR
∂R2 +
im
1
R2
Ω
Ωin
1
R
+
im
R
∂Bφ
∂R
∂uz
∂R
−(cid:18) m2
− ω(cid:19) ∂uz
R2 + k2(cid:19) ∂uz
∂R
∂
− imRe
+
∂R(cid:18) Ω
Ωin(cid:19) uz
∂R
BR
R2
∂BR
∂R
−
− k2BR+
−
R2 Bφ(cid:21) − ik(cid:20) ∂2uR
∂R2 +
R2(cid:19) uR(cid:21) − k(cid:18)mRe
Ω
Ωin
m2
−(cid:18)k2 +
km
R2 uφ − 2ikRe
−2
Ω
Ωin
1
R
∂uR
∂R
−
uR
R2
−
− ω(cid:19) uR −
uφ = 0,
(11)
1
R
∂BR
∂R
−
BR
R2
∂2BR
∂R2 +
2im
−
R2 Bφ − iPm(cid:18)mRe
Ω
Ωin
−(cid:18) m2
R2 + k2(cid:19) BR−
− ω(cid:19) BR + ikuR = 0,(12)
For given Hartmann number and magnetic Prandtl num-
ber in the present paper we shall compute with a linear
theory the critical Reynolds number of the rotation of
the inner cylinder, also for various mode numbers m.
III. BOUNDARY CONDITIONS, NUMERICS
An appropriate set of ten boundary conditions is
needed to solve the system (9). . .(13). Always no-slip
conditions for the velocity on the walls are used,
i.e.
uR = uφ = duR/dR = 0. The boundary conditions for
the magnetic field depend on the electrical properties of
the walls. The tangential currents and the radial com-
ponent of the magnetic field vanish on conducting walls
hence dBφ/dR + Bφ/R = BR = 0. These boundary con-
ditions may also hold both for R = Rin and for R = Rout.
The homogeneous set of equations (9). . .(13) together
with the boundary conditions determine the eigenvalue
problem of the form L(k, m, Re, Ha, R(ω)) = 0 for given
Pm. The real part of ω, i.e. R(ω), describes a drift
of the pattern along the azimuth which only exists for
nonaxisymmetric flows. For axisymmetric flows (m =
0) the real part of ω, i.e. R(ω), is zero for stationary
patterns of flow and field and it is nonzero for oscillating
solutions, which are called overstability. L is a complex
quantity, both its real part and its imaginary part must
vanish for the critical Reynolds number. The latter is
minimized by choice of the wave number k. R(ω) is the
second quantity which is fixed by the eigen equation.
The system is approximated by finite differences with
typically 200 radial grid points. The resulting determi-
nant, L, takes the value zero if and only if the values Re
are the eigenvalues. We can also stress that the results
are numerically robust as an increase of the number of
grid points does not change the results remarkably. For
a fixed Hartmann number, a fixed Prandtl number and a
given vertical wave number k we find the eigenvalues of
the equation system. They are always minimal for a cer-
tain wave number which by itself defines the marginally
4
unstable mode. The corresponding eigenvalue is the de-
sired Reynolds number.
The situation changes for insulating walls. The mag-
netic field must match the external magnetic field for
vacuum.
It is known for this case that the boundary
conditions for axisymmetric solutions strongly differ from
those for nonaxisymmetric solutions (see [15]). The con-
dition curlRB = 0 in vacuum immediately provides
Bφ =
m
kR
Bz
(16)
at R = Rin and R = Rout. From the solution of the
potential equation ∆ψ = 0 one finds
BR +
iBz
Im(kR)(cid:16) m
kR
Im(kR) + Im+1(kR)(cid:17) = 0
(17)
for R = Rin and
BR +
iBz
Km(kR)(cid:16) m
kR
Km(kR) − Km+1(kR)(cid:17) = 0 (18)
for R = Rout. Im and Km are the modified Bessel func-
tions (with different behavior at R → 0 and R → ∞).
One can eliminate with divB=0 the vertical component
Bz of the magnetic field in the boundary conditions
(16)...(18).
IV. RESULTS
The following results concern different aspects of
the MHD Taylor-Couette problem for small magnetic
Prandtl number Pm. In Section A the main realization
of the case a < 0 (here with resting outer cylinder, i.e.
µ = 0) is considered. There is instability even without
magnetic fields so that the bifurcation lines start at the
y-axis. In Section B the special case a = 0 is considered
with very surprising results. The Section C presents the
results for the two experiments with µ = 0 and µ = 0.33
with respect to the axisymmetry of the eigenmodes. In
Section D the existence of oscillating modes is discussed,
i.e. the case of overstability for small magnetic Prandtl
numbers.
A. Subcritical excitation for large Pm (a < 0)
Figure 3 shows the stability lines for axisymmetric
modes for containers with conducting walls and with rest-
ing outer cylinder for fluids of various magnetic Prandtl
number. Only the vicinity of the classical hydrodynamic
solution with Re = 68 is shown. There is a strong dif-
ference of the geometry of the bifurcation lines for Pm
>
∼ 1 and Pm < 1. In the latter case, for fluids with low
electrical conductivity the magnetic field only suppresses
the instability so that all the critical Reynolds numbers
exceed the value 68, and this the more the stronger the
magnetic field is.
FIG. 3: Marginal stability lines for axisymmetric modes with
resting outer cylinder of conducting material. The shaded
area denotes subcritical excitations of unstable axisymmetric
modes by the external magnetic field.
For sufficiently small magnetic Prandtl number the sta-
bility lines hardly differ, which is the situation already
considered by Chandrasekhar [2] without any indication
of magnetorotational instability.
The opposite is true for Pm >
∼ 1. Note that in Fig. 3
for materials with high electrical conductivity the result-
ing critical Reynolds numbers are smaller than Re = 68.
The magnetic field with small Hartmann numbers sup-
port instability patterns rather than to suppress them.
This effect becomes more effective for increasing Pm but
it vanishes for stronger magnetic fields. Obviously, the
MRI only exists for weak magnetic fields and high enough
electrical conductivity and/or molecular viscosity (when
the fields can be considered as frozen in and/or enough
viscosity prevents the action of the Taylor-Proudman the-
orem).
Note that the subcritical excitation of Taylor vortices
only works for weak magnetic fields. The upper limits
of the possible Hartmann numbers can be observed for
the magnetic Prandtl numbers 1 and 10 in Fig. 4. After
our computations the subcritical excitation of Taylor vor-
tices for weak magnetic fields requires rather high mag-
netic Prandtl numbers. The microscopic values for Pm
are orders of magnitudes smaller than unity, so that there
should be no chance to realize the subcritical excitation of
Taylor vortices by experiments. However, the speculation
may be allowed whether really the microscopic Pm is the
the magnetic Reynolds
for
ΩinRin(Rout − Rin)/η and
number Rm
resulting in
Ha ∝ Pm−1/2
Ha∗ ∝ const.
5
=
(23)
(24)
for Ha∗ = B0pRin(Rout − Rin)/µ0ρη2 (Lundquist num-
ber, see [18]). In case of small magnetic Prandtl num-
ber the exact value of the microscopic viscosity is totally
unimportant for the excitation of the instability. In con-
sequence, however, the corresponding Reynolds numbers
for the MRI seem to differ by 2 orders of magnitude, i.e.
104 and 106. Insofar, experiments with µ = η2 seem to
look much more promising than experiments with µ > η2.
Unfortunately, this challenging possibility cannot be
utilized in experiments. The critical Reynolds number
for µ = η2 and Pm=1 as a function of η is given in Fig.
4. The total minimum of the Reynolds number is 54.4 for
η = 0.27 so that after (20) one expects the value 1.7·104
for the Reynolds number for Pm= 10−5. Fig. 5 shows
the behavior of this result in the vicinity of µ = η2. There
is a vertical jump from 104 to 106 in an extremely small
interval of the abscissa. This sharp transition does not
exist for Pm=1, it is only due to the very small value
of Pm. For this case in Fig. 5 the coexistence of both
hydrodynamic and hydromagnetic instability is also pre-
sented. The jump profile for Pm= 10−5 in Fig. 5 (right)
FIG. 5: Critical Reynolds numbers for the Taylor-Couette
flow versus µ for η = 0.27 and Pm=1 (left) and Pm=10−5
(right). The curve for the hydrodynamic instability (Ha=0)
is dashed and the hydromagnetic curve (Ha>0) is solid. The
dotted line denotes the location of a=0, i.e. µ = η2.
FIG. 4: Critical Reynolds number versus η for µ = η2 and
Pm= 1.
basic input. The scenario is also interesting whether pos-
sible finite-amplitude hydrodynamic instabilities provide
some kind of background turbulence which can be con-
sidered as modifying the value of the magnetic Prandtl
number, [9]. The turbulence influences both the viscosity
values and the magnetic-diffusivity values so that
Pm → Pmeff =
ν + νT
η + ηT
≃
νT
ηT
(19)
with νT and ηT as the eddy viscosity and the eddy diffu-
sivity, resp. Because of the existence of the pressure term
in the momentum equation, both quantities are not iden-
tical. We do not have precise knowledge about the effec-
tive turbulent magnetic Prandtl number but it has been
demonstrated that values of order 0.1 or somewhat larger
should not be unlikely, [16]. Insofar if such speculations
are not too far from the reality, it is not completely clear
that the subcritical excitation of Taylor vortices which
we have presented in Fig. 4 is unobservable in general.
B. The case a = 0 ( µ = η 2)
There is a universal scaling on Pm for the special case
with a = 0 in the basic flow profile (1), i.e. for µ = η2.
Then the term with ∂(R2Ω)/∂R in Eq. (10) vanishes and
for m = ω = 0 one finds that the quantities uR, uz, BR
and Bz are scaling as Pm−1/2 while uφ, Bφ, k Ha scale
as Pm0. Then also the Reynolds number for the axisym-
metric modes scales as
Re ∝ Pm−1/2.
(20)
The scaling does not depend on the boundary conditions
as these for m = 0 also comply with the relations.
The result (20) has numerically been found by Willis
and Barenghi for vacuum boundary conditions,
[17].
However, Rudiger and Shalybkov [18] for a > 0 (µ > η2)
found the much steeper scaling
Re ∝ Pm−1,
(21)
resulting in the surprisingly simple relation
Rm ∝ const.
(22)
makes it clear that such experiments with µ = η2 are not
6
TABLE II: Coordinates of the absolute minima in Figs. 6 and
7 for rotating outer cylinder (µ = 0.33)
conducting walls insulating walls
Reynolds number
2.13 · 106
1.42 · 106
magnetic Reynolds number
Hartmann number
Lundquist number
21
1100
3.47
14
1400
4.42
possible. Even the smallest deviation from the condi-
tion µ = η2 drastically changes the excitation condition.
For µ smaller than η2 (negative deviations) the hydro-
dynamic instability sets in and for µ slightly exceeding
η2 (positive deviations) the Reynolds number suddenly
jumps by two orders of magnitudes.
C. Excitation of nonaxisymmetric modes
FIG. 7: Conducting walls: Stability lines for axisymmetric
(m = 0, solid lines) and nonaxisymmetric instability modes
with m = 1 (dashed). LEFT: resting outer cylinder, RIGHT:
rotating outer cylinder (33 %).
instabilities even without magnetic fields. For Ha= 0 so-
lutions for m = 0 (Re= 68) and m = 1 (Re= 75) are
known, see [13]. The axisymmetric mode possesses the
lowest eigenvalue. This is also true for nonvanishing mag-
netic field; we do not find any crossover of the instability
lines for axisymmetric and nonaxisymmetric modes. The
same is true for containers with rotating outer cylinder
(Fig. 6b). For growing µ the Reynolds number for the
hydrodynamic solution moves upwards, reaching infin-
ity for µ = η2 = 0.25 (here). The MRI is represented
by characteristic minima, in our case for µ = 0.33 at
Hartmann numbers of order 103 and Reynolds numbers
of order 106. The exact coordinates of the minima are
given in Table II.
In order to characterize the Hartmann numbers note
that for liquid sodium
B = 2.2
Ha
Rout/10cm
Gauss.
(25)
Hence, for Rout ≃ 22 the magnetic field and the Hart-
mann number have the same numerical values. With
ν = 10−2 cm2/s and η=0.5 it follows from (14) and (15)
fin = 64
Re/106
(Rout/10cm)2 Hz
(26)
for the frequency of the inner cylinder. Hence, a con-
tainer with an outer radius of 22 cm (see above) and an
FIG. 6: Insulating walls (vacuum): Stability lines for axisym-
metric (m = 0, solid lines) and nonaxisymmetric instability
modes with m = 1 (dashed). LEFT: resting outer cylinder,
RIGHT: rotating outer cylinder (33%).
Let us now concentrate to the small magnetic Prandtl
number for liquid sodium, i.e. Pm = 10−5. We start
with the results for containers with insulating walls and
outer cylinders at rest, (Fig. 6a). There are then linear
inner radius of 11 cm filled with liquid sodium and em-
bedded in vacuum requires a rotation of about 19 Hz in
order to find the MRI. Following (25) the required mag-
netic field is about 1400 Gauss.
FIG. 8: The same as in Fig. 7a but with the inclusion of
oscillating axisymmetric modes (overstability) appearing here
for lower Reynolds numbers.
The results for containers with conducting walls are
given in Fig. 7. Note that the minimal Reynolds numbers
7
given in Fig. 7b are higher than for insulating cylinder
walls. The influence of the boundary conditions is not as
small as expected. The main difference between the two
sorts of boundary conditions, however, is the existence of
crossovers of the instability lines for m = 0 and m = 1 in
case of conducting walls. For both resting and rotating
outer cylinder Hartmann numbers exist above which the
nonaxisymmetric mode possesses a lower Reynolds num-
ber than the axisymmetric mode. We have already shown
the existence of such crossovers for conducting walls for
1 ≤ Pm ≤ 0.01 in [13]. It is now clear that the occurrence
of nonaxisymmetric solutions as the preferred modes is a
rather general phenomenon for containers with conduct-
ing walls which can become important for the design of
future dynamo experiments (Cowling theorem).
D. Excitation of oscillating modes
There are not only stationary patterns of flow and field
possible but the instability can also onset in form of os-
cillating solutions. This effect is called overstability. In
case of rotating convection between two layers heated
from below the onset of the instability in form of os-
cillating solutions even possesses the lowest eigenvalues
for certain Prandtl numbers, [2]. We find a very simi-
lar behavior for the MHD Taylor-Couette flow between
conducting cylinders for resting outer cylinder (see Fig.
8). It is a pair of waves traveling in positive and negative
z-direction. Note that the cylinder considered here has
no bound in vertical direction. If the cylinder is finite,
however, the possibility exists that traveling waves might
be combined to standing waves.
[1] E. P. Velikhov, Sov. Phys. JETP 9, 995 (1959).
[2] S. Chandrasekhar, Hydrodynamic and Hydromagnetic
Stability (Clarendon, Oxford, 1961).
[3] S. A. Balbus and J. F. Hawley, Astrophys. J. 376, 214
(1991).
[13] D. A. Shalybkov, G. Rudiger and M. Schultz, Astron.
Astrophys. 395, 339 (2002).
[14] which is unknown in the hydrodynamic regime, counter
rotation excluded, see E. R. Kruger, A. Gross and R. C.
DiPrima, J. Fluid Mech. 24, 521 (1966).
[4] R. J. Donnelly and M. Ozima, Phys. Rev. Lett. 4, 497
[15] D. Elstner, R. Meinel and G. Rudiger, Geophys. Astro-
(1960).
[5] R. J. Donnelly and M. Ozima, Proc. R. Soc. Lond. A
266, 272 (1962).
[6] R. J. Donnelly and D. R. Caldwell, J. Fluid. Mech. 19,
257 (1964).
[7] A. Brahme, Physica Scripta 2, 108 (1970).
[8] J. Goodman and H. Ji, J. Fluid Mech. 462, 365 (2002).
[9] K. Noguchi, V. I. Pariev, S. A. Colgate, H. F. Beckley
phys. Fluid Dyn. 50, 85 (1990).
[16] G. Rudiger, Differential Rotation and Stellar Convection:
Sun and Solar-Type Stars (Gordon & Breach Science
Publishers, New York, 1989).
[17] A. P. Willis and C. F. Barenghi, Astron. Astrophys. 388,
688 (2002).
[18] G. Rudiger and D. A. Shalybkov, Phys. Rev. E 66,
016307 (2002).
and J. Nordhaus, Astrophys. J. 575, 1151 (2002).
[19] C.-K. Chen and M. Chang, J. Fluid Mech. 366, 135
[10] D. Richard and J.-P. Zahn, Astron. Astrophys. 347, 734
(1998).
(1999).
[11] F. Wendt, Ing. Arch. 4, 577 (1933).
[12] F. Schultz-Grunow, ZAMM 39, 101 (1959).
[20] Note that we considered in [13] only the excitation of the
instability for Hartmann number smaller than about 100
|
astro-ph/0204452 | 3 | 0204 | 2003-01-27T22:05:05 | Galactic halo cusp-core: tidal compression in mergers | [
"astro-ph"
] | We explain in simple terms how the buildup of dark haloes by merging compact satellites, as in the CDM cosmology, inevitably leads to an inner cusp of density profile $\rho \propto r^{-\alpha}$ with $\alpha \gsim 1$, as seen in cosmological N-body simulations. A flatter halo core with $\alpha <1$ exerts on the satellites tidal compression in all directions, which prevents deposit of stripped satellite material in the core region. This makes the satellite orbits decay from the radius where $\alpha \sim 1$ to the halo centre with no local tidal mass transfer and thus causes a rapid steepening of the inner profile to $\alpha >1$. These tidal effects, the resultant steepening of the profile to a cusp, and the stability of this cusp to tandem mergers with compact satellites, are demonstrated using N-body simulations. The transition at $\alpha \sim 1$ is then addressed using toy models in the limiting cases of impulse and adiabatic approximations and using tidal radii for satellites on radial and circular orbits. In an associated paper we address the subsequent slow convergence from either side to an asymptotic stable cusp with $\alpha \gsim 1$. Our analysis thus implies that an inner cusp is enforced when small haloes are typically more compact than larger haloes, as in the CDM scenario, such that enough satellite material makes it intact into the inner halo and is deposited there. We conclude that a necessary condition for maintaining a flat core, as indicated by observations, is that the inner regions of the CDM satellite haloes be puffed up by about 50% such that when they merge into a bigger halo they would be disrupted outside the halo core. This puffing up could be due to baryonic feedback processes in small haloes, which may be stimulated by the tidal compression in the halo cores. | astro-ph | astro-ph | Mon. Not. R. Astron. Soc. 000, 1 -- 19 (2002)
Printed 8 November 2018
(MN LATEX style file v2.2)
Galactic halo cusp-core: tidal compression in mergers
Avishai Dekel, Jonathan Devor & Guy Hetzroni
Racah Institute of Physics, The Hebrew University, Jerusalem 91904, Israel
8 November 2018
ABSTRACT
We explain in simple terms how the buildup of dark haloes by merging compact
satellites, as in the CDM cosmology, inevitably leads to an inner cusp of density profile
ρ ∝ r−α with α >
∼ 1, as seen in cosmological N-body simulations. A flatter halo core
with α < 1 exerts on the satellites tidal compression in all directions, which prevents
deposit of stripped satellite material in the core region. This makes the satellite orbits
decay from the radius where α ∼ 1 to the halo centre with no local tidal mass transfer
and thus causes a rapid steepening of the inner profile to α > 1. These tidal effects, the
resultant steepening of the profile to a cusp, and the stability of this cusp to tandem
mergers with compact satellites, are demonstrated using N-body simulations. The
transition at α ∼ 1 is then addressed using toy models in the limiting cases of impulse
and adiabatic approximations and using tidal radii for satellites on radial and circular
orbits. In an associated paper we address the subsequent slow convergence from either
side to an asymptotic stable cusp with α >
∼ 1. Our analysis thus implies that an inner
cusp is enforced when small haloes are typically more compact than larger haloes, as
in the CDM scenario, such that enough satellite material makes it intact into the inner
halo and is deposited there. We conclude that a necessary condition for maintaining a
flat core, as indicated by observations, is that the inner regions of the CDM satellite
haloes be puffed up by about 50% such that when they merge into a bigger halo they
would be disrupted outside the halo core. This puffing up could be due to baryonic
feedback processes in small haloes, which may be stimulated by the tidal compression
in the halo cores.
Key words: cosmology: theory -- dark matter -- galaxies: formation -- galaxies:
haloes -- galaxies: interactions -- galaxies: structure
3
0
0
2
n
a
J
7
2
3
v
2
5
4
4
0
2
0
/
h
p
-
o
r
t
s
a
:
v
i
X
r
a
1
INTRODUCTION
Cosmological N-body simulations of dissipationless hierar-
chical clustering from Gaussian initial fluctuations reveal a
relatively robust universal shape for the density profile of
dark-matter haloes,
ρ(r) = ρs (cid:16) r
rs(cid:17)−α0 (cid:16)1 +
r
rs(cid:17)α0−3
,
(1)
where rs is a characteristic inner radius and ρs a correspond-
ing inner density.1 It has an inner "cusp" ∝ r−α0 , a turn-over
near rs, and an outer envelope of r−3 (or perhaps somewhat
steeper) extending out to the virial radius Rv.2 Navarro,
Frenk & White (1995; 1996; 1997, hereafter NFW) found
1 a useful generalized functional form has been proposed by Zhao
(1996)
2 The virial radius is defined by a fixed mean overdensity ∆v
above the universal mean, with ∆v = 180 to 340, depending on
time and the cosmological model.
eq. (1) with α0 ≃ 1 to be a good fit to haloes in simula-
tions over the radius range (0.01 − 1)Rv, for a wide range
of halo masses and for a range of hierarchical cosmological
scenarios with different power spectra of initial fluctuations.
Cole & Lacey (1996) came to a similar conclusion for self-
similar scenarios with power-law power spectra, Pk ∝ kn
with n = 0, −1, −2, in an Einstein-deSitter cosmology. High-
resolution simulations of a few individual haloes in a cosmo-
logical environment (Moore et al. 1998; Ghigna et al. 2000;
Klypin et al. 2001) found that the typical asymptotic cusp
profile at r ≪ rs is sometimes somewhat steeper, closer to
α0 ≃ 1.5. A careful convergence analysis by Power et al.
(2002), who explored the robustness to numerical errors,
found for the standard ΛCDM cosmology that α0 reaches
a slope shallower than 1.2 at their innermost resolved point
of r ∼ 0.005Rv. Given these uncertainties regarding the ex-
act slope and its extent, the simulations have established the
formation of a cusp with a characteristic slope in the range
1 6 α0 6 1.5. However, a basic theoretical understanding of
its origin is still lacking.
2
A. Dekel, J. Devor & G. Hetzroni
An even more intriguing puzzle is introduced by obser-
vations of low surface-brightness (LSB) or dwarf galaxies,
whose centres are dominated by their dark matter haloes,
which indicate that at least in some cases their actual inner
halo density profiles are close to flat cores, with α0 ≃ 0 (e.g.,
van den Bosch et al. 2000; de Block et al. 2001; Marchesini
et al. 2002). Similar cores may be present in other galaxies
(Salucci & Burkert 2000; Salucci 2001; Borriello & Salucci
2001). This seems to introduce a severe challenge to the
CDM cosmological paradigm. Attempts to turn a cusp into
a core by direct stellar feedback effects in the present large
haloes, which looked promising at a first sight (Navarro, Eke
& Frenk 1996), seem not to work in a straightforward way
(e.g., Geyer & Burkert 2001; Gnedin & Zhao 2002).
In order to make progress in resolving the core problem,
we find it useful to first try to understand in simple basic
terms the origin of the universal cusp in the gravitational N-
body simulations of cold dark matter. This should provide
us with a tool for addressing the formation and survival
of flat cores by some non-gravitational mechanism, perhaps
by baryonic feedback processes still within the hierarchical
CDM framework.
We now return to the issue of dark-halo profiles in dissi-
pationless simulations. The outer slope of r−3 (and steeper)
may possibly be explained in terms of violent relaxation
(e.g., Barnes & Hernquist 1991; Pearce, Thomas & Couch-
man 1993 and references therein). In general, any finite sys-
tem would tend to have a steep density fall off at large radii
due to diffusion of particles outwards. Secondary spherical
infall is expected to produce a profile closer to the profile
of an isothermal sphere, ρ ∝ r−2, which may explain the
behavior in the intermediate regions of the halo, but is too
steep to explain the flatter inner cusp of α 6 1.5 (Lokas
& Hoffman 2000 and references therein). Thus, none of the
above mechanisms provides a natural explanation for the
characteristic cusp of α0
>
∼ 1.
By following the evolution in cosmological N-body sim-
ulations of hierarchical clustering scenarios, we know that
haloes are largely built up by a sequence of mergers of
smaller haloes (e.g., Klypin et al. 1999b; Moore et al. 1999a;
Springel et al. 2001).3 In a typical merger, a bound satellite
halo spirals into the centre of the larger halo due to grav-
ity and dynamical friction. The satellite gradually transfers
mass into the host halo due to tidal stripping or by eventu-
ally melting into the halo inner region. This process is likely
to have an important effect in shaping up the density pro-
file. Indeed, Syer & White (1998), Nusser & Sheth (1999)
and Subramanian, Cen & Ostriker (2000) argued, using cer-
tain simple models and simulations of a sequence of mergers,
that the buildup by mergers may naturally lead to a stable
profile. However, they find their predicted profile to be quite
sensitive to the power spectrum of fluctuations and to allow
an inner slope of α0 < 1, in conflict with the robust result of
the cosmological simulations. In fact, when trying to repeat
the Syer & White analysis using their simplified modeling
of the stripping process but with higher resolution, we find
that in the long run the profile does not really converge to a
stable cusp but rather continues to steepen slowly towards
α0 = 3. Either way, it seems that something is not adequate
in the simplified model adopted to describe the mass transfer
from the satellite to the halo in these studies.
We re-visit here the buildup of halo profile by merg-
ing satellites and gain an encouraging new insight. We add
two important new ingredients to the tidal effects. In the
current paper, we argue that for a flat mean halo density
profile with a logarithmic slope α 6 1 the tidal effects on
typical satellites induce three-dimensional compression with
no local mass deposit, which results in a rapid steepening of
the inner profile to α > 1. In an associated paper (Dekel et
al. 2003) we derive a useful prescription for tidal mass trans-
fer at α > 1, and obtain higher deposit efficiency at higher
α. We then show that this tends to flatten steep profiles with
large α and thus slowly leads to an asymptotic fixed point
at a certain α = αa
>
∼ 1.
The key idea in the current paper is that the local
tidal mass transfer from the merging satellite to the host
halo practically stops when the satellite's orbit has decayed
into a core region where the local logarithmic slope is flat,
α(r) 6 1. As a result, the satellite continues to sink in due
to dynamical friction without mass loss until it settles in the
halo centre. This inevitably causes a general steepening of
the core profile towards α > 1, thus explaining the asymp-
totic inner slope of the cusp in the NFW profile as seen in
the cosmological simulations. We demonstrate this behav-
ior using merger N-body simulations and interpret it using
crude analytic approximations in idealized cases.
In §2 we address the α dependence of the tidal force and
highlight the three-dimensional tidal compression in a core
of α < 1. In §3 we describe the N-body merger simulations.
In §4 we analyze the evolution of halo density profile in these
simulations, where the small satellite is more compact than
the more massive halo, as in a typical hierarchical clustering
scenario (e.g. ΛCDM). In §5 we qualitatively evaluate the
transition in mass-transfer efficiency at α ∼ 1 in the extreme
limits of impulse and adiabatic tidal effects along typical
satellite orbits. In §6 we continue to study the α dependence
of the mass transfer using the tidal radii for mergers on
radial and circular orbits. In §7 we summarize the analysis
>
leading from either side to an asymptotic cusp α → αa
∼ 1.
In §8 we demonstrate how the change from a core to a cusp
could be avoided when the merging satellites are slightly
puffed up, and discuss a possible scenario for puffing up
small haloes by baryonic feedback, perhaps assisted by tidal
compression. In §9 we discuss our results and put them in a
broader perspective.
3 Although claims were made that a substantial fraction of the
mass may actually be added by smooth accretion rather than by
mergers, a careful inspection of high resolution CDM simulations
shows that this is a wrong impression arising from difficulties
in identifying the merging clumps in many cases (e.g., Wechsler,
Dekel et al. , in preparation).
2 TIDAL FORCE: COMPRESSION AND
STRETCHING
A straightforward analysis indicates that the tidal forces ex-
erted by a dark halo on a satellite orbiting in it depend not
Halo cusp-core: tides in mergers
3
Figure 1. An illustration of the tidal acceleration at a given satellite shell of radius ℓ ≪ r, for a given ¯ρ(r) and for different values
of halo density-profile slope α [eq. (5) or eq. (6)]. Along the line connecting the centres of mass (ℓℓℓ1 = r) the tides stretch outwards for
α > 1 and compress inwards for α < 1. Along the perpendicular directions, ℓℓℓ2 or ℓℓℓ3, there is always compression, independent of α.
only on the distance from the halo centre and the halo mass
within that radius, but also on the local slope of the halo
density profile. This dependence, which is often overlooked,
affects the way mass is transfered from the satellite to the
halo, and may provide a clue for the origin of the character-
istic cusp of α > 1 in haloes.
We consider a spherical halo of mass profile M (r) and
virial mass Mv. The mean density in a sphere of radius r
is ¯ρ(r) ∝ M (r)/r3. A useful quantity in describing the tidal
forces exerted by this halo is its local logarithmic slope,
rotating)4 rest frame of the satellite, namely, by subtracting
the acceleration of the satellite centre of mass relative to the
halo,
Ft = −
GM (r + ℓℓℓ) (r + ℓℓℓ)
r + ℓℓℓ3
+
GM (r) r
r3
.
In the tidal limit ℓ ≪ r, this yields to first order in ℓ/r
Ft =
Gℓ
r3 (cid:16)[3M (r) − M ′(r) r] (r · ℓℓℓ) r − M (r) ℓℓℓ(cid:17) ,
(3)
(4)
α(r) ≡ −
d ln ¯ρ
d ln r
,
(2)
where M ′(r) ≡ dM/dr. This expression can be simplified
using the definition of α(r), which gives M ′(r) r = [3 −
α(r)]M (r). Using Cartesian coordinates about the satellite
centre, where ℓℓℓ = (ℓ1, ℓ2, ℓ3) and ℓℓℓ1 lies along r, we get
such that locally ¯ρ ∝ r−α. We assume that α is either con-
stant or monotonically increasing as a function of r, with
values in the range 0 6 α 6 3. The extreme values of
α = 0 and 3 correspond to a constant-density halo and a
point mass respectively. Note that if the profile inside r is
a power law, then the local and mean density profiles have
the same logarithmic slope. In general, they are related via
ρ(r) = [1 − α(r)/3]¯ρ(r), but they do not necessarily have
the same slope at a given r. The slope of ρ(r) is equal to or
larger than the slope of ¯ρ(r) at any r. The following analysis
refers to α as the slope of ¯ρ(r).
In the analytic part of our analysis we assume that
the density profile of the original halo mass is fixed in time
throughout the duration of the merger while the mass torn
from the satellite is gradually being added to the halo. This
is confirmed to be a reasonable approximation in our N-
body merger simulations, where the mass ratio is 1:10 (e.g.
Fig. 2).
We then consider a satellite of mass mv ≪ Mv, mov-
ing under the gravity and dynamical friction exerted by the
halo, when the satellite's centre of mass is at position r as
measured from the halo centre. The tidal acceleration ex-
erted by the halo mass distribution on a satellite particle
at position vector ℓℓℓ relative to the satellite centre of mass
is obtained by transforming the gravitational attraction ex-
erted by the halo on the particle into the accelerated (non-
Ft =
GM (r)
r3
([α(r) − 1] ℓℓℓ1 − ℓℓℓ2 − ℓℓℓ3) .
(5)
Alternatively, moving to polar coordinates, where θ is the
angle between ℓℓℓ and r and r = cos θ ℓℓℓ − sin θ θθθ, we obtain
Ft =
GM (r)ℓ
r3
(cid:16) [α(r) cos2 θ − 1] ℓℓℓ − α(r) sin θ cos θ θθθ(cid:17) .(6)
Fig. 1 illustrates the tidal acceleration at a given satel-
lite shell of radius ℓ in a plane that includes the position
vector r connecting the centres of mass, for given r and
M (r) and for different slopes α = 3, 2, 1 and 0. The tidal
components along ℓℓℓ2 and ℓℓℓ3, perpendicular to r, are always
negative, namely, they exert compression, and they do not
explicitly depend on α. Stretching outwards may occur only
along ±r. Unlike the common notion of tidal forces as pulling
outwards, the average force on a sphere of radius ℓ is actu-
ally inwards: hFti = (GM/r3)(α/3 − 1)ℓℓℓ. It vanishes for a
point-mass perturber, α = 3, and it obtains a maximum
amplitude in a core where α = 0.
According to eq. (6), the maximum radial tidal force
outwards (towards +ℓℓℓ) is always obtained along the line
connecting the centres of mass, ±r. According to eq. (5),
4 in some cases it is useful to perform the analysis in a rotating
frame, see §6
4
A. Dekel, J. Devor & G. Hetzroni
and as illustrated in Fig. 1, the pull outwards is maximal in
the limit where the tides are exerted by a point-mass halo,
α = 3, valid in the outer regions of typical haloes. For flat-
ter halo profile slopes, the tidal stretching becomes weaker in
proportion to (α − 1), until it vanishes at α = 1 and reverses
direction into compression for α < 1. This direction rever-
sal of the tides is often overlooked, because the common
context of tides is a perturbation by a concentrated mass
distribution. It is a simple result of the fact that the ampli-
tude of the gravitational attraction by the halo, GM (r)/r2,
is not a decreasing function of r where the halo mass pro-
file is rising rapidly enough (faster than M ∝ r2). Thus,
while for α > 1 there is always a tidal component pulling
outwards, for α < 1 the tidal forces are of compression for
any θ, namely everywhere in the satellite. In the limit of
a constant-density core, α = 0, the tides induce symmetric
compression in all three dimensions.
We note that the critical slope of α = 1 is the asymp-
totic inner slope of the cusp in the NFW profile. We suggest
that this is not a coincidence. The key idea is that if the local
tidal mass transfer from the satellite to the halo stops when
the satellite's orbit has decayed into a core region where
α(r) 6 1, the satellite would continue to sink in due to dy-
namical friction until it settles in the halo centre. This would
inevitably cause a general steepening of the core profile to-
wards α > 1. For this scenario to be valid one should verify
that, indeed, there is only little local tidal transfer of mass
from the satellite to the halo in a region where α 6 1. We
first demonstrate this using N-body simulations, and then
apply crude analytic approximations in simple cases in order
to gain better qualitative understanding of this effect.
3 MERGER SIMULATIONS
For the purpose of investigating the proposed scenario, we
run and analyze N-body simulations of isolated mergers
between a large halo and a satellite halo of mass ratio
m/M = 0.1, and then try to interpret them using analytic
toy models. We use the Tree code by Mihos & Hernquist
(1996 and references therein) with dark-matter haloes only
(no gaseous disks). In our default suit of simulations, the
large host halo is represented by N = 105 equal-mass parti-
cles and the satellite by n = 104 particles. We test the sensi-
tivity of our results to resolution by repeating one case with
N + n = 0.55 × 106 particles and a correspondingly higher
force resolution. The simulation units were chosen quite ar-
bitrarily to be: length 3.5kpc, mass 5.6 × 1010M⊙, and time
13.06Myr. The force softening length in the default case is
0.08 units, i.e. 0.28kpc. The units can be properly scaled up
or down according to desire. In most cases we refer hereafter
to distances and masses in the simulation units, but to time
in Myr.
The initial halo density profile, as measured in the un-
perturbed initial conditions, is a truncated isothermal sphere
with a flat core,
ρ(r) =
ρs e−(r/rt)2
1 + (r/rs)2 ,
(7)
Figure 2. The halo profile in the simulation represented by the
logarithmic slope α of the mean density profile as a function of
radius r (in units of 3.5kpc). Shown are the halo profile at the
initial time (solid) and the profiles of the original halo mass at
the final times of the radial (long dash) and circular (short dash)
merger simulations.
with ρs = 10.36, rs = 1 and rt = 10. The density at the
characteristic radius rs is thus ρ(rs) = 5.13. The internal ve-
locities are constructed to fulfill the isotropic Jeans equation
which ensures an equilibrium configuration as discussed in
Mihos & Hernquist (1996). When run in isolation, the halo
profile has been tested to be very stable for many dynamical
times. As shown in Fig. 2, the logarithmic slope of the den-
sity profile spans the range of interest between α = 0 and 3,
and its variation as a function of radius can be described to
a good approximation by α(r) ≈ 1.73 log r + 0.67 through-
out the range 0.3 6 α 6 2.9. The initial halo density profile
will be shown in Fig. 5 below in comparison with the post-
merger profile. It resembles the generalized NFW profile of
eq. (1) with a core of α ≪ 1.
The satellite initial density profile is fit by a Hernquist
profile,
σ(ℓ) =
σs
(ℓ/ℓs) [1 + (ℓ/ℓs)]3 ,
(8)
with the default choice σs = 19.2 and ℓs = 1 defining our
typical "compact" satellite. The initial satellite density pro-
file can also be seen in Fig. 5 below. In the inner region it
convergence to the NFW profile, α = 1. If we fit this Hern-
quist profile with an NFW profile by matching the charac-
teristic radii where the local logarithmic slope is −2, we find
that the radius corresponding to the characteristic radius of
the NFW profile is ℓ = ℓs/2. At this radius the density is
16σs/27, which for our default satellite is σ(ℓs/2) = 11.38.
The satellite parameters were chosen to roughly mimic
a typical compact satellite according to the distribution
of halo properties in the ΛCDM scenario. In a hierarchi-
cal clustering scenario with a fluctuation power spectrum
Pk ∝ kn, the halo characteristic radii and densities scale
like ℓs/rs ∝ m(1+ν)/3 and σs/ρs ∝ m−ν, with ν = (3 + n)/2.
For ΛCDM on galactic scales one has ν ≃ 0.33. This is
Halo cusp-core: tides in mergers
5
Figure 3. The 104 satellite particles in 6 snapshots during the radial (left), circular (right) and elongated (bottom) mergers, projected
onto the orbital plane. The 105 live halo particles are not shown. The centre of mass is at the origin and the dot marks the temporary
halo maximum density. The circle is of radius r = 20 ≃ 2rt about the halo maximum density, corresponding to where the initial halo
practically ends and where the satellite is at the onset of the simulation. The three thirds of the mass, in concentric shells about the
satellite bound centre at each time, are marked by different colors.
confirmed by cosmological simulations (NFW; Bullock et al.
2001a), where the typical halo profile is NFW. The haloes
of lower masses are thus typically more compact; they tend
to have lower characteristic radii and higher corresponding
mean densities within these radii (corresponding to higher
virial concentration parameters). For CDM haloes of mass
ratio m/M = 1/10, the typical ratio of characteristic radii
and corresponding densities are expected to be roughly 0.4
and 2.1. The corresponding ratios in the initial conditions
of our simulations are approximately 0.5 and 2.2, providing
a reasonable match. It may also be interesting to note that
the satellite mean density interior to ℓs/2 (or ℓs) is about
3.9 (or 1.09) times the halo mean density interior to rs.
We simulated three cases of initial merger orbits: a ra-
dial orbit, a circular orbit, and an elongated orbit with an
initial pericentre to apocentre ratio of rp/ra ≃ 1/6. The un-
perturbed satellite is put initially at r = 20, i.e. at about
2rt, where the initial circular period is about 230Myr (or
about 17.6 simulation time units). In the circular and radial
cases the magnitude of the initial satellite velocity was set to
equal the circular velocity of the halo at that radius, namely
a bound orbit with an orbital kinetic energy that equals half
the absolute value of the total energy. For the elongated or-
6
A. Dekel, J. Devor & G. Hetzroni
Figure 4. Time evolution of the satellite spherical mass profile during the radial merger (left) and the circular merger (right). Shown
in the bottom panels are the mean radii of concentric spherical shells about the momentary satellite centre, each encompassing a given
fraction of the original satellite mass, as labeled on the right. The top panels show the time evolution of position r of the satellite centre
relative to the halo centre (dotted), and the corresponding local slope α(r) (solid). For the radial merger, the times of halo-centre crossing
are marked in the bottom panel by vertical lines. For the circular merger, representative values of α(r) along the orbit are marked in
the bottom panel. The stripping of a shell can be crudely identified by a rapid increase of its radius. Lack of stripping and slight overall
contraction of bound shells is noticed whenever the satellite enters the halo core.
bit the initial tangential velocity was about one half of the
circular velocity at r = 20, in fact v/vc = 0.49. Each merger
has been followed until the satellite's bound core has prac-
tically settled at the halo centre. Fig. 3 shows the satellite
mass distribution in 6 snapshots during the merger process
in the three cases, projected onto (one of) the orbital planes.
The radial-orbit merger has been followed for more than
200Myr (about 15 time units), with equal steps of 1.306Myr
(0.1 time units) between output times. The satellite is
stretched and stripped along the line connecting the centres
of mass and the stripping produces the familiar bridge-and-
tail tidal structure. An overall shrinking is seen near the first
centre crossing, followed by a three-dimensional re-bounce
and significant mass loss about the following apocentre (dis-
cussed in §5 below). The stripped material shows a sequence
of arc-like caustic structures corresponding to each apocen-
tre, reflecting the initial confinement in phase space. The
oscillations of the satellite remnant about the halo centre
decay rapidly due to dynamical friction until it becomes con-
fined to the halo core after about 115Myr, and it practically
melts into the halo central region after ∼ 130−140Myr. The
final distribution of satellite mass extends quite smoothly
about the halo centre in a puffy ellipsoid; it is moderately
prolate, with the major axis parallel to the original merger
line.
The circular-orbit merger has been followed for 836Myr
(64 time units) with 81 output times spaced by 10.45Myr.
The satellite gradually spirals inwards due to dynamical fric-
tion. The continuous tidal stripping process in the outer
parts is obvious. Particles escape from the satellite in the
two directions roughly along the line connecting the centres
of mass of halo and satellite, and produce the familiar trail-
ing and leading tidal tails because of the differential rotation
in the halo potential. The remaining bound satellite seems
to be slightly elongated in a direction not far from the line
connecting the centres of mass, crudely reflecting the shape
of the Roche lobe (see §6 below). After entering the halo
core, at about 380Myr, the dynamical friction weakens and
the satellite orbit continues to decay very slowly towards the
halo centre. The final distribution of satellite mass extends
quite smoothly in an oblate configuration about the halo
centre.
In the elongated-orbit merger, the satellite oscillates
about the halo centre through a sequence of pericentre and
apocentre passages. The shape of the orbit remains roughly
constant while it shrinks in scale; when measuring the ratio
of pericentre to the following apocentre it is rp/ra ∼ 1/3.5,
and when measuring the ratio of pericentre to the preced-
ing apocentre it is rp/ra ∼ 1/3.5, but these ratios remain
roughly the same for all detectable pericentres. The oscil-
lations decay due to dynamical friction until the satellite
becomes confined to the halo core after ∼ 125Myr and 5
pericentre passages. By the first pericentre, the satellite is
already stretched and stripped along its orbit, while it is
temporarily shrunk in the perpendicular direction. This is
followed by a re-bounce and significant mass loss about the
following apocentre. The visual impression confirms the no-
tion that the particles that are torn away near a pericentre
radius continue on orbits that reflect on average the satel-
lite orbit at the time of stripping, while the bound rem-
nant is gradually sinking into smaller radii due to continu-
ing dynamical friction (see §5.1 below). For example, in the
snapshots corresponding to the second and third apocentres
we clearly see a large amount of stripped satellite material
spread about the location of the previous apocentre. We
can crudely say, in all the mergers simulated, that mass is
stripped near pericentre and is practically "deposited" about
the following apocentre radius. This crudely validates a toy-
model concept that each satellite shell ℓ is being practically
deposited at a certain halo radius r. The final distribution of
satellite mass extends quite smoothly about the halo centre
in a puffy oblate ellipsoid that looks quite symmetric in the
orbital plane.
Fig. 2 displayed earlier also shows the α(r) profiles of
the original halo material at the final times after the merg-
ers. The relatively small changes in this profile compared to
the changes in the profile of the total halo mass (original
halo plus stripped satellite material) indicate that the as-
sumption of a fixed halo to which the satellite mass is being
added can serve for a crude toy model.
4 PROFILE BUILDUP IN THE SIMULATIONS
Fig. 4 describes the time evolution of the satellite spherical
mass profile during the mergers by showing the mean radii
of concentric spherical shells about the satellite maximum-
density centre, each encompassing a given fraction of the
satellite mass (and not necessarily the same population of
particles at different times).
In the circular merger case, to a good approximation,
the satellite position r is a monotonically decreasing func-
tion of time, and so is α(r), as shown at the top and marked
along the time axis of Fig. 4. The stripping moment of each
shell is marked by the onset of a steep rise in the corre-
sponding curve. Proceeding from the outside shells inwards,
the stripping point of each shell can be identified with a
specific halo radius r and a corresponding α, moving down
monotonically from α ∼ 3 to flatter slopes. However, as the
stripping point passes beyond the radius where α ∼ 1 (after
∼ 380Myr), when the satellite is left with less than ∼ 30%
of its original mass, the stripping slows down and eventually
stops. The 20% and inner shells are never really stripped,
while the satellite's orbit continues to decay very slowly to-
wards the halo centre.
In the radial merger case the satellite oscillates about
r = 0. It crosses the halo core back and forth a few times
while the oscillation amplitude is decaying due to dynamical
friction until the bound remnant settles at the halo centre.
Center crossings occur near the times 22, 62, 83, 99, 110Myr
and so on. The satellite enters the core region in every os-
cillation a couple of Myr before it crosses the halo centre.
Overall contraction of bound satellite shells seems to start
roughly at these times, as indicated by the reversal of the
tidal forces near α = 1 (§2). Each major contraction is fol-
Halo cusp-core: tides in mergers
7
lowed by a re-bounce as the satellite exits the core towards
apocentre in the opposite side, which results in overall ex-
pansion and stripping of the outer shells.
After about 90Myr the satellite orbit becomes confined
to inside the region where α < 1.5, and after about 107Myr it
is confined to the inner halo where α < 1. Once the satellite
becomes confined to this halo core, there is no apparent
overall shell expansion anymore, indicating that stripping
has stopped.
In the elongated merger case (not shown to avoid redun-
dancy), pericentre passages are identified five times, and the
behavior is qualitatively similar to the radial merger case.
Overall contraction of bound satellite shells seems to start
roughly at these times. Each major contraction is followed
by a re-bounce as the satellite moves towards apocentre,
which results in overall expansion and stripping of the outer
shells. Once the satellite becomes confined to the inner halo
where α < 1, after ∼ 125Myr and 5 apocentres, there is no
apparent overall shell expansion anymore, indicating that
stripping has stopped.
It is important to note that the halo radius r where
α = 1 is more than three times larger than the radius ℓ of the
satellite's 20% shell, indicating that the cease of stripping
occurs significantly before the centre of the bound satellite
remnant coincides with the halo centre.
Fig. 5 addresses the cusp formation straightforwardly
by showing the density profile of the halo before and after
the merger in the three different cases. For the standard,
compact satellite, either on radial, circular or elongated or-
bit, the figure demonstrates the inevitable steepening of the
profile in the core region, starting near the core boundary
where α ≃ 1. In the case of radial merger, the slight deple-
tion of final density in the region near α ∼ 1, compared to
the slight increment in halo density at larger radii, is consis-
tent with no mass transfer in this region while the orbit of
the remaining satellite continues to decay into smaller radii.
A similar effect is also seen in the circular case, though it
is somewhat weaker. Not surprisingly, the steepening seen
at α <
∼ 1 in the elongated merger case is not very different
from the steepening seen in the two other cases.
In order to evaluate the sensitivity of our results to the
resolution of the simulation, we ran a case identical to the
merger on an elongated orbit described above, but now with
5 times more particles (N + n = 0.55 × 106) and a softening
length smaller by a factor of 51/3 accordingly. The results are
found to be practically identical in all respects, including the
decay rate of the satellite radius within the halo, the mass
loss from the satellite, and the final halo density profile. The
latter is shown in the bottom-right panel of Fig. 5. This
indicates that the resolution of our simulations is adequate
in the range of radii of relevance, well inside the core/cusp
region and down to below 0.2rs.
We thus learn that one single merger with a 1:10 com-
pact satellite, typical of the ΛCDM cosmology, is enough
for turning a core into a cusp of α >
∼ 1. How stable is this
cusp under further mergers with similar satellites? To test
this, we performed a series of mergers following each other.
The main progenitor halo in each merger is taken to be the
8
A. Dekel, J. Devor & G. Hetzroni
Figure 5. Halo density profile before (solid) and after (dashed) the simulated mergers with a compact satellite. The initial halo core
becomes a cusp for all types of merger orbits (specified within each panel). A significant fraction of the satellite settles intact at the halo
centre without depositing mass near α <
∼ 1, causing the steepening of the core into a cusp. The error bars refer to Poisson errors in the
r bins, and they are roughly spaced by the bin size. The thin dotted curves mark for reference the initial satellite profile (lower) and a
straightforward sum of the initial profiles of halo and satellite (upper). The bottom-right panel corresponds to the same elongated-orbit
merger as in the bottom-left panel, but simulated with 5 times more particles (0.55 × 106) and a softening length smaller by a factor of
51/3 accordingly. It demonstrates that our default resolution is adequate.
outcome of the previous merger, except that its total mass
is scaled down back to the original halo mass (such that all
the density profiles can be directly compared). We achieve
this by letting each simulation involve the same numbers of
N halo particles and n satellite particles (each of a fixed
mass m), where the halo particles are selected at random
from the N + n particles of the halo produced in the previ-
ous merger. In these tandem mergers we used N = 20, 000
and n = 2, 000. The halo profile in the range of interest
is found to remain stable for at least several hundred Myr
when the halo is run in isolation. The satellite starts with
the same profile as the original compact satellite. It is put
at the same distance of r = 20 as before but in an elongated
orbit of a random spatial orientation. The initial satellite
velocity is tangential to the line connecting the centres of
mass of halo and satellite, with an amplitude in units of the
circular velocity at r = 20 chosen at random in the range
v/vc = 0.2 − 0.6 (compared to the typical case simulated
before, where v/vc = 0.49 led to peri/apocentre ratio of
roughly 1:6). Each merger was followed for 325Myr before
the following merger started. The density profiles after each
of the first 5 mergers are shown in Fig. 6. We see that the first
merger reproduces a cusp very similar to the cusp produced
when the merger was simulated with N = 100, 000 (and with
N = 500, 000), Fig. 5, indicating that even N = 20, 000 is
adequate for crude results in the regime of interest here. We
see that inside the initial core radius the cusp is stable at
a slope of α ≃ 1.5, with no systematic tendency to deviate
from the power-law profile resulting already after the first
merger. At the radius where α >
∼ 1.5 the profile flattens, and
the amplitude becomes lower partly because of the renormal-
ization of the total mass between every two mergers.
We conclude that the NFW inner slope of α = 1 is in-
deed a robust lower bound, as indicated by the reversal of
the tidal forces there (§2). Any flatter density core does not
survive as long as bound satellites bring enough mass into
Halo cusp-core: tides in mergers
9
5.1 Orbits, stripping and deposit
The validity of each approximation depends on the nature
of the orbit of the satellite within the halo. Along a typical
orbit, the satellite distance from the halo centre oscillates pe-
riodically between the radii of apocentre ra and pericentre
rp, while their amplitudes gradually decay due to dynam-
ical friction. Ghigna et al. (1998) studied the distribution
of satellite orbits in a high-resolution N-body simulation of
a cluster emerging from a CDM cosmological background
(mostly in an extended range where α ∼ 2). They found
that the median ratio ra/rp is 6:1, with about 25% of the
orbits more eccentric than 10:1, and concluded that radial
orbits are common while circular orbits are rare. The ex-
pected more rapid tidal disruption of satellites on radial or-
bits may indicate that the actual initial distribution of orbits
tended even more towards radial orbits.
Ghigna et al. (1998) also demonstrated that the tidal
radii of the satellites, ℓt, are consistent with being deter-
mined near pericentre, under the general resonance condi-
tion ¯ρ(rp) ∼ ¯σ(ℓt), where ¯σ(ℓ) ∝ m(ℓ)/ℓ3 is the mean density
profile of the satellite. Particles that escape from the satellite
can be assumed, on average, to continue on an orbit about
the halo centre with apocentre and pericentre radii "frozen"
at their values near the time of escape, suffering no further
decay due to dynamical friction. Since a particle spends most
of its time near the apocentre of its orbit, we can vaguely say
that the escapers are effectively "deposited" in the halo near
the apocentre radius valid at the time of escape. This effect
is clearly seen in our simulation of the radial and elongated
merger orbits, e.g., Fig. 3, where a large cloud of stripped
particles is seen left behind at each successive apocentre.
Very eccentric orbits may involve "penetrating" encoun-
ters near pericentre, where rp < ℓ for ℓ a characteristic ra-
dius of the satellite. In this case the tidal approximation
becomes invalid. The tides are weakened when the centre
of the perturber lies inside the satellite, until they vanish
when the satellite centre coincides with the halo centre. It
can be shown that the effect of a fast penetrating encounter
is comparable to the effect of an encounter with rp ∼ ℓ as
computed using the impulse approximation in the tidal limit
(BT §7.2.e). In a slow penetrating situation the tidal effects
simply weaken gradually as r → 0.
5.2 The impulse limit
The impulse approximation may be partly valid for satel-
lites on elongated orbits during their first quick pericentre
passages. As long as α > 1 along the orbit, the tidal force,
Ft ∝ ¯ρ(r), has a peaked maximum at the ∼ rp vicinity of
the pericentre, where the satellite spends a small fraction of
its orbital period. Since ¯ρ(rp) ∼ ¯σ(ℓt), and since the typical
orbital period is inversely proportional to the square-root of
the density, the satellite orbital period is comparable to the
internal orbital periods of outer satellite particles. Therefore,
the impulse approximation may be partly valid for such par-
ticles near their apocentres within the satellite.
In the impulse approximation, each satellite particle is
assumed to obtain an instantaneous velocity kick ∆v, which
Figure 6. Halo density profile after tandem mergers with com-
pact satellites of mass ratio 1:10 on random elongated orbits. The
initial halo profile, with a core, is shown for comparison (solid
line). After each merger, the total mass of the halo is scaled down
back to the original mass. We see that the cusp which forms in
the first merger remains stable under the subsequent mergers.
the inner halo and deposit it near the halo centre. This pro-
vides a simple explanation for why the cusp slopes seen in
cosmological N-body simulations of CDM models are typi-
cally α > 1. It implies that the only way which may possibly
enable a flatter core is by suppressing the settling of satel-
lites in the core, as demonstrated by simulations with puffy
satellites (§8).
Next, we continue our effort to understand the tidal
mass transfer seen in the simulations. We build upon the α
dependence of the tidal forces discussed in §2, and try to
evaluate the resultant mass transfer using simple dynamical
considerations and idealized toy models.
5 THE IMPULSE AND ADIABATIC LIMITS
In this section we analyze the tidal mass transfer from satel-
lite to halo by referring to particles in orbits within the satel-
lite. The α dependence of the tidal process in the central
regions of a halo can be qualitatively evaluated in two ex-
treme limits. If the tidal forces vary on a time scale shorter
than the particle orbital periods within the satellite, then
one can use the impulse approximation (Spitzer 1958; Bin-
ney & Tremaine 1987, hereafter BT, §7.2) to estimate the
energy input into the satellite and the resulting mass loss. If
the tidal forces vary on a time scale longer than the orbital
periods within the satellite, then one can appeal to adia-
batic invariants along the particle orbits (e.g., BT §3.7) in
order to estimate the gradual energy change of the satellite
particles, the associated structural changes in the satellite,
and the escape rate.
10
A. Dekel, J. Devor & G. Hetzroni
is the integral R Ftdt over the short duration of effective
tides near pericentre along the satellite orbit. The energy
change per unit mass of a particle moving with momentary
velocity v is thus ∆E = (1/2)(v + ∆v)2 − (1/2)v2, which
consists of linear and quadratic terms in ∆v, namely
∆E = v · ∆v + (1/2)(∆v)2 .
(9)
Some particles may be kicked by this impulse to above the
escape velocity at their position and become unbound.
If the satellite is spherical and the particle motions out-
wards and inwards are symmetric, a kick inwards is as effec-
tive in feeding energy into the satellite as a kick outwards.
The contribution of the linear term to the total energy van-
ishes by the symmetry of incoming and outgoing particles,
but the linear term may have an important contribution for
the escapers which come from the high-end tail of the ∆E
distribution.
The dominant tidal kick ∆v is due to the systematic
force towards the satellite centre along the direction per-
pendicular to the orbital plane, where the tidal force is α-
independent and persistently inwards [eq. (5)]. A somewhat
smaller kick is obtained when integrating the force compo-
nent along the position vector of the satellite (r) at pericen-
tre (it is a net kick inwards if α < 2), while the contribution
along the position vector at apocentre tends to average out
when integrated along the orbit (BT §7.2, eq. 7-54; Gnedin,
Hernquist & Ostriker 1999).
If indeed ¯ρ(rp) ∼ ¯σ(ℓt), then the velocity kick integrated
along a path of length ∼ rp near pericentre is comparable
to the typical particle velocity, ∆v ∼ v ∼ ¯ρ(rp)1/2ℓt. This
makes the energy change maximal in the outer satellite re-
gions, with comparable contributions from the linear and
quadratic terms. The kick inwards causes a delayed escape,
because the escapers need to cross the satellite before depar-
ture, and the crossing time is comparable to the orbital pe-
riod in the halo. Therefore, the actual stripping is expected
to occur in the halo near the following apocentre, where the
particle is "deposited" anyway (§5.1).
When apocentre is at α > 1 but pericentre is already
in the α < 1 regime, the duration of maximum tides be-
comes longer and the impulse approximation less valid. Any
stripping that still occurs along such an orbit ends up with
"deposit" outside the α < 1 core.
The shrinkage near pericentre and the following re-
bounce and later mass deposit near apocentre are clearly
seen in the simulations of the radial and elongated mergers,
Fig. 3 and Fig. 4.
Once the satellite orbit becomes confined to the inner
halo where α < 1, we expect no impulse stripping. If the
slope near pericentre is α, the tidal force drops to half its
peak magnitude at a distance of ∼ 21/αrp from pericentre.
At α = 1/2 this is already 4rp, so the tidal force is not
much weaker than its pericentre value throughout a large
fraction of the orbit. Furthermore, adiabatic compression
makes ¯σ become larger than ¯ρ(rp) (§5.3 below). These two
effects make the impulse approximation invalid. The effect
of the inwards tidal force accelerating an incoming particle
is roughly balanced by the inwards tidal force of similar
magnitude working to decelerate the particle when it is going
out. In the limit where α = 0, the tidal force is the same
in all directions and constant in time, so R Ftdt = 0 along
each particle orbit and there is no net impulse.
We conclude that we expect no impulse tidal mass de-
posit in the core region where α < 1: neither during the
early quick core passages nor after the decay of the orbit
into the core region. This is consistent with the behavior in
the radial and elongated merger simulations.
5.3 The adiabatic limit
The adiabatic invariance of actions, which is formally valid
when the internal orbital periods are much shorter than the
characteristic time over which the external potential is vary-
ing, is known to be crudely applicable even when these time
scales are comparable to each other (BT, §3.6, Fig. 3-29).
Since ¯ρ(rp) ∼ ¯σ(ℓt), and since for bound satellite particles
along the satellite orbit ℓ 6 ℓt and r > rp, The satellite
particles inwards of the tidal radius are likely to obey the
adiabatic approximation at all times. Once the decayed orbit
becomes confined to a central halo region of α < 1, where
the tidal force varies slowly along the orbit and actually
tends to a constant when α → 0, the adiabatic approxima-
tion becomes crudely applicable also for the outer satellite
particles. Any adiabatic compression of the satellite would
strengthen the applicability of this approximation.
In this case, the tides gradually distort the satellite,
and may cause slow mass loss mostly outside a momentary
tidal radius. Now tidal forces outwards cause stretching and
energy gain which may lead to mass loss, while tidal forces
inwards cause contraction, make the satellite more bound
and do not lead to stripping.
In order to gain some insight into the α dependence
of adiabatic tidal effects we consider the following simple
example. We see in eq. (5) that the tidal perturbation in
each direction (i = 1, 3) mimics the force of a harmonic
oscillator about the satellite centre,
Fti = −ktiℓi ,
kti ∝ ¯ρ(r) [(1 − α), 1, 1] ,
(10)
except that the force constant along ℓℓℓ1 is negative for α > 1.
If we assume as a toy model that the satellite core is a ho-
mogeneous sphere with density σ0, the orbits under its own
self gravity are also three-dimensional harmonic oscillators,
with
F0i = −k0iℓi ,
k0i ∝ σ0 .
(11)
This can be generalized to a homogeneous ellipsoid, with
different force constants in the different directions, based
on Newton's third theorem (e.g. BT, §2.3, Table 2-1). The
perturbed system is thus a 3-dimensional harmonic oscilla-
tor, with force constants ki = k0i + kti, and corresponding
frequencies w2
i = ki. If the motion of the satellite within
the halo introduces slow variations in the tidal perturba-
tion, and correspondingly in ki, the particles should conserve
adiabatic invariants along their orbits. The case of a radial
satellite orbit within the halo is particularly simple because
the principle directions remain stationary. The adiabatic in-
variant of the harmonic oscillator along each direction is the
action integral
Z v2
i dt ∝ wiL2
i ∝ −Ei/wi ≃ const. ,
(12)
where the integral is over the period of the oscillation, Li is
the amplitude of the oscillation, and Ei is the corresponding
energy (e.g, BT, §3.6.b). Thus ∆Ei ∝ −∆ωi, meaning that
strengthening (weakening) of the force constant ki, corre-
sponding to an increase (decrease) in wi, leads to shrinking
(stretching) and negative (positive) energy changes. This im-
plies that escape may be caused only for particles moving
roughly along the radial-orbit line (±ℓℓℓ1) and only as long as
k1 ∝ ¯ρ(r)[1 − α(r)] is decreasing while r is decreasing. When
the satellite is decaying within a halo core where α < 1 and
where the forces are towards the satellite centre in all di-
rections, all the force constants are clearly increasing with
decreasing r, so there is systematic shrinking in all direc-
tions. The corresponding energy changes are negative for all
the particles, implying no stripping.
When the satellite orbit becomes confined to a flat core
of α = 0 and density ρ0, the restoring force is constant and
fully spheri-symmetric, F ∝ −(ρ0 + σ0)ℓℓℓ, independent of
the actual orbit of the satellite within the halo core. The
satellite is then in its most tightly bound configuration. We
may expect the situation not to be very different for small
α values in the range α < 1, as the deviation from spherical
symmetry is limited to one direction, along which the force
is still restoring and is getting stronger as r decreases.
The general behavior in the merger simulations is con-
sistent with the behavior in the above toy model. We noticed
in Fig. 4 that once the satellite becomes confined to the halo
core there is no further overall shell expansion, indicating
that stripping has stopped. at α < 1 can be qualitatively
reproduced in the extreme limiting cases where the impulse
or the adiabatic approximations may be crudely valid. This
provides some intuitive understanding for why the slope of
α ∼ 1 is a critical value below which mass transfer is ineffi-
cient.
6 TIDAL RADIUS
We now turn to a complementary approach, where the satel-
lite is addressed as a spherical "onion" from which the outer
layers are gradually peeled off by the tides. The α depen-
dence of a slow stripping process can be estimated by means
of the momentary tidal radius. This depends on the nature
of the orbit of the satellite within the halo, which dictates,
for example, whether the most useful frame of reference for
defining the tidal radius should be rotating or not.
A lower bound to the tidal radius is obtained rigorously
in the (rare) case of a circular orbit. If the circular angular
velocity is Ω, the potential is stationary in a frame which
co-rotates with the same angular velocity, in which the cor-
responding fictitious centrifugal force should also be taken
Halo cusp-core: tides in mergers
11
into account. In this case, the Jacobi integral J is conserved
along the orbit of each satellite particle (see BT, §3.3 and
§7.3). This integral is the energy as measured in this ro-
tating frame including in the effective potential the term
corresponding to the centrifugal force, −(Ω × x)2. The out-
ermost closed zero-velocity surface of constant J defines a
firm closed boundary around the satellite, the Roche lobe,
which cannot be crossed by particles from the inside out.
Therefore, the distance from the satellite to the Lagrange
point, the saddle point of the effective potential, provides a
lower bound for the tidal radius. This Lagrange point can
be obtained by balancing in the rotating satellite frame the
self-gravity of the satellite with the sum of the tidal and cen-
trifugal forces along the line connecting the centres of mass.
The self-gravity pull inwards by the satellite mass interior
to ℓ is −[Gm(ℓ)/ℓ3]ℓℓℓ and the centrifugal force is of an op-
posite sign and the same amplitude because of co-rotation
[Ω2 ∝ M (r)/r3 = m(ℓ)/ℓ3]. Using the ℓℓℓ1 component of the
tidal force from eq. (5) we obtain for the tidal radius when
the satellite is on a circular orbit of radius r:
α(r) ¯ρ(r) = ¯σ(ℓt)
(13)
(compare to BT, eq. 7-84, for the case α = 3). In this case the
combined effect of the tidal and centrifugal forces along the
line connecting the centres of mass is always pointing away
from the satellite centre, towards +ℓℓℓ. This stretching force
becomes weaker as α decreases, but it completely vanishes
only at α = 0. In the direction perpendicular to the line
connecting the centres of mass but still in the merger plane
the tidal compression along −ℓℓℓ is exactly balanced by the
centrifugal force, independent of r and α. In the direction
perpendicular to the merger plane, the centrifugal force is
zero so the net force is the usual tidal compression, with no
explicit α dependence.
In the other extreme (more realistic) case of a merger
on a radial orbit, the only forces working on the particle
in the satellite (non-rotating) rest frame are the tidal force
and the self-gravity pull inwards. The direction of maximum
tides is stationary, and only the magnitude of the force varies
as a function of r. Even though there is no exact analog of
the Jacoby integral in this case, one may still define a tidal
radius ℓt by the point along the line of maximum tides at
which the tidal and self-gravity forces balance each other
such that the net force in the satellite frame is zero. The
condition replacing eq. (13) for a radial orbit is thus
[α(r) − 1] ¯ρ(r) = ¯σ(ℓt) .
(14)
In the above expressions for the tidal radius, the satel-
lite profile is crudely assumed to be fixed in time for ℓ < ℓt
and its mass distribution is assumed to remain spherical; the
shells outside the tidal radius are assumed to be peeled like
layers off an onion while the tidal distortions are neglected.
We now investigate the evolution of the tidal radius,
starting with the radial-orbit case. As r is decreasing from
the virial radius inwards, where α is large, the function
[α(r) − 1] ¯ρ(r) is increasing. Assuming that ¯σ(ℓ) is a mono-
tonically decreasing function, this means that ℓt is decreas-
ing. More and more mass of the satellite is being torn away
as the satellites is moving in. However, as the satellite ap-
12
A. Dekel, J. Devor & G. Hetzroni
proaches the position where α approaches unity, the function
[α(r)−1] ¯ρ(r) approaches zero, implying that the tidal radius
grows to infinity. This means that there is a halo radius rm
and a corresponding slope αm > 1 at which [α(r) − 1] ¯ρ(r)
obtains a maximum. If the inner satellite density is high
enough, then eq. (14) implies that ¯σ(ℓt) obtains a maximum
value ¯σm at rm, and therefore ℓt obtains a minimum value
ℓtm there. Since at r < rm (and α < αm) the tidal radius
derived by eq. (14) is larger than ℓtm, we expect no further
stripping inwards of rm. The slope αm thus characterizes
the point of minimum tidal radius inside which the strip-
ping stops, and ℓtm marks the boundary of the central part
of the satellite which remains intact, not to be disrupted.
The subsequent sinking in of this whole remnant to the
halo centre due to dynamical friction causes a steepening
of the profile at r < rm where α < αm. This steepening is
effective at least as long as the inner slope is flatter than
α = 1. It can be avoided only if the inner satellite density is
low enough such that ℓt = 0 already at r > rm, namely the
satellite is fully disrupted outside the halo core region (see
§8 and Fig. 10).
As an example, assume that the mean halo density
profile is given by eq. (1) with ρs = rs = 1, namely
¯ρ(r) = r−α0 (1 + r)α0−3, and allow the inner slope α0 to
take a value between 0 and 3.5 The corresponding α profile
is α(r) = (α0 +3r)/(1+r) and we find that [α(r)−1] ¯ρ(r) ob-
tains a maximum at rm = (1−α0)/2+(3α2
0 −12α0 +9)1/2/6.
Fig. 7 shows, for a range of α0 values, the corresponding val-
ues of the minimum-tidal-radius quantities rm, αm, and the
corresponding ¯σ(ℓt) according to eq. (14). We see that for a
flat core, α0 = 0, the minimum tidal radius is obtained at
αm = 1.5 and rm = 1. For larger α0 in the range (0, 1), the
value of αm gradually decreases towards αm = 1, namely
it converges to the asymptotic value α0 = 1 at r → 0. For
α0 > 1, the minimum tidal radius is obtained at rm = 0,
with αm = α0. For α0 in this range the stripping continues
until the tidal radius shrinks to zero and the whole satellite
is disrupted.
An overall tidal compression at α < 1 should actually
enhance the steepening effect. It would compactify the satel-
lite while in the halo core and add halo mass to it, which
would make the dynamical friction more effective and speed
up the orbit decay to the halo centre. Moreover, the halo
would react to the addition of satellite mass to its centre by
further contraction, preferentially at smaller radii, roughly
obeying the adiabatic invariant that requires M r ∼const.
inside each shell of material.
In the case of a circular orbit, eq. (13), an analogous
analysis to the one summarized in Fig. 7, for the example of a
mean halo profile given by eq. (1), yields for a core of α0 = 0
a minimum tidal radius at αm = 3/4 (the analog of αm = 1.5
in the radial case). The minimum tidal radius becomes zero
and αm = α0 for any α0 > 3/13 (the analog of α0 > 1 in the
radial case). Since eq. (13) provides a lower bound to the
tidal radius, these values of αm should be regarded as lower
bounds for the circular-orbit case.
5 The corresponding local density profile is a Hernquist profile,
ρ(r) = (1 − α0/3)r−α0 (1 + r)α0−4.
Figure 7. Quantities when the tidal radius obtains a minimum,
as a function of the halo inner slope α0 in eq. (1). Shown are the
position in the halo rm/rs, the slope αm and the mean density
within the minimum tidal radius ¯σm(ℓt). Note that αm > 1.
Since at a given position in a given halo central re-
gion (given α) the stripping seems to be more efficient for
satellites on circular orbits than on radial orbits, and the
compression is correspondingly less effective, the dynamical
friction becomes weaker for circular orbits. This should slow
down the dynamical-friction decay of the orbit to the halo
centre and the steepening process, as noticed in our N-body
simulations.
Thus, based on the current simplified analysis, the satel-
lites on circular orbits are not necessarily as effective as those
on radial orbits in steepening cores to cusps. However, we ar-
gue in Dekel et al. (2003) (as summarized in §7 below) that
the α dependence of the mass-transfer efficiency is enough to
cause steepening of the profile even by satellites on circular
orbits, as is demonstrated by the simulations (§4).
The tidal radius of a satellite on a realistic, elongated
orbit is likely to lie in between the radii obeying eq. (14) and
eq. (13). At large r (compared to pericentre), an elongated
orbit can be approximated as a radial orbit. Near pericentre
of an orbit about a point-mass halo the orbit might have
resembled a circular orbit, but in a flat halo core the orbit
near pericentre is better approximated by a straight line.
Thus, the steepening effects of typical satellites on the inner
halo profile are more likely to be similar to the effects eval-
uated for radial orbits. Furthermore, the fact that even in
the simulation of a merger on a circular orbit the mass de-
posit seems to practically stop near α ∼ 1, not far from the
behavior in the radial-orbit case, indicates that the estimate
of tidal radius based on the Jacoby integral may indeed be
an underestimate.
We conclude that the general behavior in the simula-
tions of tidal mass transfer as a function of α(r) can be
qualitatively understood also in terms of the idealized con-
cept of stripping outside a tidal radius.
7 CONVERGENCE TO AN ASYMPTOTIC
CUSP
The tidal mass transfer in halo regions where the slope is
α > 1, and the actual convergence from either side to an
asymptotic cusp, are analyzed in an associated paper (Dekel
et al. 2003). We provide a quick preview of this analysis here
for the completeness of the discussion.
We develop a simple prescription for tidal mass transfer
by relating every shell ℓ in the initial satellite with a "de-
posit" radius r in the halo. This is done by equating the
encompassed mass within ℓ in the initial satellite with the
mass of stripped satellite material found inside halo radius r
after the merger, m(ℓ) = mfin(r). This correspondence can
be expressed in terms of α:
¯ρ(r)
¯σ(ℓ)
= ψ[α(r)] ,
(15)
where ¯ρ(r) and ¯σ(ℓ) are as before the initial mean density
profiles of the halo and the satellite. What makes this pre-
scription useful is that ψ(α) is found to be quite insensitive
to the specific nature of the merger (see below). Several au-
thors (e.g., Syer & White 1998; Klypin 1999a) applied the
crude resonance condition ψ = 1. However, the α depen-
dence of the tidal force, and in particular its vanishing at
low α, imply that ψ(α) should be a monotonically decreasing
function of α. When we adopt the approximation of peeling
layers off a fixed onion as above, we obtain from eq. (14) and
eq. (13) that ψ ∝ (α − 1)−1 and ∝ α−1 respectively. When
we also take into account the deposit at apocentre due to
stripping at pericentre and the stretching and distortion of
the inner satellite mass before it is being torn away6, we
expect ψ to be smaller than unity even near α ∼ 1 and to
decline further with α at α > 1.
We then measure ℓ(r) and obtain ψ(α) in several differ-
ent merger simulations like the ones described above. The
measured ψ(α) is found to be qualitatively similar for the
different merger cases, despite the very different orbits and
initial density contrasts between satellite and halo. [This
similarity can be seen from the resemblance of the final pro-
files r(ℓ) in Fig. 4]. An approximate empirical fit is obtained,
for example, by ψ(α) = 0.5/α, with a spread of less than
±0.1 about it in the different cases studied. The robustness
of ψ(α) indicates that eq. (15) can serve as a useful approx-
imate recipe for tidal mass transfer in the general case. We
conclude that the simplified resonance condition, ψ = 1,
does not provide a good approximation for where the satel-
lite mass ends up. The actual mass transfer is more efficient
(ψ < 0.5), and its relative efficiency gets higher in steeper
regions of the halo profile.
If the mass transfer is described by eq. (15) with ψ(α)
declining rapidly enough, we show in Dekel et al. (2003)
that the profile evolves slowly towards an asymptotic sta-
>
ble power law with αa
∼ 1. We assume that the halo and
satellite are drawn from a cosmological distribution (as pre-
dicted theoretically and seen in simulations); they are ho-
6 a decrease in satellite density before stripping has been seen in
simulations (e.g. Klypin et al. 1999a, Fig. 6; Hayashi et al. 2002)
Halo cusp-core: tides in mergers
13
Figure 8. A schematic illustration of satellite mass deposit in
the halo. Shown are an NFW halo profile ¯ρ(r), and a homologous
satellite ¯σ(ℓ) properly shifted to the left and upwards. The arrows
connect shell radii ℓ to the halo radii where they are deposited r.
The horizontal dashed arrows refer to stripping when stretching
is ignored, ψ(α) = 1. This would steepen the profile, as steep
regions of ¯σ(ℓ) are deposited at flatter regions of ¯ρ(r). The solid
arrows illustrate realistic stripping after stretching. The vertical
displacements, which grow with r, refer to ψ(α) < 1. The slope
at ℓ may be flatter than the slope at r such that the mass tends
to be deposited at larger r and the result is flattening of the halo
profile.
mologous, with their characteristic radii and densities scal-
ing like ℓs/rs ∝ m(1+ν)/3 and σs/ρs ∝ m−ν, where ν ≃ 0.33
for ΛCDM. Fig. 8 helps understanding intuitively the ori-
gin of an asymptotic slope due to the decreasing nature of
ψ(α). We write ¯ρfinal(r) = ¯ρ(r) + ¯σ(ℓ)ℓ3/r3, and obtain for
the change of the local slope in a merger
∆α(r) ∝ −
d
dr (cid:18) ¯σ(ℓ)
¯ρ(r)
ℓ3
r3(cid:19) .
(16)
While every power law is a self-similar solution, ∆α(r) =
0,
it is not necessarily a stable one. For example, with
ψ =const. one would have obtained ∆α(r) > 0 everywhere
(because ℓ/r is decreasing with r), namely a continuous
steepening towards α = 3. On the other hand, a realistic
mass transfer where ψ is decreasing with r may produce a
stable fixed point where ∆α = 0 and the second derivative is
negative. A rigorous linear perturbation analysis determines
the rate of convergence and yields an equation for the value
of the asymptotic slope αa under a sequence of mergers with
the same mass ratio m/M :
∆α ∝ α(α − 3)ψ′(α)/ψ(α) + 3 ln[(m/M )−νψ(α)] = 0 .
The solutions for m/M large enough are typically in the
range 1 < αa 6 1.5. For a sequence of mergers with a cosmo-
logical distribution of mass ratios we obtain an asymptotic
slope comparable to the solution of eq. (7) with m/M ≃ 0.3.
In order to test the linear analysis, we perform toy
simulations of the profile buildup by cosmological mergers,
where we implement the mass-transfer recipe, eq. (15), with
ψ(α) = 0.5/α. Fig. 9 shows the convergence of α at a fixed r
to the asymptotic value. The profile actually evolves through
14
A. Dekel, J. Devor & G. Hetzroni
Figure 10. Halo density profile before (solid) and after (dashed) the merger with a puffy satellite, whose density at the characteristic
radius has been scaled down by a factor of about 2.5 compared to the compact satellite. Left: the radial merger simulation. Right: the
circular merger simulation. Notation is as in Fig. 5. The puffy satellite loses most of its mass in the outer halo, leaving the halo core
practically unaffected.
only if the cores of satellite haloes were puffed up signifi-
cantly by processes other than the gravity of the cold dark
matter. In this case the satellites would practically disrupt
while the apocentre of their orbit is still in the outer halo,
and the halo core could remain undamaged.
How much puffing is needed? In order to obtain a first
clue, we have performed simulations similar to those de-
scribed in §3 and §4, except that the compact satellites of
CDM have been replaced by more puffy satellites of the
same total mass. In the Hernquist profile, eq. (8), where the
default compact satellites had σs = 19.2 and ℓs = 1, the
parameters of the puffy satellites are now σ′
s = σs/8 and
ℓ′
s = 2ℓs. With this choice, the density at ℓ = 1/2, the char-
acteristic radius of the original compact satellite, is scaled
down by a factor of 2.3, to just below the density of the
host halo at its characteristic radius. This corresponds to
a reduction by a factor of 2.8 in the mean density interior
to ℓ = 1/2. Fig. 10 shows the effect of such a merger with
a puffy satellite on the halo density profile in the case of
circular and radial mergers. When the satellite is puffy, we
see that almost all the satellite mass is stripped before the
satellite orbit decays to the α 6 1 zone, and as a result the
halo core is practically unaffected. We learn that a modest
reduction in the initial satellite inner density is enough for
preventing the cusp formation seen in Fig. 5.
In fact, mergers with puffy satellites is a plausible mech-
anism for turning a halo cusp into a flat core. A sequence of
such mergers, in which the stripped satellite mass is added
to the halo outside its inner region, may lead to a continuous
increase of density in the outer halo while the inner density
remains unchanged, and thus flatten the inner profile. This
process is a subject of ongoing work beyond the scope of
the present paper; at the moment it is only a sensible spec-
ulation. Our solid result so far is limited to the fact that
puffing of the satellites is a necessary condition for halo core
survival in the CDM scenario.
Could baryonic feedback effects be responsible for the
Figure 9. Toy-simulation evolution of slope α at r = 0.1rs
due to a sequence of mergers with mass ratio m/M = 0.3. The
initial profile is eq. (1) with α0 either zero or 2. When α < 1, the
slope steepens rapidly to α > 1 within a few mergers and then it
converges slowly from either side towards an asymptotic value.
momentary profiles which are more relevant for comparison
with real haloes at different times during their buildup pro-
cess; they resemble in shape the generalized NFW profile,
with an inner cusp growing slowly from α = 1 to α = αa.
We find a weak sensitivity to the cosmological power spec-
trum of perturbations, partly because of the robust tidal
effect driving the slope to α > 1, as discussed in the current
paper.
8 A FLAT CORE BY PUFFY SATELLITES
Our robust result so far is that a cusp of α > 1 develops in
dark-matter haloes whenever mergers with compact satel-
lites bring enough bound clumps to reside in the inner parts
of haloes. It implies that a necessary condition for maintain-
ing a flat core is preventing bound clumps from settling in
the core. This can be achieved within the ΛCDM scenario
puffing up of haloes that could help the haloes maintain their
cores, or even turn halo cusps into cores? It is known, for ex-
ample, that the dissipative infall of the baryons into the halo
centre causes adiabatic contraction of the inner dark-matter
halo. If a significant fraction of the baryons is then blown out
of the galaxy with a velocity much higher than the escape
velocity (e.g., as observed by Adelberger et al. 2002), then
the inner halo may expand in response to a configuration
more extended than the original configuration before bary-
onic infall.7 It has been estimated that even in the extreme
limit of instantaneous blowout of all the gas the effect on
the halo is not strong enough for straightforwardly turning
a cusp into a core in a final large halo (Geyer & Burkert
2001; Gnedin & Zhao 2002; though Navarro, Eke & Frenk
1996 find a stronger effect). However, this effect is possibly
sufficient for the necessary indirect puffing-up of the merg-
ing small haloes. Gnedin & Zhao (2002) estimate that direct
feedback effects may reduce the central halo densities by a
factor of 2 to 6. Based on our simulations with puffed-up
satellites, this may be enough by itself to avoid the steepen-
ing from a core to a cusp in the framework of the merging
scenario. Furthermore, the strength of the effect depends on
how deep in the potential well the gaseous disk resides be-
fore it is blown away, which is determined by the baryonic
spin. The possibility that baryons in the merging satellites
have lost angular momentum due to over-cooling before re-
gaining it by feedback (Navarro & Steinmetz 2000 and ref-
erences therein; Maller & Dekel 2002, see §9 below) allows
one to consider lower spin values than measured in today's
disk galaxies, and thus obtain stronger effects on the dark
matter.
In fact, one can think of special circumstances within
the hierarchical buildup scenario that may boost up the feed-
back effects even further and help maintaining flat cores.
We sketch here such a scenario, based on enhanced feed-
back in merging satellites due to the tidal compression of
the baryons. Consider a satellite made of dark-matter and
a typical fraction of baryons merging with the halo on a
typical eccentric orbit, which takes the satellite through the
halo core to an apocentre in the outer halo a few times
before it decays to inside the core region (as in our radial
and elongated merger simulations). Assume that cooling in
the satellite is efficient such that before the satellite passes
through the halo core the baryons are already concentrated
in the satellite core, making, say, a half-and-half mixture
with the dark matter there. In every passage of the satel-
lite through the halo core, where α < 1, the tides compress
the satellite into high densities (Fig. 4), creating shocks and
stimulating an efficient burst of star formation. (The tides
may also induce accretion of material including massive stars
from the core of the host galaxy onto the satellite). By the
time the satellite is turning around in the following apocen-
tre outside the halo core, the massive stars have produced
supernova-driven winds which drive much of the gas out
of the satellite. The satellite have already lost much of its
outer dark-matter envelope at this point and is basically
made of the original satellite core plus most of the satellite
baryons. If the satellite core loses half its bound mass in this
7 Note, for example, that a virial system that instantaneously
loses one half of its mass becomes totally unbound.
Halo cusp-core: tides in mergers
15
gas blow-out, and the gas expulsion is instantaneous, then
the satellite becomes completely unbound. But even if the
expulsion is slow, the adiabatic invariants imply a density
drop by a factor of ∼ 8. Thus, the remaining satellite is now
much more susceptible to tidal stripping, which could dis-
rupt it completely before it manages to settle in the halo
core. This scenario should be investigated in detail before
one can consider it viable. At this point it is just an intrigu-
ing speculative proposal for enhanced feedback effects, based
on the wisdom gained so far regarding the tidal effects, and
very crude common wisdom regarding star formation and
feedback (e.g. Dekel & Silk 1986).
9 DISCUSSION
The main result of this paper is that tidal compression en-
forces an inner cusp of slope α > 1 in dark-matter haloes
which are subject to tandem mergers with relatively com-
pact satellites as in the ΛCDM cosmology. The lower bound
of α ≃ 1 coincides with the transition from tidal forces out-
wards (at α > 1) to tidal forces inwards (at α < 1), resulting
in no local mass transfer from the satellite to the halo near
its core boundary and therefore steepening of the core into
a cusp. This robust effect is one way to explain the halo in-
ner structure as seen in cosmological N-body simulations of
the hierarchical clustering scenario. It then helps providing
a tool for addressing other processes which may explain the
observed flat cores in (at least some) galaxies. More details
of the cusp formation are provided in an associated paper
(Dekel et al. 2003), where we derive a simple prescription
for tidal mass transfer between satellite and halo, and use it
to show that successive cosmological mergers slowly lead to
a stable asymptotic cusp that is only slightly steeper than
ρ ∝ r−1.
The buildup of a cusp by a single merger with a compact
satellite, and in particular the dependence of the tidal mass
transfer on the slope of the halo density profile, have been
demonstrated using controlled N-body simulations of single
isolated mergers. The simulations also indicate that the cusp
remains stable under successive, similar mergers. We tried
to interpret in simple qualitative terms the detected phe-
nomenon of little mass transfer at α <
∼ 1, which gives rise to
the steepening to a cusp. For this purpose we used several
idealized toy-model considerations based on standard ap-
proximations, each valid in another limit and all pointing in
the same direction. It is evident that these are simplified ap-
proximations, and we expect each to be crudely valid at most
during part of the time and for limited parts of the satellite.
Nevertheless, we recall that these approximations sometime
provide useful qualitative estimates even where there is no
formal justification for them to be strictly valid. The fact
that in these different regimes the different approximations
all point to the phenomenon we see in the simulations may
be interpreted as a hint for a qualitative understanding of
this phenomenon. One should not use these approximations
for making accurate quantitative predictions away from their
formal range of validity, and our toy-model analysis is not
meant to replace a detailed and definitive exploration us-
ing exact analytic dynamical techniques, which should be
16
A. Dekel, J. Devor & G. Hetzroni
the target of subsequent work. However, at this point the
cosmological N-body simulations do not give us a clue for
the actual origin of the cusp, and the exact methods have
not provided a satisfactory understanding yet either. Given
that, the controlled merger simulations which highlight a
specific key effect, and the idealized approximations which
provide hints in the same direction, are useful for making
progress in the understanding of the phenomenon associ-
ated with the merger process, and in particular for putting
forward for further investigation a possible mechanism for
cusp formation.
Syer & White (1998) (and independently Nusser &
Sheth 1999 and Subramanian, Cen & Ostriker 2000) also ad-
dressed the profile resulting from mergers. Although they all
see the formation of a cusp in many cases, our results show
several differences which one should try to understand. Syer
& White implemented the more simplified model of tidal
stripping at resonance, ψ = 1 in our terms, ignoring the α
dependence (and in particular the compression at α < 1 and
the gradual stretching at α > 1). According to their algo-
rithm, the steepening or flattening of the profile is deter-
mined solely by whether the satellite inner density is higher
or lower than that of the halo. They argue that their pre-
scription allows a long-term survival of a relatively flat core
of α < 1 and it introduces a strong explicit dependence
on the power spectrum of fluctuations -- both in conflict
with the findings in cosmological simulations. For each of
several different power spectra they followed a sequence of
mergers and saw an apparent convergence to a different self-
similar profile. When we substitute the oversimplified recipe
ψ = 1 in eq. (16), we find that ∆α is always positive, for
any merger and at any r. Indeed, when trying to repeat the
Syer & White toy simulations with higher resolution and fol-
lowing more mergers we actually find that the profile does
not really converge to a stable cusp but rather continue to
steepen slowly towards α0 = 3. Only when using the im-
proved stripping prescription where ψ is decreasing with α
do we obtain convergence to a flatter asymptotic profile.
Furthermore, the asymptotic profile obtained via the revised
stripping recipe is more robust to the cosmological model.
This is partly because the effect of tidal compression drives
the profile to α >
∼ 1 independently of the fluctuation power
spectrum, and because the cosmological dependence of the
ultimate asymptotic slope is weakened by the fact that the
profile is determined by mergers of relatively large mass ra-
tio (see Dekel et al. 2003).
Somewhat puzzling is the finding of power-spectrum de-
pendence in the profile resulting from a sequence of N-body
merger simulations by Syer & White, and especially the ap-
parent slope flatter than α = 1 found in one of their cases. It
is puzzling because these simulations should have automat-
ically included the correct tidal effects. We can only suspect
that the main shortcomings of their N-body simulations are
the limited resolution within the cusp region (only 8000 par-
ticles in the whole halo) and perhaps the fixed narrow range
of satellite masses.
The gravitational processes leading to a cusp are mod-
eled in our analysis as tidal effects during the buildup of
the halo by a sequence of mergers. This picture is likely to
be valid in the CDM hierarchical clustering scenario, where
numerous sub-galactic haloes exist on all scales and are con-
tinuously merging (e.g., Klypin et al. 1999b; Moore et al.
1999a; Springel et al. 2001). This merger picture is confirmed
by a careful inspection of high resolution CDM simulations,
where a special effort is made to identify the merging clumps
which otherwise could have been easily missed (e.g., Wech-
sler, Dekel et al.
, in preparation). It is therefore clear that
cores, independently of how they form, cannot survive in a
pure CDM scenario without some modification -- they ef-
ficiently turn into cusps as a result of the mergers with the
relatively compact building blocks.
Nevertheless, a cusp, though somewhat flatter, is re-
ported to be seen also in simulations where the initial fluc-
tuations had less power on small scales, thus suppressing the
number of sub-galactic satellites and the associated merger
rate (Moore et al. 1999b; Avila-Reese et al. 2001; Bullock,
Kravtsov & Colin 2002). Further indications for the general-
ity of cusp formation comes from simulations by Huss, Jain
& Steinmetz (1999), who find that cusps also form as a result
of collapse from roughly spherical initial density perturba-
tions with random velocity perturbations, as well as from
the simulations of Alvarez, Shapiro & Martel (2002), who
find that cusps arise from the gravitational instability and
fragmentation of cosmological pancakes. One way to explain
this is by noticing that the asymptotic cusp formed in the
CDM scenario is driven by mergers with relatively massive
satellites (of typical mass ratio 1:3, see Dekel et al. 2003),
and realizing that such mergers do happen even when small-
scale power is suppressed and when pancakes fragment. It is
therefore possible that the cusp is actually driven by merg-
ers to a certain extent also in these cases. If it turns out that
a cusp also forms when the merger picture is strictly invalid,
it would imply that the gravitational processes involved in
the halo buildup somehow mimic a behavior similar to the
merger case. We note in particular that the tidal compres-
sion discussed above is expected to amplify density pertur-
bations and possibly make them behave in certain ways like
merging satellites.
More generally speaking, however, one should accept
the fact the halo buildup is a complex gravitational process,
whose different aspects can probably be modeled in more
than one way, e.g., as a violent relaxation process driven by
fluctuations in comparison with a sequence of mergers and
substructure accretion.8 There is also the intriguing possi-
bility that general statistical considerations in phase space
may provide some clue for the origin of a universal halo pro-
file (e.g., Taylor & Navarro 2001). Unfortunately, we cur-
rently know of no viable alternative to the tidal effects in
mergers as a simple model for the origin of α >
∼ 1 cusps in
the cosmological simulations. The idealized merger picture
provides one possible toy model, properly valid at least in
hierarchical clustering scenarios, within which the origin of
the cusp is understood in simple terms.
Our result implies a necessary condition for the survival
8 We encounter an analogous duality, for example, when one
manages to explain the origin of galactic spin alternatively via
tidal torque theory applied to shells and by summing up the or-
bital angular momenta in a cosmological sequence of mergers (see
Maller, Dekel & Somerville 2002).
of cores in haloes independent of their origin -- that satel-
lites should be prevented from adding mass to the halo cores.
This could be avoided in a CDM scenario if feedback pro-
cesses manage to puff up the small haloes and make them
disrupt before they merge with the halo cores. We have not
explicitly addressed in this paper the sufficient conditions for
the formation of cores, but one can imagine that a sequence
of mergers with low-density satellites, where the mass is pre-
dominantly deposited outside the inner halo region, would
indeed flatten the inner profiles (work in progress).
Supernova feedback effects are probably not strong
enough for turning cusps into cores in haloes with rotation
velocities higher than ∼ 100 km s−1, but the feedback may
be sufficient for lowering the inner densities in the small
halo progenitors such that when they merge they give rise
to cores. However, if the (so far inconclusive) observational
clues for cores in clusters of galaxies are confirmed, simple
supernova feedback is unlikely to provide a viable explana-
tion for their origin. In this case one may search for higher
efficiency in supernova feedback either due to microscopic
effects such as porosity in a muliphase ISM or due to hy-
pernova from very massive stars (Silk 2002). Alternatively,
one may appeal to stronger feedback mechanisms, perhaps
associated with radio jets from AGNs. This process may be
indicated by an observed correlation between AGN activ-
ity and bright galaxies in SDSS (Kauffmann et al. 2003, in
preparation), and it may be needed independently in order
to explain the missing baryons in big galaxies (Klypin, Zhao
& Somerville 2002) and in clusters. Otherwise, such large
cores may present a real challenge for the standard hierar-
chical clustering scenario.
Other processes may also contribute to the development
of halo cores. One proposed scenario is based on scattering
of inner halo particles off central, massive black holes (e.g.,
Merritt & Cruz 2001). This could be useful in explaining the
centres of early type galaxies, but one might suspect that it
would be difficult to apply this scenario on the scales of
∼ 10kpc corresponding to halo cores, which are much larger
than the scales associated with the black hole(s). A second
scenario addresses the heating of halo particles by gas clouds
spiraling in due to dynamical friction (El-Zant, Shlosman &
Hoffman 2002). This is based on assumed specific gradients
of velocity dispersion in the inner haloes, which may or may
not be valid. A third explanation is based on an efficient
angular-momentum transfer from a big rotating bar into the
halo (Weinberg & Katz 2002)./
However, other simulations indicate that this effect is
inefficient (Valenzuela & Klypin 2002). It has been demon-
strated (Sellwood 2002) that the Weinberg & Katz result is
likely to be an artifact of the unrealistically large bar as-
sumed, and the representation of this bar by a rigid body.
We stress again that, no matter what the origin of the core
might be, our analysis implies that such cores could survive
only if they are not perturbed by significant mass transfer
from merging compact satellites, which implies that small
CDM haloes must be puffed up before they merge into big-
ger haloes. We also stress that none of these scenarios seem
to be capable of explaining the formation of cores in haloes
on the scales of clusters of galaxies within the CDM scenario.
A possible caveat, which could in principle jeopardize
Halo cusp-core: tides in mergers
17
our results as well as the whole current standard paradigm
regarding the validity of N-body simulations in demonstrat-
ing robust cusp formation, has been raised by Katz & Wein-
berg (2002). They claim that all current cosmological sim-
ulations fail to recover the actual formation of a core in
dark-matter haloes because the N-body noise smears out
a delicate resonant reaction of halo-core orbits to the tidal
perturbation by the satellite. Based on linear analysis and
smoothed-potential simulations, they argue that an induced
global mode dominates the process and invalidates the stan-
dard dynamical friction approximation by Chandrasekhar.
As a result, satellites do not decay into the core but are
rather disrupted outside it, thus maintaining a flat core pro-
file at small radii. The dominance of this effect in realistic
cases has not been conclusively demonstrated. It is hard
to imagine how the resonances can remain so isolated and
stable in systems where phase mixing and broadening of res-
onances are expected. The linear analysis is far from being
valid in the nonlinear case in hand, and the over-smoothing
in their simulations may hide physical gradients that may
spread out the resonances (A. Klypin, private communica-
tion). Even if this resonant behavior turns out to be im-
portant in the N → ∞ limit, its relevance in real galaxies
is questionable, where the baryons are likely to introduce
clumpiness at a level comparable to the granularity in cur-
rent N-body simulations. Still, this is a caveat worth pursu-
ing.
The cusp/core problem is only one of the difficulties
facing galaxy formation theory within the CDM cosmology.
It turns out that other main problems can also be mod-
eled by tidal effects in mergers, and may also be resolved
by the inevitable feedback processes. For example, Maller
& Dekel (2002) addressed the angular-momentum problem,
where simulations including gas produce disks smaller than
the galactic disks observed (Navarro & Steinmetz 2000 and
references therein; Governato et al. 2002), and with a dif-
ferent internal distribution of angular momentum (Bullock
et al. 2001b; van den Bosch, Burkert & Swaters 2001). A
toy model has been constructed for the angular-momentum
buildup by mergers based on tidal stripping and dynamical
friction, which helps us understand the origin of the spin
problem as a result of over-cooling in satellites. A simple
model of feedback has then been incorporated, motivated by
Dekel & Silk (1986). This model can remedy the discrepan-
cies, and in particular explain simultaneously the low baryon
fraction and angular-momentum profiles in dwarf disk galax-
ies.
Various feedback effects may also provide the cure to
the missing dwarf problem, where the predicted large num-
ber of dwarf haloes in CDM can possibly match the observed
number of dwarf galaxies only if the mass-to-light ratio in
these objects is very high (Klypin et al. 1999b; Moore et
al. 1999a; Springel et al. 2001; Kochanek 2001). Bullock,
Kravtsov & Weinberg (2000), Somerville (2002) and Tully
et al. (2002) appeal to radiative feedback effects which pre-
vent the formation of small dwarfs after cosmological reion-
ization at z ∼ 7, Scannapieco, Ferrara & Broadhurst (2000)
and Scannapieco & Broadhurst (2001) address the destruc-
tive effect of outflows from one galaxy on neighboring pro-
togalaxies via ram pressure, and Dekel & Woo (2002) study
the role of supernova feedback in determining the relevant
18
A. Dekel, J. Devor & G. Hetzroni
global properties of dwarfs and larger low-surface-brightness
galaxies. We note that while the requirements from feedback
in explaining the dwarf-galaxy properties and the angular-
momentum problem are not too demanding, the solution to
the core problem requires that the dark-matter distribution
be affected by feedback, which is a non-trivial requirement.
Nevertheless, the successes of such toy models in match-
ing several independent observations indicate that they in-
deed capture the relevant basic elements of the complex
processes involved, and in particular that feedback effects
may indeed provide the cure to some or all the main prob-
lems of galaxy formation theory within the ΛCDM cosmol-
ogy that does so well on larger scales. The alternative solu-
tion involving Warm Dark Matter (e.g., Hogan & Dalcanton
2000; Avila-Reese et al. 2001; Bode, Ostriker & Turok 2001)
seems to still suffer to some extent from the core problem,
it may still fail to reproduce the angular-momentum profile
in galaxies (Bullock, Kravtsov & Colin 2002), and it may be
an overkill where the formation of dwarf galaxies is totally
suppressed once the inevitable feedback effects are included
(Bullock 2001). The speculative alternatives involving self-
interacting dark matter are even more problematic (Spergel
& Steinhardt 2000; Dave et al. 2001; Hennawi & Ostriker
2002).
ACKNOWLEDGMENTS
We acknowledge stimulating discussions with Itai Arad,
George Blumenthal, Andi Burkert, Doug Lin and Gary Ma-
mon. This research has been supported by the Israel Sci-
ence Foundation grant 213/02, by the US-Israel Bi-National
Science Foundation grant 98-00217, by the German-Israel
Science Foundation grant I-629-62.14/1999, and by NASA
ATP grant NAG5-8218.
REFERENCES
Adelberger, K.L., Steidel, C.C., Shapley, A.E. & Pettini,
M. 2002, ApJ, astro-ph/0210314
Alvarez, M.A., Shapiro, P.R. & Martel, H. 2002, AAS, 200,
4103
Avila-Reese, V., Colmn, P., Valenzuela, O., D'Onghia, E.
& Firmani, C. 2001, ApJ, 559, 516
Barnes, J.E. & Hernquist, L. 1991, ApJL, 370, L65.
Binney, J. & Tremaine, S. 1987, Galactic Dynamics Prince-
ton University Press), BT
Bode, P., Ostriker, J.P. & Turok, N. 2001, ApJ, 556, 93
Borriello, A. & Salucci, P. 2001, MNRAS, 323, 285
van den Bosch, F.C., Burkert, A. & Swaters, R.A. 2001,
MNRAS, 326, 1205
van den Bosch, F.C., Robertson, B.E., Dalcanton, J.J. &
de Blok, W.J.G. 2000, AJ, 119, 1579
Bullock, J.S. 2001, astro-ph/0111005
Bullock, J.S. Dekel, A., Kolatt, T.S., Kravtsov, A.V.,
Klypin, A.A., Porciani, C. & Primack, J.R. 2001b, ApJ,
555, 240
Bullock, J.S., Kolatt, T.S., Sigad, Y., Somerville, R.S.,
O.Y.
&
Zhao,
H.
2002,
2002, MNRAS
I. & Hoffman, Y.
Kravtsov, A.V., Klypin, A.A., Primack, J.R. & Dekel, A.
2001a, MNRAS, 321, 559
Bullock, J.S., Kravtsov A.V. & Colin P. 2002, ApJL, 564,
1
Bullock J.S., Kravtsov A.V. & Weinberg D.H. 2000, ApJ,
539, 517
Cole, S. & Lacey, C. 1996, MNRAS, 281, 716
Dave, R., Spergel, D.N., Steinhardt, P.J. & Wandelt, B.D.
2001, ApJ, 547, 574
de Blok, W.J.G., McGaugh, S.S., Bosma, A. & Rubin, V.C.
2001, ApJL, 552, L23
Dekel, A., Arad, I., Devor, J. & Birnboim, Y. 2003, ApJ,
588, in press (astro-ph/0205448)
Dekel, A. & Silk J. 1986, ApJ, 303, 39
Dekel, A. & Woo, J. 2002, astro-ph/0210454
El-Zant, A., Shlosman,
astro-ph/0103386
Geyer, M.P. & Burkert, A. 2001, MNRAS, 323, 988
Ghigna, S., Moore, B., Governato, F., Lake, G., Quinn, T.
& Stadel, J. 1998, MNRAS, 300, 146
Ghigna, S., Moore, B., Governato, F., Lake, G., Quinn, T.
& Stadel, J. 2000, ApJ, 544, 616
Gnedin,
(astro-ph/0108108)
Governato, F., Mayer, L., Wadsley, J., Gardner, J.P., Will-
man, B., Hayashi, E., Quinn, T., Stadel, J. & Lake, G.
2002, astro-ph/0207044
Hayashi, E., Navarro, J.F., Taylor, J.E., Stadel, J. & Quinn,
T. 2002, astro-ph/0203004
Hennawi, J.F. & Ostriker, J.P. 2002, ApJ, 572, 41
Hogan C.J. & Dalcanton J.J. 2000, PRD, 62
Huss, A., Jain, B. & Steinmetz, M. 1999, ApJ, 517, 64
Katz, N. & Weinberg, M.D. 2002, in Cozumel
Klypin, A., Gottlber, S., Kravtsov, A.V. & Khokhlov, A.M.
1999a, ApJ, 516, 530
Klypin, A.A., Kravtsov, A.V., Bullock, J.S. & Primack,
J.R. 2001, ApJ, 554, 903
Klypin, A., Kravtsov, A.V., Valenzuela, O. & Prada, F.
1999b, ApJ, 522, 82
Klypin A., Zhao H., Somerville R.S., 2002, ApJ, 573, 597
Kochanek, C.S. 2001, astro-ph/0108160
Lokas, E.L. & Hoffman, Y. 2000, ApJL, 542, L139
Maller A.H. & Dekel A. 2002, MNRAS, 335, 487
Maller A.H., Dekel A. & Somerville, R.S. 2002, MNRAS,
329, 423
Marchesini, D., D'Onghia, E., Chincarini, G., Firmani, C.,
Conconi, P., Molinari, E. & Zacchei, A. 2002, ApJ, in press
(astro-ph/0202075)
Merritt, D & Cruz, F. 2001, ApJL, 551, L41
Mihos, J.C. & Hernquist, L. 1996, ApJ, 464, 641
Moore, B., Ghigna, S., Governato, F., Lake, G., Quinn, T.,
Stadel, J. & Tozzi, P. 1999a, ApJL, 524, L19
Moore, B., Governato, F., Quinn, T., Stadel, J. & Lake, G.
1998, ApJL, 499, L5
Moore, B., Quinn, T., Governato, F., Stadel, J. & Lake, G.
1999b, MNRAS, 310, 1147
Navarro J.F., Eke, V.R. & Frenk C.S. 1996, MNRAS, 283,
L72
Navarro J.F., Frenk C.S. & White S.D.M. 1995, MNRAS,
275, 56
Navarro J.F., Frenk C.S. & White S.D.M. 1996, ApJ, 462,
563
Halo cusp-core: tides in mergers
19
Navarro J.F., Frenk C.S. & White S.D.M. 1997, ApJ, 490,
493
Navarro J.F. & Steinmetz M. 2000, ApJ, 538, 477
Nusser, A. & Sheth, R. 1999, MNRAS, 303, 685
Pearce, F.R., Thomas, P.A. & Couchman, H.M.P. 1993,
MNRAS, 286, 865
Power, C., Navarro, J.F., Jenkins, A., Frenk, C.S., White,
S.D.M., Springel, V., Stadel, J & Quinn, T. 2002, MNRAS,
submitted (astro-ph/0201544)
Salucci, P. 2001, MNRAS, 320, L1
Salucci, P. & Burkert, A. 2000, ApJL, 537, L9
Scannapieco, E. & Broadhurst, T. 2001, ApJ, 549, 28
Scannapieco, E., Ferrara, A. & Broadhurst, T. 2000, ApJL,
536, L11
Sellwood, J.A. 2002, astro-ph/0210079
Silk J., 2002, MNRAS, submitted (astro-ph/0212068)
Somerville, R.S. 2002, ApJL, 572, L23
Spergel, D.N. & Steinhardt, P.J. 2000, PRL, 84, 3760
Springel, V., White, S.D.M., Tormen, G. & Kauffmann, G.
2001, MNRAS, 328, 726
Subramanian, K., Cen, R., & Ostriker, J.P. 2000, ApJ, 538,
528
Syer, D. & White, S.D.M. 1998, MNRAS, 293, 337
Taylor, J.E. & Navarro, J.F. 2001, ApJ, 563, 483
Tully R.B., Somerville R.S., Trentham N., Verheijen,
M.A.W., 2002, ApJ, 569, 573
Valenzuela, O. & Klypin, A. 2002, astro-ph/0204028
Weinberg, M.D. & Katz, N. 2002, astro-ph/0110632
Zhao, H. 1996, MNRAS, 278, 488
3
2.5
2
α
1.5
1
0.5
0
0
50
100
150
200
250
Mergers
300
350
400
450
500
|
astro-ph/0209169 | 1 | 0209 | 2002-09-09T21:29:48 | New insights in high-resolution spectroscopy: a wide theoretical library of R=500000 stellar spectra | [
"astro-ph"
] | We present an extended theoretical library of over 800 synthetic stellar spectra, covering energy distribution in the optical range (lambda = 3500-7000 angstrom), at inverse resolution R=500000. The library, based on the ATLAS9 model atmospheres, has been computed with the SYNTHE code developed by R.L. Kurucz. The grid spans a large volume in the fundamental parameter space (i.e. effective temperature, surface gravity, metallicity), and can be profitably applied to different research fields dealing both with the study of single stars and stellar aggregates, through population synthesis models. A complementary project, in progress, will extend the wavelength range to the ultraviolet, down to 850 angstrom, at an inverse resolution of R=50000. | astro-ph | astro-ph |
Modelling of Stellar Atmospheres
IAU Symposium, Vol. xxx, xxxx
N. E. Piskunov, W. W. Weiss, D. F. Gray, eds.
New insights in high-resolution spectroscopy: a wide
theoretical library of R=500 000 stellar spectra
E. Bertone
Instituto Nacional de Astrof´ısica, Optica y Electr´onica, A.P. 51 y 216,
72000 Puebla, Mexico
A. Buzzoni
Telescopio Nazionale Galileo, A.P. 565, E-38700 Santa Cruz de La
Palma, Canary Islands, Spain; and Osservatorio Astronomico di Brera,
Milano, Italy
L.H. Rodr´ıguez-Merino and M. Ch´avez
Instituto Nacional de Astrof´ısica, Optica y Electr´onica, A.P. 51 y 216,
72000 Puebla, Mexico
Abstract.
We present an extended theoretical library of over 800 synthetic
stellar spectra, covering energy distribution in the optical range (λ =
3500 − 7000 A), at inverse resolution R=500 000. The library, based on
the ATLAS 9 model atmospheres, has been computed with the SYNTHE
code developed by R. L. Kurucz. The grid spans a large volume in the
fundamental parameters space (i.e. Teff , log g, [M/H]), and can be prof-
itably applied to different research fields dealing both with the study of
single stars and stellar aggregates, through population synthesis models.
A complementary project, in progress, will extend the wavelength range
to the ultraviolet, down to 850 A, at an inverse resolution of R=50 000.
1.
Introduction
New-generation spectrographs, at the major ground-based telescopes, have be-
gun to pour in the hard-disks of astronomers' computers an increasing mass of
high-quality spectroscopic data. Inverse resolutions as high as λ/∆λ = 100 000
can now be easily attained, at least for the brightest (V ∼< 15 mag) objects in
the sky, and this is pushing the observation of local and extragalactic stellar
systems to an ever unimagined resolution level. The outstanding performance
of instruments like UVES and VIRMOS at the ESO Very Large Telescope, or
HIRES and SARG at the Keck Observatory and Telescopio Nazionale Galileo,
respectively, urges therefore theoretical tools of comparable accuracy level in
order to consistently match and analyse such a huge amount of observational
data.
In this framework, and to help filling the gap, we undertook a long-term
project aimed at providing the community with a systematic theoretical library
1
A theoretical library of R=500 000 stellar spectra
2
of high-resolution stellar spectra (virtually the largest sample currently available
in the literature) in the optical range (λ = 3500 → 7000 A) and at an inverse
resolution of R = 500 000.
2. The optical grid of synthetic spectra
This first set of spectral energy distributions relied on the SYNTHE code of Ku-
rucz (1993), using ATLAS 9 model atmospheres (Kurucz 1995) as input. Com-
putations have been carried out at the Brera Observatory in Milan (Italy). The
input models follow the classical approximations of steady-state, homogeneous,
LTE, plane-parallel layers, with a microturbulence velocity of 2 km/s and a
mixing-length value ℓ/Hp = 1.25. They make use of an improved treatment
of convective overshooting in the calculation of the transfer equations (Castelli,
Gratton, & Kurucz 1997).
Over 46 million absorption lines are accounted for in the code, including
all atomic elements at different ionization states and the most important di-
atomic molecules. All line data were extracted from the Kurucz' (1992) original
database, with the major improvement of the TiO contribution, for which we
adopted the new list of lines computed by Schwenke (1998).
The wavelength interval spans the whole optical range, from the Balmer
break to Hα. All the passbands of the Lick/IDS spectrophotometric indices
(Worthey et al. 1994) are included, as well as the high-resolution index set
defined by Rose (1994). Spectra are sampled at variable wavelength step main-
taining a constant resolution (∆λ/λ = 2 × 10−6).
The library extends to effective temperatures as hot as Teff = 50 000 K, and
provides a suitable match to O → K spectral types. M stars, cooler than Teff <
4000 K, are however still lacking in our dataset given the missing contribution
of tri-atomic molecular opacity in the input model atmospheres (Kurucz 1992).
Gravity spans a wide range across the H-R diagram (5 ≥ log g ≥ 0), namely
from dwarf (MK class V) to supergiant stars (MK class I), while metallicity
accounts for stars at the two extreme edges of the Galaxy [Fe/H] distribution
with −3.0 < [M/H] < +0.3 as boundary limits. A total of 832 synthetic spectral
energy distributions (SEDs) have been computed, as summarized in Fig. 1 and
Table 1. An illustrative example of the model output, sampling the spectral
emission around the atomic Mg b triplet, is shown in Fig. 2 and 3, with varying
physical parameters.
A complementary project, which will extend our analysis to the ultravio-
let interval, between 850 -- 4750 A with an inverse resolution R = 50 000, is in
progress at INAOE and will be completed soon (Rodr´ıguez-Merino 2002). The
extended dataset consists of over 1000 SEDs covering the effective temperature
range between 3500 and 50 000 K at surface gravity log g = 1.0 → 5.0 and chem-
ical compositions [M/H] = +0.5, 0.0, −0.5 and −1.5 (Rodr´ıguez-Merino et al.
2001).
In Fig. 4 we display a subset of spectra computed for solar chemical
composition, log g = 5 and different effective temperatures.
A theoretical library of R=500 000 stellar spectra
3
Table 1.
The properties of the optical library of synthetic spectra.
Code:
Input models:
Wavelength range:
Resolution:
Wavelength step:
Wavelength points:
Total absorption lines:
Molecules:
Total number of spectra:
Effective temperature:
Surface gravity:
Metallicity:
SYNTHE (Kurucz 1993)
ATLAS9 (Kurucz 1995)
3500 ≤ λ ≤ 7000 A
R = λ/∆λ = 500 000
0.007 ≤ ∆λ ≤ 0.014 A
346 645
∼ 46 millions
C2, CN, CO, CH, NH, OH, MgH, SiH,
H2, SiO, TiO
832
4000 ≤ Teff ≤ 50000 K
0.0 ≤ log g ≤ 5.0
−3.0 ≤ [M/H] ≤ +0.3
Figure 1.
The optical library of synthetic high-resolution spectra.
Each panel shows the model grid for fixed metallicity. The whole set
consists of 832 SEDs.
A theoretical library of R=500 000 stellar spectra
4
Figure 2.
An illustrative sample of the synthetic spectra with solar
metallicity, spanning the wavelength region around the atomic Mg b
triplet. Increasing temperatures (botton to top) from Teff = 4000 →
8000 K at step of 1000 K are explored, with fixed gravity log g = 4.
Same as in Fig. 2, but with varying metallicity,
Figure 3.
for
Teff = 4000 K and log g = 4. Metallicity values are [M/H] =
−3.0, −2.0, −1.0, −0.3, 0.0, +0.3 (bottom to top, respectively).
A theoretical library of R=500 000 stellar spectra
5
Figure 4.
A sequence of high temperature spectra longward of the
Lyman limit from the UV library. The models have fixed gravity
(log g = 5) and solar chemical composition, while the effective tem-
perature increases (bottom to top) from 20 000 to 50 000 K at step of
10 000 K.
Figure 5. Difference in the pseudo-residual flux between the syn-
thetic and the observed spectrum of the Sun (Kurucz et al. 1984) in
the spectral region 3500 -- 7000 A. The features about 6300 and 6900 A
are due to telluric absorption bands in the observed spectrum.
A theoretical library of R=500 000 stellar spectra
6
3. The solar spectrum validation
A comparison with the solar spectrum represents a first mandatory test in order
to assess the degree of reliability of our theoretical library. We present in Fig. 5
the match of the Solar Flux Atlas by Kurucz et al. (1984) and the corresponding
synthetic solar spectrum, which we computed at similar resolution (i.e. R =
522 000). The input model atmosphere is from the ATLAS 9 and assumes Teff =
5777 K, log g = 4.4377, a microturbulent velocity of 1.0 km/s, and chemical
abundance according to Anders & Grevesse (1989).
The figure is made up by a dense sequence of positive and negative spikes,
which mainly track the residual flux differences in the absorption line profiles.
While positive spikes mainly originate from observed lines that are not present
in the synthetic model, the negative peaks come from theoretical absorption
lines that are too weak or undetected in the observed spectrum. Note, as a
stricking feature in the plot, that both positive and negative spikes tend to
decrease at longer wavelength, and no systematic drift is present in the data
distribution. Fig. 5 gives also a direct measure of the intrinsic uncertainty in
the input physics of the models; excluding the telluric bands, clearly affecting
the residual distribution about 6300 and 6900 A, the rms value in the plot is
±0.086.
References
Anders, E., & Grevesse, N. 1989, Geochim. Cosmochim. Acta, 53, 197
Castelli, F., Gratton, R. G., & Kurucz, R. L. 1997, A&A, 318, 841
Kurucz, R. L. 1992, in IAU Symp. 149, The Stellar Populations of Galaxies, ed.
B. Barbuy & A. Renzini, (Dordrecht: Kluwer), 225
Kurucz, R. L. 1993, CD-ROM No. 18, SYNTHE spectrum synthesis programs
and line data
Kurucz, R. L. 1995, CD-ROM No. 13, ATLAS9 Stellar Atmosphere Programs
and 2 km/s Grid, revised
Kurucz, R. L., Furenlid, I., Brault, L., & Testerman, L. 1984, National Solar
Observatory Atlas, Sunspot, New Mexico: National Solar Observatory
Rodr´ıguez-Merino, L.H. 2002, Ph. D. Thesis, in preparation
Rodr´ıguez-Merino, L. H., Ch´avez, M., Buzzoni, A., Bertone, E. 2001, in proc. of
New Quests in Stellar Astrophysics: A Link between Stars and Cosmol-
ogy, ed. M. Chavez, A. Bressan, A. Buzzoni & D. Mayya, (Dordrecht:
Kluwer), 39
Rose, J. A. 1994, AJ, 107, 206
Schwenke, D. W. 1998, Faraday Discuss., 109, 321
Worthey, G., Faber, S. M., Gonz´ales, J. J., & Burnstein, D. 1994, ApJS, 94, 687
|
0802.0462 | 1 | 0802 | 2008-02-04T18:15:37 | Life and times of dwarf spheroidal galaxies | [
"astro-ph"
] | We propose a cosmological scenario for the formation and evolution of dwarf spheroidal galaxies (dSphs), satellites of the Milky Way (MW). An improved version of the semi-analytical code GAMETE (GAlaxy Merger Tree & Evolution) is used to follow the dSphs evolution simultaneously with the MW formation, matching the observed properties of both. In this scenario dSph galaxies represent fossil objects virializing at z = 7.2 +/- 0.7 (i.e. in the pre-reionization era z > z_rei = 6) in the MW environment, which at that epoch has already been pre-enriched up to [Fe/H] ~ -3; their dynamical masses are in the narrow range M = (1.6 +/- 0.7) x 10^8 M_sun, although a larger spread might be introduced by a more refined treatment of reionization. Mechanical feedback effects are dramatic in such low-mass objects, causing the complete blow-away of the gas ~100 Myr after the formation epoch: 99% of the present-day stellar mass, M_* = (3 +/- 0.7) x 10^6 M_sun, forms during this evolutionary phase, i.e. their age is >13 Gyr. Later on, star formation is re-ignited by returned gas from evolved stars and a second blow-away occurs. The cycle continues for about 1 Gyr during which star formation is intermittent. At z = 0 the dSph gas content is M_g = (2.68 +/- 0.97) x 10^4 M_sun. Our results match several observed properties of Sculptor, used as a template of dSphs: (i) the Metallicity Distribution Function; (ii) the Color Magnitude Diagram; (iii) the decrement of the stellar [O/Fe] abundance ratio for [Fe/H] > -1.5; (iv) the dark matter content and the light-to-mass ratio; (v) the HI gas mass content. | astro-ph | astro-ph | Mon. Not. R. Astron. Soc. 000, 1–?? ()
Printed 12 February 2018
(MN LATEX style file v1.4)
Life and times of dwarf spheroidal galaxies
Stefania Salvadori1, Andrea Ferrara1 & Raffaella Schneider2
1SISSA/International School for Advanced Studies, Via Beirut 4, 34100 Trieste, Italy
2INAF/Osservatorio Astrofisico di Arcetri, Largo Enrico Fermi 5, 50125 Firenze, Italy
8
0
0
2
b
e
F
4
]
h
p
-
o
r
t
s
a
[
1
v
2
6
4
0
.
2
0
8
0
:
v
i
X
r
a
ABSTRACT
We propose a cosmological scenario for the formation and evolution of dwarf
spheroidal galaxies (dSphs), satellites of the Milky Way (MW). An improved version
of the semi-analytical code GAMETE (GAlaxy Merger Tree & Evolution) is used
to follow the dSphs evolution simultaneously with the MW formation, matching the
observed properties of both. In this scenario dSph galaxies represent fossil objects
virializing at z = 7.2 ± 0.7 (i.e. in the pre-reionization era z > zrei = 6) in the MW en-
vironment which at that epoch has already been pre-enriched up to [Fe/H]∼ −3; their
dynamical masses are in the narrow range M = (1.6 ± 0.7) × 108M⊙, although a larger
spread might be introduced by a more refined treatment of reionization. Mechanical
feedback effects are dramatic in such low-mass objects, causing the complete blow-
away of the gas ∼ 100 Myr after the formation epoch: 99% of the present-day stellar
mass, M∗ = (3 ± 0.7) × 106M⊙, forms during this evolutionary phase, i.e. their age is
> 13 Gyr. Later on, star formation is re-ignited by returned gas from evolved stars and
a second blow-away occurs. The cycle continues for about 1 Gyr during which star for-
mation is intermittent. At z = 0 the dSph gas content is Mg = (2.68 ± 0.97) × 104M⊙.
Our results match several observed properties of Sculptor, used as a template of dSphs:
(i) the Metallicity Distribution Function; (ii) the Color Magnitude Diagram; (iii) the
decrement of the stellar [O/Fe] abundance ratio for [Fe/H]> −1.5; (iv) the dark matter
content and the light-to-mass ratio; (v) the HI gas mass content.
Key words: stars: formation, population II, supernovae: general - cosmology: theory
- galaxies: evolution, stellar content -
1 BACKGROUND
The lack of a comprehensive scenario for the formation and
evolution of dwarf spheroidal galaxies (dSphs) contrast with
the large amount of available data for these nearby Local
Group satellites. Several authors have focused on different
aspects of the dSph evolution and on the observed properties
related with them, giving important contributions to our ac-
tual understanding of such puzzling objects. However, many
questions remain unanswered and models able to match si-
multaneously different observed properties are still missing.
Recent observations by Helmi et al. (2006) opened
new questions on the origin of dSph galaxies. By observ-
ing the stellar Metallicity Distribution Function (MDF) in
four nearby dSphs, they found a significant lack of stars
with [Fe/H]< −3. On the contrary, the Galactic halo MDF
shows a low [Fe/H]-tail extending down to [Fe/H]∼ −4
(Beers & Christlieb 2006) and even below (Christlieb et al.
2002, 2006; Frebel et al. 2005, Christlieb 2007). Does such
result imply that dSph and MW progenitors are different? Is
the dSph birth environment pre-enriched or does the Initial
Mass Function (IMF) behave differently in Galactic building
c(cid:13) RAS
blocks and in dSphs at earliest times? If the pre-enrichment
can solve the problem it must be completed before the be-
ginning of the star formation (SF) in dSphs .i.e. > 13 Gyr
ago. The dSph star formation histories (SFH) derived by the
analysis of the observed color-magnitude diagram (CMD),
in fact, show that all dSph have an ancient stellar popu-
lation formed more that 13 Gyr ago (Grebel & Gallagher
2004). Which mechanism can be responsible of such rapid
metal-enrichment?
In addition to these new open questions there is still
a crucial unresolved problem related to the dSph's SFHs.
In fact, although all dSphs display an ancient stellar pop-
ulation, their SFHs appear to be considerably different. In
most of them the bulk of stars is made by ancient stars
(> 10 Gyr old) i.e. their SF activity is concentrated dur-
ing the first Gyrs (Dolphin et al. 2005). However Fornax,
LeoII and Sagittarius show very different features: in these
objects the bulk of the stars was formed much less than
10 Gyr ago and their SF activity proceeds until z ∼ 1, or
even to lower redshifts (Grebel & Gallagher 2004). Carina,
finally, exhibits a clearly episodic SFH, with a pause of sev-
2
Stefania Salvadori, Andrea Ferrara & Raffaella Schneider
eral Gyrs after the old population formed and a massive
formation of stars younger than 10 Gyr (Smecker-Hane et
al. 1994, Hurley-Keller, Mateo & Grebel 1999). As pointed
out by Grebel & Gallagher (2004) such variety of SFHs can-
not be explained with a drop in the SF activity in response
to photoionization; only Carina in fact displays the gap ex-
pected if the SF were truncated during reionization and then
restarted. If reionization cannot explain the inferred variety
of SFHs, local processes, such as mechanical feedback, tidal
stripping, gas infall etc., must be invoked in order to produce
these observational features. However, it is difficult that lo-
cal processes can lead to different SFHs if the dSph host halo
mass has a "Universal" value, as predicted by Mateo et al.
(1998), Gilmore et al. (2007) and Walker et al. (2007), and
if the assumed initial gas-to-mass ratio is taken to be equal
to the cosmic mean value. Will more sophisticated observa-
tions and analysis reveal a spread in the dSph DM content?
If this is the case, different host halo mass could be more eas-
ily linked to different SFHs by internal physical processes.
In addition to these issues there are a number of related
aspects that deserve attention, as for example the observed
elemental abundance patterns (like the [α/Fe] ratio and the
s-elements abundance).
Despite all of these unresolved questions, the relevance
of stellar feedback for the evolution of dSph galaxies (Ferrara
& Tolstoy 2002 and references therein) can be considered a
milestone of our actual understanding of these objects, given
the general consensus among theoretical studies and the con-
tinuous confirmation by observational evidences. Given their
low DM content and therefore their shallow potential wells,
mechanical feedback driven by SN explosions has dramatic
effects in dSphs. Such intense mechanical feedback activity
is indirectly confirmed by the low light-to-dark matter ratios
observed in dSphs (Mateo et. al 1998, Gilmore et al. 2007).
Moreover, observations of neutral hydrogen (HI), reveal that
the Local Group dSphs are all relatively HI poor, suggesting
that maybe most of the gas have been removed from these
galaxies. The Sculptor dSph is one of the few with HI emis-
sion; a lower limit of MHI > 3 × 104M⊙ has been derived
by Carignan et al. (1998) using radio observations. In addi-
tion, winds are presumably metal-enhanced, as suggested by
several theoretical studies (Vader 1998; MacLow & Ferrara
1999; Fujita et al. 2004) and by recent X-rays observations
of the starburst galaxy NGC 1569 (Martin, Kobulnicky &
Heckman 2002).
Many theoretical works have attacked some aspects of
the dSph evolution separately:
• Origin & DM content of dSphs: Bullock et al. (2001);
Kravtsov et al. (2004); Ricotti & Gnedin (2005); Gnedin &
Kravtsov (2006); Read, Pontzen & Viel (2006); Moore et al.
(2006); Metz & Kroupa (2007).
• DSph MDFs: Ripamonti et al. (2006); Lanfranchi &
Matteucci (2007)
• SFH & abundance ratios of dSphs: Ikuta & Arimoto
(2002); Fenner et al. (2006); Lanfranchi & Matteucci (2007);
Stinson et al. (2007).
• Stellar feedback & dSph gas content: Ferrara & Tolstoy
(2000); Tassis et. al (2003); Fujita et al. (2004); Lanfranchi
& Matteucci (2007).
In this study we analyze the formation and evolution
of dSph galaxies, satellites of the MW, in their cosmological
context by using the improved GAlaxy MErger Tree & Evo-
lution code (GAMETE), which allows to build-up the SFH
and chemical enrichment of the MW along its hierarchical
merger tree (Salvadori, Schneider & Ferrara 2007, hereafter
SSF07). This approach gives a self-consistent description of
the dSphs evolution and MW formation: dSphs form out
from their natural birth environment, the Galactic Medium
(GM), whose metallicity evolution is completely determined
by the history of SF and by mechanical feedback processes
along the build-up of the Galaxy. The star formation and
mechanical feedback efficiencies of dSphs are assumed to be
the same as for all the Galactic building blocks and cali-
brated to reproduce the observable properties of the MW.
The model allows to predict the following observable prop-
erties of a typical dSph galaxy: (i) the formation epoch; (ii)
the MDF; (iii) the SFH and the derived CMD diagram; (iv)
the stellar [O/Fe] abundance ratio with respect to [Fe/H];
(v) the DM content; (vi) the stellar-to-mass ratio; (vii) the
final gas content. These are compared with observations of
Sculptor, which represents the best studied nearby dSph.
A global scenario for the formation and evolution of dSph
galaxies is presented.
The plan of the paper is the following: a recap of the
general properties of the improved code GAMETE is pre-
sented in Sec. 2. The life of a dSph galaxy is described in
subsequent sections: the birth environment and the selec-
tion criteria from the MW building blocks are presented in
Sec. 3; the evolution until the blow-away, driven by mechan-
ical feedback, is described in Sec. 4; the subsequent and final
stages of the dSph life are traced in Sec. 5. Sec. 6 is devoted
to a comparison between model results and some observa-
tional properties of the Sculptor dSph. Finally, a summary
and discussion of the main results is given in Sec. 7.
2 MODEL DESCRIPTION
In this Section, we first summarize the main features of the
model introduced in SSF07; following that we discuss the
modifications and improvements made for the purpose of
this work.
2.1 Summary of the model
The semi-analytic code GAMETE used in SSF07 allows to
follow the star formation history and chemical enrichment
of the MW throughout its hierarchical merger tree. Its main
features can be summarized along the following points (for
detailed explanations see SSF07). The code reconstructs the
hierarchical merger history of the MW using a Monte Carlo
algorithm based on the extended Press & Schechter theory
(Bond et al. 1991; Lacey & Cole, 1993); it adopts a binary
scheme with accretion (Cole et al. 2000, Volonteri, Haardt
& Madau 2003) to decompose the present day MW dark
matter halo (MMW ∼ 1012M⊙ Binney & Merrifield 1998)
into its progenitors, running backward in time up to z = 20.
At any time a halo of mass M0 can either loose part of
its mass (corresponding to a cumulative fragmentation into
haloes with M < Mres) or loose mass and fragment into two
progenitor haloes with random masses in the range Mres <
M < M0/2. The mass below the resolution limit accounts
for the Galactic Medium (GM) which represents the mass
c(cid:13) RAS, MNRAS 000, 1–??
reservoir into which haloes are embedded. During the star
formation history of the MW, the progenitor haloes accrete
material from the GM and virialize out of it. We assume
that feedback effects rapidly suppress star formation in the
first mini-haloes and that only Lyα cooling haloes (Tvir >
104K) contribute to the star formation history and chemical
enrichment of the Galaxy. This motivates our choice of a
resolution mass Mres = M4(z)/10, where
M4(z) = M (Tvir = 104K, z) ∼ 108M⊙(cid:16) 10
1 + z(cid:17)3/2
(1)
represents the halo mass corresponding to a virial equilib-
rium temperature Tvir = 104K at a given redshift z. At the
highest redshift of the simulation, z ≈ 20, the gas present
in virialized haloes is assumed to be of primordial compo-
sition. The star formation rate is taken to be proportional
to the mass of gas. Following the critical metallicity sce-
nario (Bromm et al. 2001; Omukai 2000, Omukai et al. 2005;
Schneider et al. 2002, 2003, 2006; Bromm & Loeb 2004)
we assume that low-mass star formation is triggered by the
presence of metals (and dust) exceeding Zcr = 10−5±1Z⊙.
When the gas in star forming haloes has a metallicity
Z ≤ Zcr Pop III stars form with a reference mass value
of mpopIII = 200M⊙. If on the contrary Z > Zcr, Pop II/I
stars form according to a Larson initial mass function (IMF):
Φ(m) =
dN
dm
∝ m−1+x exp(−mcut/m),
(2)
with x = −1.35, mcut = 0.35M⊙ and m in the range
[0.1−100]M⊙ (Larson 1998). Due to the lack of spatial infor-
mation, when metals and gas are returned to the interstel-
lar medium (ISM) through stellar winds and SN explosions,
they are assumed to be instantaneously and homogeneously
mixed with the ISM. The same instantaneous perfect mixing
approximation is applied to material ejected out of the host
halo into the external GM.
2.2 New features
We now discuss the additional features we have incorporated
in the model. The aim of introducing these new physics is
to obtain a more complete description of the evolution of a
single dSph galaxy. These can be summarized as follows:
• Infall rate. The gas in newly virialized haloes is accreted
with an infall rate given by
dMinf
dt
= A(cid:18) t
tinf (cid:19)2
exp(cid:18)−
t
tinf(cid:19).
The selection of this particular functional form has been
guided by the results of simulations presented in Keres et
al. (2005). For reasons that will be clarified in Sec. 6.1, the
infall time is assumed to be proportional to the free-fall time,
tinf = tf f /4 where tf f = (3π/32Gρ)1/2, G is the gravita-
tional constant, and ρ is the total (dark + baryonic) mass
density of the halo. The normalization constant is set to be
A = 2(Ωb/ΩM )M/tinf so that for t → ∞ the accreted gas
mass reaches the universal value Minf (∞) = (Ωb/ΩM )M .
No infall is assumed after a merging event i.e. all the gas is
supposed to be instantaneously accreted. Hydrodynamical
simulations in fact show that galaxy mergers can drive sig-
nificant inflow of gas raising the star formation rate by more
c(cid:13) RAS, MNRAS 000, 1–??
Life and times of dwarf spheroidal galaxies
3
than an order of magnitude (Mihos & Hernquist, 1996 and
references therein).
• Finite stellar lifetimes. We follow the chemical evolution
of the gas taking into account that stars of different masses
evolve on characteristic time-scales (Lanfranchi & Matteucci
2007). The rate at which gas is returned to the ISM through
winds and SN explosions is computed as:
dR(t)
dt
= Z 100M⊙
m1(t)
(m − wm(m))Φ(m)SF R(t − τm)dm,
(4)
where τm = 10/m2 Gyr is the lifetime of a star with mass
m, wm is the remnant mass and m1(t) the turnoff mass
i.e. the mass corresponding to t = τm. Similarly, the total
ejection rate of an element i , newly synthesized inside stars
(first term in the parenthesis) and re-ejected into the ISM
without being re-processed (second term), is
dYi(t)
= Z 100M⊙
(5)
dt
m1(t)
(cid:2)(m − wm(m) − mi(m, Z))×
Zi(t − τm) + mi(m, Z)(cid:3)Φ(m)SF R(t − τm)dm,
where mi(m, Z) is the mass of element i produced by a star
with initial mass m and metallicity Z and Zi(t − τm) is the
abundance of the i − th element at the time t − τm. The SN
rate is simply computed as
dNSN (t)
dt
= Z 40M⊙
m1(t)>8M⊙
Φ(m)SF R(t − τm)dm.
(6)
We used the grid of values wm(m) and mi(m, Z) by Heger
& Woosley (2002) for 140M⊙ < m < 260M⊙, Woosley &
Weaver (1995) for 8M⊙ < m < 40M⊙ and van der Hoek &
Groewengen (1997) for 0.9M⊙ < m < 8M⊙.
• Mechanical feedback. Assuming a continuous mass loss
prescription (Larson 1974), the mass of gas ejected into the
surrounding GM is regulated by the equation,
e = ESN
Mej v2
1
2
where
ESN = ǫwNSN hESN i
(7)
(8)
is the kinetic energy injected by SN-driven winds and v2
e =
GM/rvir = 2Eb/M is the escape velocity of the gas from a
halo with mass M and binding energy Eb given by (Barkana
& Loeb 2001)
(3)
Eb =
1
2
GM 2
rvir
= 5.45 × 1053erg(cid:16) M8
h−1(cid:17)5/3(cid:16) 1 + z
10 (cid:17) h−1.
In eq. 8, ǫw is a free parameter which controls the conversion
efficiency of SN explosion energy in kinetic form, NSN is the
number of SN, and hESN i is the average explosion energy;
the latter quantity is taken to be equal to 2.7 × 1052erg for
Pop III stars and to 1.2×1051erg for Type II SNe. Differenti-
ating eq. 7 we find that the gas ejection rate is proportional
to the SN explosion rate,
dMej
dt
=
2ǫwhESN i
dNSN
v2
e
dt
.
(9)
• Differential winds. The gas ejected out of the host halo
is assumed to be metal-enhanced with respect to the star
forming gas. According to Vader (1986), who studied SN-
driven gas loss during the early evolution of elliptical galax-
4
Stefania Salvadori, Andrea Ferrara & Raffaella Schneider
ies, the SN ejecta suffer very limited mixing before they
leave the galaxy, playing a minor role in its chemical evo-
lution. Such hypothesis implies different ejection efficiency
for gas and metals. This result has been later confirmed by
numerical studies (Mac Low & Ferrara, 1999; Fujita et al.
2004). Adopting a simple prescription, we assume that the
abundance of the i − th element in the wind is proportional
to its abundance in the ISM, Zw
, and we take
α = 10 only for newly virialized haloes (M < 109M⊙) oth-
erwise α = 1.
i = αZISM
i
For any star forming halo of the MW hierarchy, we
therefore solve the following system of differential equations:
dM∗
dt
dMg
dt
dMZi
dt
= SF R = ǫ∗
Mg
tf f
,
= −SF R +
dR
dt
+
= −ZISM
i
SF R +
−
dMej
dt
,
(10)
(11)
+ Zvir
i
dMinf
dt
− Zw
i
dMej
dt
. (12)
dMinf
dt
dYi
dt
The first equation is the star formation rate; Mg is the mass
of cold gas inside haloes, ǫ∗ the free parameter which controls
the star formation efficiency and tf f the free-fall time. The
second equation describes the mass variation of cold gas:
it increases because of gas infall and/or returned from stars
and decreases because of star formation and gas ejection into
the GM. The third equation, analogous to the second one,
regulates the mass variation of an element i; ZISM
, Zvir
,
and Zw
i are the abundance of the i − th element in the ISM,
in the infalling gas (i.e. in the hot gas at virialization), and
in the wind, respectively.
i
i
2.3 Model parameters
The model has six free parameters: ǫ∗, ǫw, Zcr, mP opIII, tinf
and α. We use the observed global properties of the MW in
order to fix ǫw and ǫ∗. To do so we compare the results of
the simulations at redshift z = 0 with: (i) the gas metallicity
Zgas ∼ Z⊙; (ii) the stellar metallicity Z∗ ∼ Z⊙; (iii) the
stellar mass M∗ ∼ 6 × 1010M⊙; (iv) the gas-to-stellar mass
ratio Mg/M∗ = 0.1; (v) the baryon-to-dark matter ratio
fb = 0.07; (vi) the GM metallicity ZGM ∼ 0.25Z⊙. The last
quantity has been estimated using the observed value for
[O/H] in high-velocity clouds (Ganguly et al. 2005) which
are supposed to be gas leftover from the Galactic collapse
and currently accreting onto the disk. The MDF of Galactic
halo stars observed by Beers & Christlieb (2006) is instead
used in order to fix the values of Zcr and mP opIII (Fig. 1, left
panel). As we will discuss in detail in Sec. 6.1, the additional
parameters tinf and α are fixed to match the Sculptor MDF
without altering the MW properties. Our fiducial model is
characterized by the following parameters values⋆: ǫ∗ = 1,
ǫw = 0.002, Zcr = 10−3.8Z⊙, mP opIII = 200M⊙, tinf =
tf f (zvir)/4 and α = 10.
The comparison between the best-fitting model and the
observed MDF is shown in the left panel of Fig. 1. The model
provides a good fit to the data; in particular, the selected
Zcr value allows to reproduce the peculiar MDF cut-off al-
though it cannot account for the two isolated hyper iron-
poor stars ([Fe/H]= −5.3 and [Fe/H]= −5.7) and for the
recent observations of a star with [Fe/H]= −4.8, the first
detection in what was previously called "metallicity desert"
(Christlieb 2007). These stars can only be reproduced as-
suming Zcr ≤ 10−6Z⊙, at the expenses of overpredicting the
number of stars in the range −5.3 < [Fe/H]< −4, loosing the
agreement with the observed MDF cut-off (see SSF07 for a
detailed analysis of the dependence of the predicted MDF on
Zcr and mP opIII ). We find that most of the iron-poor stars
([Fe/H]< −2.5) form in haloes which originally contain gas
of primordial composition but which accrete material from
the GM, Fe-enhanced by previous SN explosions. The initial
[Fe/H] abundance within a virializing halo is then fixed by
the corresponding GM iron abundance at the virialization
redshift. The evolution of GM iron and oxygen abundance
predicted by the fiducial model is also shown in Fig.1 (mid-
dle panel).
3 THE BIRTH ENVIRONMENT
Once the fiducial model has been fixed, its parameters are
used to solve the system of equations (10)-(12) for all the
progenitor haloes of the MW. The next point to address is
the selection criteria to identify dSph galaxies among various
MW progenitors. We use two criteria: the first is based on
dynamical arguments and the second on reionization.
We want to select virializing haloes which could become
dSph satellites. Using N-body cosmological simulations, Die-
mand, Madau & Moore (2005) show that in present-day
galaxies, haloes corresponding to rare high-σ(M, z) density
peaks† are more centrally concentrated. The probability of
a protogalactic halo to become a satellite increases if it is as-
sociated with lower-σ density fluctuations. This result, com-
bined with the fact that at each redshift 95% of the to-
tal dark matter lies in haloes with mass M < M2σ, which
correspond to < 2σ fluctuations, suggest that most satel-
lites originate from such density peaks. Therefore, we select
dSph candidates from haloes with masses M4 < M < M2σ.
In Fig. 1 (right panel) we show the redshift evolution of
M4(z), defined in eq. 1 as M4(z) = 108M⊙[(1 + z)/10]−3/2,
and of the halo masses corresponding to 1 − 3 σ(M, z) den-
sity peaks. Note that, in addition, the adopted dynamical
criterion can be used to set an upper limit to the dSph can-
didates formation epoch of zvir < 9 (see Fig. 1, right panel).
The second criterion is based on reionization. During
this epoch, the increase of the Inter Galactic Medium (IGM)
temperature causes the growth of the Jeans mass and conse-
quent suppression of gas infall in low-mass objects. In partic-
ular, cosmological simulations by Gnedin (2000) show that
below a characteristic halo mass-scale the gas fraction is
drastically reduced compared to the cosmic value. We adopt
a simple prescription and assume that after reionization the
formation of galaxies with circular velocity vc < 30 km/s
⋆ The difference of ǫ∗, ǫw values with respect to those found in
SSF07 is a result of model improvements. Note however that the
integrals of the star formation rate and the gas ejection rate from
progenitor haloes remain unaltered.
† The quantity σ(M, z) represents the linear theory rms density
fluctuations smoothed with a top-hat filter of mass M at redshift
z.
c(cid:13) RAS, MNRAS 000, 1–??
Life and times of dwarf spheroidal galaxies
5
Figure 1. Left panel: Cumulative number of stars below a given [Fe/H] observed by Beers & Christlieb (2006) in the Galactic halo, with
the inclusion of the three hyper-iron poor stars (Christlieb et al. 2002, 2006; Frebel et al. 2005, Christlieb 2007) (points), and simulated
by using the fiducial model (histogram). The histogram is the average value over 100 realizations of the merger tree re-normalized to
the number of observed stars with [Fe/H]≤ −2. The shaded area represents ±1σ errors. Middle: Evolution of GM iron (solid line) and
oxygen (dotted line) abundance. Lines are the average values over 100 realizations of the merger tree. The shaded area delimits the ±1σ
dispersion region for [Fe/H]. The point is the measured [O/H] in high-velocity clouds (Ganguly et al. 2005). Right: Evolution of the mass
corresponding to 1 (long dashed line), 2 (short dashed line) and 3σ(M, z) (dotted line) density peaks; the solid line show the evolution
of M4(z) (eq.1), the dotted-dashed line the evolution of M30(z). The selected reionization redshift zrei = 6 is also shown in the Figure.
The shaded area delimits the region M4(z) < M < M2σ for z > zrei, M4(z) < M < M2σ for z < zrei.
is suppressed, i.e. we assume that when z < zrei haloes
with masses below M30(z) = M (vc = 30 km/s , z) have
no gas (for a thorough discussion of radiative feedback see
Ciardi & Ferrara 2005 and Schneider et al. 2007). Since
M30(z) > M4(z) (see the right panel of Fig. 1) and the
probability to form a newly virialized halo with M > M30
is very low, the second criterion implies that the formation
of dSph candidates is unlikely to occur below zrei. As we
will discuss in Sec. 6.4 this fact may have important conse-
quences as, for example, the possible existence of a universal
mass of dSph host haloes.
Following the two above criteria, dSph candidates can
only form in the redshift range zrei < zvir < 9. From the
middle panel of Fig. 1 it is evident that in this redshift range
the mean GM iron abundance is −2.5 <
∼ − 3. This
implies that the birth environment of dSph candidates is
pre-enriched to [Fe/H] values consistent with those implied
by the MDF observations of Helmi et al. (2006).
∼ [Fe/H] <
In what follows, we will present the results obtained by
our fiducial model averaged over 100 different realizations
of the hierarchical merger tree of the MW. In each single
realization, dSph candidates are selected from haloes with
masses and redshifts corresponding to the shaded area in the
right panel of Fig. 1, that is M4 < M < M2σ for z > zrei
and M30 < M < M2σ otherwise. Their subsequent evolution
is followed in isolation with respect to the forming Galaxy:
they neither merge nor accrete material from the GM. Fol-
lowing Choudhury & Ferrara (2006), we vary the reioniza-
tion redshift within the range 5.5 < zrei < 10. The total
number of dSph candidates depends on zrei and it is typi-
cally larger than the number of observed ones. Therefore, for
each zrei, it is necessary to randomly extract a sub-sample
in order to match the total number of known MW satellites,
∼ 15. The average properties of a dSph galaxy presented
in the following Sections refer to the case zrei = 6 (see the
discussion in Sec. 6.1) and are obtained averaging over the
c(cid:13) RAS, MNRAS 000, 1–??
selected satellites from all the 100 realizations of the MW
merger tree (about ∼ 2000 objects).
4 FEEDBACK-REGULATED EVOLUTION
The life of a dSph is very violent in the first hundred Myr,
due to mechanical feedback effects which are more intense
in low mass objects. The evolution of the mass of cold gas
(eq. 11) helps in understanding this rapid evolution. Fig. 2
shows the evolution of several properties of an average dSph
galaxy (M = 1.6×108M⊙) that virialized at zvir = 7.2, with
respect to the formation time (age) T = t − tvir. Three main
evolutionary phases are identified in the Figure, depending
on the dominant physical processes.
An increasing fraction of cold gas is collected during
Phase I (T < 40 Myr) dominated by the infall rate. The
mass of the infalling gas rapidly increases during this epoch,
reaching a maximum when T = tinf ∼ 25 Myr. The mass of
ejected and returned gas start to contribute to eq. (11) only
when the most massive SNe of 40M⊙ explode‡, ∼ 6 Myr
from the formation of the first stellar generation. There-
after, the mass of ejected and returned gas rapidly grow,
due the raising number of SNe and evolving low mass stars.
The ISM metallicity and iron abundance evolve accordingly
during this phase: they are steadily equal to the values of
the infalling gas (Zvir ∼ 10−3Z⊙, [Fe/H]∼ −2.9) before the
first SNe explodes and then rapidly increase.
During Phase II (40 Myr <
∼ T <
∼ 60 Myr) the gain of
‡ Stars with 40M⊙ < m < 100M⊙ are predicted to collapse
to black holes (Woosley & Weaver, 1995) while massive Pop III
stars cannot be produced in dSph galaxies since their birth envi-
ronment is pre-enriched to Zvir ∼ 10−3Z⊙ > Zcr = 10−3.8Z⊙
6
Stefania Salvadori, Andrea Ferrara & Raffaella Schneider
Figure 2. Evolution of Mg/M (solid line), M∗/M (long dashed
line), Minf /M (dotted line), Mej /M (long-short dashed line),
R/M (short dashed line), MZ /Mg (dotted-short dashed line),
MF e/Mg (dotted-long dashed line) for a typical dSph with to-
tal mass M = 1.6 × 108M⊙, that virialize at redshift zvir = 7.2,
with respect to its age T = t − tvir. The three main evolutionary
phases (Phase I T < 40 Myr, Phase II 40 Myr < T < 60 Myr,
Phase III T > 60 Myr) are also shown.
Figure 3. SFR of a typical dSph with total mass M = 1.6 ×
108M⊙, which virialize at redshift zvir = 7.2, as a function of
its age T = t − tvir . The arrow shows the occurrence of the first
blow-away.
by numerical simulations (Fujita et al. 2004). Eventually at
T ∼ 100Myr our template dSph is a gas free system.
cold gas by infall is mostly used to form stars and Mg re-
mains constant. Note that the M∗/M curve in Fig. 2 repre-
sents the total stellar mass at time T .
Finally, during Phase III (T >
∼ 70 Myr), the mass of the
ejected gas overcomes the infalling gas and Mg starts to de-
crease. Because of the metal-enhanced wind prescriptions
MZ and MF e should in principle decrease earlier and faster
than Mg. This is the case for MZ: in Fig. 2 the metallic-
ity is a slowly decreasing function both during Phase II and
MZ < Mg. Conversely, the MF e/Mg
Phase III so that
ratio is enhanced during these epochs: the mass of newly
synthesized iron released by a SN with a m = 12M⊙ progen-
itor is ∼ 2 orders of magnitude bigger than for m = 40M⊙
(Woosley & Weaver, 1995)§; for this reason, when lower
mass SNe evolve, a larger amount of iron is injected into
the ISM and the second right-term in eq. 12 can contrast
the high ejection rate.
When T ∼ 100 Myr the mass of gas lost due to winds
becomes larger than the remaining gas mass and Mg drops
to zero. During this blow-away metals and iron are also
ejected out of the galaxy. Moreover, since SN explosions con-
tinue at subsequent times, even the infalling gas can rapidly
acquire enough energy to escape the galaxy. The infall is first
reversed and in few Myr, when the remaining mass of hot
gas has blowed-away, definitively stopped. The occurrence
of reversal infall in high-redshift dwarf galaxies is confirmed
§ This result is virtually independent of the initial metallicity
of the star. In the same mass range, the total mass of metals
produced remains constant.
5 BEYOND BLOW-AWAY
In Fig. 3 we show the star formation rate (SFR) of a typi-
cal dSph galaxy as a function of its age. The highest peak
corresponds to the star formation activity during the first
100 Myr i.e. before the blow away. After the blow-away
the galaxy remains gas free and star formation is suddenly
halted i.e. SF R(T ) = 0. The gas returned by evolved stars
represents the only source of fresh gas for the galaxy af-
ter the blow away. However, until the latest SN explodes,
this low mass of gas is easily ejected outside the galaxy by
SN winds (Fig. 2); the dSph remains dormant (SF R = 0)
for the subsequent ∼ 150 Myr (this time-lag corresponds to
the life time of the lowest m = 8M⊙ SN progenitor formed
just before the blow-away). Observationally, mass loss from
evolved stars has been invoked by Carignan et al. (1998)
in order to explain the detection of neutral hydrogen (HI)
associated with the Sculptor dSph galaxy.
After the latest SN explosion, the return rate dR/dt
by evolved stars with m < 8M⊙ becomes the only non-zero
term of eq. 11 and the dSph enters a rejuvenation phase: the
recycled gas is collected into the galaxy and star formation
starts again.
From the beginning of the rejuvenation phase the sub-
sequent evolution of the galaxy proceeds like in the first 100
Myr of its life. However (Fig. 3) SFR is now more than 2
orders of magnitude lower than before the first blow-away,
due to the paucity of returned gas. Almost 100 Myr later
the galaxy is drained of the whole mass of gas and metals:
a new blow-away has occurred and the cycle starts again.
In Fig. 3 we note that the repetition of blow-away and
c(cid:13) RAS, MNRAS 000, 1–??
rejuvenation phases causes an intermittent SF activity with
a typical blow-away separation of ∼ 150 Myr. Such burst-
like SFH is similar to that inferred from the CMD observed
in dSph galaxies such as Carina (Smecker-Hane et al. 1994)
although the typical duration of active and quiescent phases
is ∼ (1 − 2) Gyr. In agreement with the present work, recent
simulations for the collapse of an isolated dwarf galaxy (Stin-
son et al. 2007), show that feedback effects cause a periodic
SF activity, with a typical duration of active and quiescent
phases of ∼ 300 Myr.
Fig. 3 also shows that about 1 Gyr after the formation
of the galaxy, this burst-like SFH ends. Because of the grad-
ually smaller mass of gas returned by stars, fewer SNe are
produced during this epoch. For this reason, gas ejection is
less efficient and can be easily counteracted by the continu-
ous input of returned gas.
Although the SF activity continues until the present-
day, the mass of stars formed after the first blow-away (M∗ ∼
2×105M⊙) is only 1% of the mass of stars formed before the
blow-away (M∗ ∼ 2 × 107M⊙). Consistent with this result,
the analysis of the dSph CMD diagrams by Dolphin et al.
(2005) shows that dSphs typically form most of their stars
over 10 Gyr ago.
After the first blow-away subsequent stellar generations
formed out of gas recycled by low mass star. The character-
istic iron-abundance of this gas is [Fe/H]∼ −1.5, as can be
inferred using the results of van der Hoek & Groewengen
(1997).
6 OBSERVABLE PROPERTIES
In the following Sections we will compare our numerical re-
sults for dSph galaxies with the most relevant observations.
Given the amount of available data we take Sculptor as the
best case of a dSph template; however, the validity of model
results is general.
6.1 Metallicity distribution function
In this Section we analyze the Metallicity Distribution Func-
tion (MDF) of Sculptor, i.e. the number of relic stars as a
function of their iron abundance [Fe/H], a quantity com-
monly used as a metallicity tracer. In Fig. 4 we compare
the MDF observed by Helmi et al. (2006) with the simu-
lated one, normalized to the total number of observed stars
(513). The theoretical MDF is obtained as follows: we adopt
a reionization redshift of zrei = 6; given this choice, the
average number of dSph candidates in each realization is
Ntot ∼ 200, among which 10% become MW satellites, hence
naturally matching the number of observed satellites. A
higher reionization redshift of zrei = 8.5 would reduce the
number of dSph candidates to Ntot ∼ 5, well below the ob-
served value. This allows to put a solid constraint on the
reionization redshift of zrei < 8.5. As can be inferred from
the Figure, the model shows a good agreement with the ob-
served MDF, particularly for [Fe/H]< −1.5. A marginally
significant deviation is present at larger [Fe/H] values.
We have already discussed in the previous Section that
the bulk of stars (∼ 99%) in a dSph galaxy is formed during
the first 100 Myr of its life when [Fe/H]< −1.5. Essentially,
stars formed after the first blow away ([Fe/H]> −1.5), are
c(cid:13) RAS, MNRAS 000, 1–??
Life and times of dwarf spheroidal galaxies
7
unnoticeable in the normalized MDF. For this reason the
physical processes regulating the MDF shape are mostly
those responsible for the cold gas mass evolution analyzed in
Sec. 4. We can use the evolution of MF e/Mg shown in Fig. 2
in order to convert time in [Fe/H] variable and identify the
three main evolutionary phases into the MDF.
∼ [Fe/H] <
We find that stars with [Fe/H] <
∼ − 2 formed during
the infall-dominated Phase I; the MDF shape at low [Fe/H]
values is then essentially regulated by the functional form of
the infall rate. Stars with −2 <
∼ − 1.6 (around the
MDF maximum) are formed during Phase II i.e. when the
mass of cold gas remains approximately constant; the maxi-
mum of the MDF is instead fixed by the values of tinf and α.
In particular, tinf determines the beginning of Phase II and
α its end. Their values (tinf = tf f (zvir)/4, α = 10) have
been selected in order to match the Sculptor MDF maxi-
mum/shape without altering the global MW properties and
Galactic halo MDF (the same parameter are in fact applied
to all the virialized MW building blocks). Finally, stars with
[Fe/H] >
∼ − 1.6, are formed during the feedback-dominated
Phase III. Note in particular that the value of the MDF cut-
off ([Fe/H]∼ −1.5) corresponds to the gas iron-abundance
at the blow-away.
At [Fe/H] >
∼ − 1.5 our model slightly underpredicts the
data as the theoretical MDF drops very steeply. The ex-
planation for such disagreement is likely to reside in our
simplified dynamical treatment of mechanical feedback. In-
terestingly, Mori, Ferrara & Madau (1999), investigated the
dynamics of SN-driven bubbles in haloes with M = 108M⊙
at z = 9 using 3D simulations. They found that less than
30% of the available SN energy gets converted into kinetic
energy of the blown away material, the remainder being ra-
diated away. A large fraction of gas remains bound to the
galaxy, but is not available to form stars before it cools and
rains back onto the galaxy after ∼ 200 Myr. Such effect is
not included in our modeling. Qualitatively we do expect
that such "galactic fountain" would increase the amount of
Fe-enriched gas to restart SF after blow-away, and hence the
number of [Fe/H]≥ −1.5 stars.
The total number of relics stars shown in the MDF cor-
responds to a total stellar mass of M∗ = (3 ± 0.7) × 106M⊙.
Using the total (dark+baryonic) dSph mass, derived from
our simulations M = (1.6 ± 0.3) × 108M⊙ we can compute
the mass-to-luminosity ratio
(cid:16) M
L∗(cid:17) = (cid:16) M
M∗(cid:17) ×(cid:16) M∗
L∗ (cid:17) ∼ 150 ,
(13)
having assumed (M∗/L∗) = 3, in agreement with the results
by Ricotti & Gnedin 2005. This result is consistent with the
most recent estimate for Sculptor (Battaglia 2007; Battaglia
et al. 2008), that gives a very high value (M/L) = 158 ± 33.
6.2 Color-magnitude diagram
Another comparison with data can be done in terms of color-
magnitude diagram (CMD) of the Sculptor stellar popu-
lation observed by Tolstoy et al. (2004). CMD represents
one of the best tools to study the star formation history of
a galaxy. Starting from our numerical results for a typical
dSph, we have computed the corresponding synthetic CMD
using the publicly available IAC-STAR code by Aparicio
& Gallart (2004). Given the IMF, the SFR and the ISM
8
Stefania Salvadori, Andrea Ferrara & Raffaella Schneider
Figure 4. Comparison between the Sculptor MDF observed by
Helmi et al. (2006), (points) and simulated one obtained by as-
suming tinf = tf f /4, α = 10, zrei = 6 (histogram). Error bars are
the Poissonian errors. The histogram is the averaged dSph MDF
over the surviving satellites (∼ 20) in all the 100 realization of
the merger tree (∼ 2000 objects). The shaded area represents the
±1σ Poissonian error.
metallicity evolution, IAC-STAR allows to calculate several
properties of the relic stellar population and, in particular,
the stellar magnitudes. We have used the stellar evolution
library by Bertelli (1994) and the bolometric correction li-
brary by Lejeune et al. (1997). Note that the IAC-STAR
input parameters for the ISM metallicity evolution must
be ZIAC−ST AR > 0.005Z⊙. No binary stars have been in-
cluded.
We adopt a randomization procedure in order to simu-
late the observational errors in the synthetic CMD and com-
pare numerical results with data. To this aim, we first derive
the normalized error distribution for the magnitude MI and
the color index V − I from the data sample by Tolstoy (pri-
vate communication). Errors have been randomly assigned
at every synthetic star, identified by a (MI , V −I) pair using
a Monte Carlo method and randomly added or subtracted.
Note that more accurate (and complicated) randomization
procedures exist (see for example Aparicio & Gallart 2004);
however, we consider the simple approach adopted here ade-
quate for our present purposes. In Fig. 5 we compare the syn-
thetic and observed CMDs. Data by Tolstoy et al. (∼ 10300
stars into the relevant MI , V − I range) have been normal-
ized to the total number of synthetic stars derived by IAC-
STAR (∼ 2300). In order to do so stars have been randomly
selected from the data sample.
The match between theoretical and experimental points
is quite good. We note however that the number of red giant
branch (RGB) stars in the synthetic CMD is lower than the
observed one. This discrepancy can be explained with the
contamination of the data sample by Galactic foreground
stars (see Tolstoy et al. 2004). The synthetic CMD repro-
duces reasonably well the blue/red horizontal branch stars
Figure 5. Comparison between the CMD of the Sculptor stellar
population observed by Tolstoy et al. (2004) (triangles) and the
synthetic CMD (open points) derived for a typical dSph galaxy
with total mass M = 1.6 × 108M⊙ which virialize at redshift
zvir = 7.2.
(BHB/RHB stars) i.e. stars residing in the CMD branch
(0 < MI < 1, 0 < (V − I) < 1). A well populated HB in the
CMD diagram might be interpreted as an indication of an
old stellar population (age > 10 Gyr). The interpretation of
the blue and red HB, on the contrary, is quite controversial:
due to the age-metallicity degeneracy of the CMD stellar
colors become bluer when stars are younger and/or poorer
in metallicity. For this reason the position of a star in the
CMD cannot be unequivocally interpreted. In our model the
majority of the stars are formed during the first 100 Myr of
the dSph life; this means that all the stars have basically the
same age >
∼ 13 Gyr; so the HB morphology, in our model,
reflects the metallicity gradient of the stellar populations:
BHB stars belong to metal-poor stars formed during the
Phase I (see Fig. 2) while RHB stars to the more metal-rich
stars formed during the Phase II.
6.3 Key abundance ratios
A method commonly used to break the age-metallicity de-
generacy and derive accurate SFH from the CMD diagrams,
is the analysis of the stellar elemental abundances. In most
of the observed dSph galaxies the abundance ratio of α el-
ements (O, Mg, Si, Ca) relative to iron ([α/Fe]) shows a
strong decrease when [Fe/H]> −2 (Venn et al. 2004). Since
α-elements are primarily generated by SN II while a substan-
tial fraction of iron-peak elements (Fe, Ni, Co) are produced
by type Ia SNe (SNe Ia), the decline of [α/Fe] is usually in-
terpreted as a contribution by SNe Ia. Using this argument
and the assumption that the lifetime of SNe Ia is around
1.5 Gyr, Ikuta & Arimoto (2002) inferred an age spead of 1-
2 Gyr in the dominant stellar population of Draco, Sextans
and Ursa Minor dSphs. However, the issue of the lifetime of
SNe Ia remains quite debated and uncertain, with timescales
c(cid:13) RAS, MNRAS 000, 1–??
Life and times of dwarf spheroidal galaxies
9
the Sculptor chemical evolution including differential winds.
In conclusion we find that the trend of [α/Fe] does not re-
quire a prolonged star formation phase (> 1 Gyr) but can
be satisfactorily explained even if 99% of the stars formed
during the first 100 Myr of the dSph lifetime.
Additional constraints on the SFH may also come from
the analysis of the abundances of s-elements associated with
the slow neutron-capture process. These are produced by
low-mass stars during Asymptotic Giant Branch (AGB)
phases. From an analysis of [Ba/Y] Fenner et al (2006) con-
cluded that most of the stars must be formed over an inter-
val of at least several Gyr to allow time for metal-poor AGB
stars to enrich the ISM up to the observed values. Our model
does not make specific prediction on s-elements; it is likely
however that since the bulk of stars is predicted to formed on
a time-scale of ∼ 100 Myr there would not be enough time
for the ISM to be enriched with the products of AGB stars.
Nevertheless this may not be the only scenario to explain s-
elements abundances. For example binary systems in which
the lower mass, long-living star, accretes s-enhanced gas di-
rectly from the companion rather than from the ISM can
equally explain the observed high s-elements abundances.
Internal production during the dredge-up phase can repre-
sent yet another possibility. These alternative scenarios are
supported by the observed stellar [s/Fe] ratio with respect to
[Fe/H]. The data (Venn et al. 2004) show that the [s/Fe] val-
ues do not increase at higher [Fe/H], as expected if the ISM
is gradually enriched by the contribution of lower mass stars.
Moreover, a large [s/Fe] spread is observed for any [Fe/H]
which is expected if the efficiencies of accretion, dredge-up
and s-element production are functions of stellar mass.
6.4 Dark matter content
DSph galaxies represent the most dark matter-dominated
systems known in the Universe. It is then very interesting to
determine their dark matter mass. Observationally, the mass
content of dSph galaxies is derived by measuring the velocity
dispersion profile of their stellar populations and comparing
it with the predictions from different kinematic models. The
latter step strongly depends on the adopted stellar kine-
matics (in particular the assumed velocity anisotropy radial
profile), on the dark matter mass distribution, and on the
nature of the dark matter itself.
Recently, Battaglia (2007), Battaglia et al. (2008) have
derived the velocity dispersion profile of Sculptor measuring
the velocities of ∼ 470 RGB stars. They model Sculptor as
a two component system with a metal poor and a metal
rich stellar population that show different kinematics. They
use these two components as distinct tracers of the same
potential and find that the best model is a cored profile
with rc = 0.5kpc and M (< rlast)k= (3.4 ± 0.7) × 108M⊙
which gives an excellent representation of the data assuming
an increasing radial anisotropy. Interestingly, the values of
M (< rlast) obtained assuming a NFW model for the dark
matter distribution or a constant radial anisotropy are M (<
rlast) = (2.4+1.1
−0.7) × 108M⊙ and M (< rlast) = (3.3 ± 0.8) ×
108M⊙, respectively, consistent with the above result within
1σ.
Figure 6. Oxygen-to-iron stellar abundance with respect to
[Fe/H] for the Sculptor dSph. Points refer to 8 Sculptor stars
observed by Gaisler et al. (2005) and Shetrone et al. (2003); con-
tours represent the probability, respectively equal to 0.6 and 0.8
and 0.99, to find a star into the [O/H]-[Fe/H] plane.
as short as 40 Myr having been suggested (see Ricotti &
Gnedin 2005 for a thorough discussion) under starburst for-
mation conditions.
In Fig. 6 we compare the oxygen-to-iron stellar abun-
dance with respect to [Fe/H] for 8 stars observed in Sculp-
tor by Gaisler et al. (2005) and Shetrone et al. (2003), with
the results of our model. In spite of the poor statistics the
data show a clear indication of the [O/Fe] decrement for
[Fe/H]> −1.8; in particular, subsolar values are observed.
Even if SNe Ia are not included in our model, a drop in
the [O/Fe] occurs as a result of having released the IRA
approximation (see also Fenner et al. 2006) and of differ-
ential winds. However, subsolar [O/Fe] values can only be
accounted for by differential winds. This is because when
[O/Fe] reaches the maximum value¶, the "effective oxygen
yield" (dYO/dt − Zw
O dMej /dt) is strongly reduced with re-
spect to iron due the effect of differential winds, which,
as can be deduced from eq. (12), have a larger impact on
more abundant elements (Zw
). This causes a pro-
nounced and rapid decrease of [O/Fe] to subsolar values. We
note that model results tend to underpredict the observed
stellar abundances: essentially, differential winds are too ef-
ficient. We recall, however, that the value α = 10 have been
selected in order to match the Sculptor MDF. The prob-
lem of the lack of [Fe/H]> −1.5 stars noted in Sec. 6.1, is
evident also here. A more sophisticated treatment of dif-
ferential winds and/or the inclusion of the missing physical
effects discussed in Sec. 6.1, should presumably remove such
discrepancies. Noticeably the result of Fig. 6 is fully consis-
tent with the analysis by Fenner et al. (2006), who studied
i = αZISM
i
¶ The relative production rate of oxygen with respect to iron is
larger in low-mass SNe II.
k M (< rlast) is the mass enclosed within the last measured point.
c(cid:13) RAS, MNRAS 000, 1–??
10
Stefania Salvadori, Andrea Ferrara & Raffaella Schneider
The average mass of dSph galaxies that we infer from
our simulations, M = (1.6 ± 0.8) × 108M⊙. Because the last
measured points in the Battaglia (2007), Battaglia et al.
(2008) typically reach 1-2 kpc, one could suspect that ad-
ditional dark matter could be located outside this radius,
thus turning their determination into a lower mass limit.
However, for the mean mass value and formation redshift
that we have obtained M ≈ 108M⊙, zvir ≈ 7, the virial ra-
dius of such a halo is 1 kpc. Thus, the agreement (at 1 − 2σ
level) between our prediction and the actual mass deter-
minations might not be coincidental, but reflects the fact
that in these small objects star formation has propagated up
to the most remote galactocentric regions. This prediction
could be eventually checked by deeper observations and/or
other techniques.
The narrow dispersion around the average mass value
found is an indication that dSphs may have a universal host
halo mass. This finding agrees with the results by Mateo et
al. (1998), Gilmore et al. (2007) and Walker et al. (2007),
who suggest that dSph galaxies might have a common mass
scale M0.6 = (2 − 7) × 107M⊙, where M0.6 is the dark mat-
ter mass within a radius of 0.6 kpc. Assuming a NFW den-
sity profile, zvir = 0 and a concentration parameter c = 35
(Battaglia 2007) we found M0.6 = 2.3 × 107M⊙.
6.5 Gas footprints of feedback
A final comparison with data can be done in terms of the
observed gas properties. In the previous Sections we have
shown that metal-enhanced winds driven by SN explosions
play a fundamental role in determining the evolutionary
times scales and properties of a dSph galaxy. Based on ob-
servations obtained with the Chandra X-Ray Observatory,
Martin, Kobulnicky & Heckman (2002) provide the first di-
rect evidence for metal-enhanced winds from dwarf starburst
galaxies. They have observed the hot X-ray-emitting gas
around the nearby dwarf galaxy NGC 1569 which entered
in a starburst phase (10-20) Myr ago. The X-ray spectrum
they find presents strong emission lines from α-process ele-
ments, that require the wind metallicity to be Zw > 0.25Z⊙
i.e. larger than ZISM = 0.2Z⊙, supporting our assumption
of metal enhanced winds Zw = 10ZISM . In particular, their
best fit models predict the ratio of α-elements to Fe to be
2-4 times higher than the solar value; it is then likely that
the ISM is preferentially depleted in α-elements consistent
with the findings shown in Fig. 6. We stress that these obser-
vations confirm the idea that mechanical feedback processes
start to play a significant role in the dSph evolution on a
very short time-scale (10-20 Myr after the beginning of the
starburst phase).
Alternatively, the efficiency of mechanical feedback pro-
cesses can be tested using observations of neutral hydrogen
(HI). The Local Group dSph galaxies are all relatively HI
poor (Mateo 1998) suggesting that little gas has remained
after the main SF phase. Within the known dSph galaxies,
Sculptor is one of the few with detectable HI emission. Us-
ing radio observation, Carignan et al. (1998) derived a lower
limit for the HI mass of MHI > 3 × 104M⊙. Our simulation
predicts an average mass of gas Mg = (2.68±0.97)×104 M⊙,
in very good agreement with the observed value if indeed
this gas is in neutral form. According to our model, the HI
mass detected in Sculptor can be associated to gas returned
by evolved stars, an explanation also offered by Carignan et
al. (1998).
7 SUMMARY AND DISCUSSION
We have proposed a global scenario for the formation and
evolution of dSph galaxies, satellites of the MW, in their
cosmological context using an improved version of the semi-
analytical code GAMETE (GAlaxy Merger Tree & Evolu-
tion, SSF07). This approach allows to follow self-consistently
the dSph evolution and the MW formation and match, si-
multaneously, most of their observed properties. In this con-
text dSphs formed within the Galactic environment, whose
metallicity evolution depends on the history of star for-
mation and mechanical feedback along the build-up of the
Galaxy. The star formation and mechanical feedback effi-
ciencies of dSphs are assumed to be the same as for all the
Galactic building blocks; they are calibrated to reproduce
the observable properties of the MW.
DSph candidates are selected among the MW progen-
itors following a dynamical and a reionization criteria; we
choose haloes with masses (i) M4 < M < M2σ if z > zrei,
(ii) M30 < M < M2σ if z < zrei i.e. we assumed the forma-
tion of galaxies with circular velocity vc < 30 km/s to be
suppressed after reionization, where 5.5 < zrei < 10 (Choud-
hury & Ferrara 2006). As the number of dSph candidates
found varies with zrei, we determine the fraction that will
become MW satellites requiring that their number matches
the observed one (∼ 15). Once formed, dSphs are assumed
to evolve in isolation with respect to the merging/accreting
Galaxy. In this work we present the results obtained assum-
ing zrei = 6. This value provides a good agreement between
the Sculptor MDF and the simulated one and gives a total
number of dSph candidates of Ntot ∼ 200; hence, we suppose
that ∼ 10% of them become MW satellites.
The results of our model, supported by the comparison
with observational data and previous theoretical studies, al-
low to sketch a possible evolutionary scenario for dSphs.
In our picture dSph galaxies are associated with Galac-
tic progenitors corresponding to low-sigma density fluctu-
ations (M4 < M < M2σ), that virialize from the MW
environment before the end of reionization, typically when
z = 7.2 ± 0.7. Their total (dark+baryonic) mass results to
be M = (1.6 ± 0.7) × 108M⊙.
At the virialization epoch the dSph birth environment
is naturally pre-enriched due to previous SN explosions up
>
to [Fe/H]GM
∼ − 3, a value fully consistent with that in-
ferred from observations by Helmi et al. (2006). The sub-
sequent dSph evolution is strongly regulated by mechanical
feedback effects, more intense in low mass objects (MacLow
& Ferrara 1999). We take winds driven by SN explosions to
be metal enhanced (Zw = 10ZISM ) as also confirmed by
numerical simulations (Fujita et al. 2004) and by the X-ray
observations of the starburst galaxy NGC1569 (Martin et
al. 2002). Typically, ∼ 100 Myr after the virialization epoch
a complete blow-away of the gas caused by mechanical feed-
back is predicted. The 99% of the present-day stellar mass,
M∗ = (3 ± 0.7) × 106M⊙, is expected to form during the first
100 Myr. The stellar content of dSphs is then dominated by
an ancient stellar population (> 13 Gyr old), consistent with
c(cid:13) RAS, MNRAS 000, 1–??
the analysis of the dSph CMD diagrams by Dolphin et al.
(2005).
After the blow-away the galaxy remains gas-free and
SF is stopped. Fresh gas returned by evolved stars allows to
restart the SF ∼ 150 Myr after the blow-away. The SFR,
however, is drastically reduced due to the paucity of the
returned gas. Mass loss from evolved stars has been also
invoked by Carignan et al. (1998) to explain the detection
of HI in the Sculptor dSph. About ∼ 100 Myr later, a sec-
ond blow-away occurs and the cycle starts again. Such in-
termittent SF activity is similar to those observed in Ca-
rina by Smecker-Hane et al. (1994) and to the one de-
rived by Stinson et al. (2007) using numerical simulations.
Roughly 1 Gyr after the virialization this burts-like SFH
ends while the SF activity proceeds until the present-day
with a rapidly decreasing rate. At z = 0 the dSph gas con-
tent is Mg = (2.68 ± 0.97) × 104M⊙.
Our model allows to match several observed properties
of Sculptor dSph:
• The Metallicity Distribution Function (Helmi et al.
2006). The pre-enrichment of the dSph birth environment
accounts for the lack of observed stars with [Fe/H]< −3,
a striking and common feature of the four dSph galaxies
observed by Helmi et al. (2006).
• The stellar Color Magnitude Diagram (Tolstoy et al.
2003) and the decrement of the stellar [O/Fe] abundance
ratio for [Fe/H]> −1.5 (Gaisler et al. 2005, Shetrone et al.
2003). The agreement found between models and observa-
tions support the SFH we have predicted.
• The DM content M = (3.4 ± 0.7) × 108M⊙ and the high
mass-to-light ratio (M/L) = 158 ± 33 recently derived by
Battaglia (2007), Battaglia et al. (2008); we find (M/L) ∼
150 using the predicted dark matter to stellar mass ratio
and assuming (M/L)∗ = 3.
• The HI gas mass content. The value derived by radio
observations (MHI > 3 × 104M⊙, Carignan et al. 1998) is in
agreement with our findings.
Interestingly, the model can also be used to put an up-
per limit on epoch of reionization, zrei < 8.5. The total
number of selected dSph candidates in fact, is reduced be-
low the observed one (Ntot ∼ 5) if zrei = 8.5. In addition,
the imprint of reionization lies in the suppression of dSphs
formation below the reionization redshift, zrei = 6. This re-
sult is fully consistent with the presence of an ancient stellar
population (> 13 Gyr old) in all the observed dSph galaxies
(Grebel & Gallagher 2004).
Despite the success of the model in producing a coher-
ent physical scenario for the formation of dSphs in their
cosmological context and matching several of the Sculptor
and MW properties, several aspects deserve a closer inspec-
tion. Although Sculptor represents the best template to com-
pare with because of its average properties and the large
amount of available data, examples of deviations are already
known. In particular the SFHs differ considerably among
dSph galaxies (Dolphin et al. 2005, Grebel & Gallagher
2004). The Fornax CMD diagram, for example, indicates a
massive presence of younger stars than in other dSphs (Dol-
phin et al. 2005; Stetson et al. 1998; Buonanno et al. 1999);
the peculiarity of this object is also evident in the observed
MDF, which is a monotonically increasing function up to
[Fe/H]∼ −1 ( Pont et al. 2004; Helmi et al. 2006; Battaglia
c(cid:13) RAS, MNRAS 000, 1–??
Life and times of dwarf spheroidal galaxies
11
et al. 2006). The dSph properties inferred in our model, in-
cluding the SFH and the MDF, are instead "Universal". This
is a consequence of the selection criteria, that gives a Uni-
versal dSph host halo mass, and of the assumed cosmological
gas fraction in all virializing haloes. Since M4(z) < M2σ only
for z < 9 (see Fig. 1, right panel) and the typical mass of
newly virializing halo is ∼ M4(z) < M30, dSphs are forced
to form in the redshift range 6 < zrei < 9. Due to the small
variation of M4(z) in such a range, the dSph dark matter
content is very similar in all objects and equal to ∼ 108M⊙.
A refinement of the reionization criterion should allow
larger deviations from the average evolutionary trend with-
out altering it and possibly allow reionization imprints to
emerge in their SFHs. Cosmological simulations by Gnedin
(2000) show in fact that after reionization the gas fraction in
haloes below a characteristic mass-scale is gradually reduced
compared to the cosmic value. If included in our model such
prescription might allow to form, just after the reionization
epoch, massive dSphs (M ≥ M4(z = 6)) with a lower initial
gas-to-dark matter ratio. Since mechanical feedback depends
on the available mass of gas powering the SF and on the halo
binding energy, this should translate in a less efficient me-
chanical feedback and a consequent more regular SF activ-
ity. In particular, if blow-aways do not occur, the SF could
proceed until the present day with a higher rate, allowing
massive formation of younger and more Fe-rich stars. Alter-
natively, random episodes of mass accretion and/or merging
with primordial composition haloes, can be invoked as exter-
nal gas sources to power the SF at lower redshifts. Another
physical mechanism, tidal stripping by the gravitational field
of the Galaxy, might be invoked (Ibata et al. (2001); Mayer
et al. (2004, 2006)) to explain the paucity of remnant gas in
dSph, and perhaps as a mechanism of star formation sup-
pression. In our model, we see no need to resort to such effect
as the large majority of the gas is expelled by SN feedback
within the first 100 Myr of dSph evolution (we recall that the
amount of newly born stars after that time is only ≈ 1% of
the final stellar mass). Hence, by the time dSphs find them-
self embedded in the MW gravitational potential, there is
little gas left to be stripped.
Finally, we like to comment on some of the model as-
sumptions that could affect the results of the present work.
The most relevant one certainly resides in the perfect mixing
approximation which determines the metallicity evolution of
the dSph birth environment and therefore of the low [Fe/H]
tail of the dSph MDF. Our persisting ignorance on all the
physical effects regulating such process cannot allow to im-
prove this assumption at the moment. The problem, how-
ever, is partially alleviated by the spread of the GM metallic-
ity evolution (Fig. 1, middle panel) induced by the stochastic
nature of the merger histories, which appears to be simi-
lar to what found by sophisticated numerical simulations
of mixing in individual galaxies (Mori & Umemura 2006).
DSph galaxies formed out of a metal-poor birth environment
[Fe/H]< −3 are found in our model; however, their num-
ber is very small and their statistical impact on the average
MDF negligible. Finally, the assumed PopIII IMF (here a δ-
function centered in mP opIII = 200M⊙) might in principle
affect the chemical evolution of the MW environment given
the large iron production of massive stars (see also SSF07
for a more detailed description). A comparison of results ob-
tained using PopIII masses in the range (140 − 260)M⊙ and
12
Stefania Salvadori, Andrea Ferrara & Raffaella Schneider
a Larson IMF, shows that the GM evolution is independent
of the assumed PopIII IMF below z = 10. The dSph birth
environment is therefore not affected by this hypothesis. We
refer the reader to SSF07 for a detailed analysis of the PopIII
IMF impact on the Galactic halo MDF.
ACKNOWLEDGEMENTS
We are grateful to G. Battaglia, N. Gnedin, E. Grebel, A.
Helmi, E. Tolstoy and K. Venn for providing us their data
and enlightening discussions. This work has made use of the
IAC-STAR Synthetic CMD computation code. IAC-STAR
is supported and maintained by the computer division of
the Instituto de Astrofsica de Canarias. We are grateful to
DAVID⋆⋆ members for discussions.
REFERENCES
Aparicio A., Gallart C., 2004, AJ, 128, 1465
Barkana R., Loeb A., 2001, Phys. Rep., 349, 125
Battaglia G. et al., 2006, A&A, 459, 423
Battaglia G., 2007, Phd Thesis, University of Groningen, The
Netherlands
Battaglia G., Helmi A., Tolstoy E., Irwin M. J., in preparation
Beers T. C., Christlieb N., 2006, private communication
Bertelli G., Bressan A., Chiosi C., Fagotto F., Nasi E., 1994, A&A,
106, 275
Binney J., Merrifield M., 1998, Galactic Astronomy. Princeton
Univ. Press, Princeton
Bond J. R., Cole S., Efstathiou G., Kaiser N., 1991, ApJ, 379,
440
Bromm V., Ferrara A., Coppi P. S., Larson R. B., 2001, MNRAS,
328, 969
Bromm V., Loeb A., 2004, New Astron., 9, 353
Bullock J. S., Kravtsov A. V., Weinberg D. H., 2001, ApJ, 548,
33
Buonanno R., Corsi C. E., Castellani M., Marconi G., Fusi Pecci
F., Zinn R., 1999, AJ, 118, 1671
Carignan C., Beaulieu S., Cote S., Demers S., Meteo M., 1998,
ApJ, 116, 1690
Choudhury T. R., Ferrara A., 2006, MNRAS, 371, L55
Christlieb N., et al., 2002, Nat, 419, 904
Christlieb N., Bessel M. S., Eriksson K., 2006, in preparation
Christlieb N., 2007, private communication
Ciardi B., Ferrara A., 2005, Space Sci. Rev., 116, 625
Cole S., Lacey C. G., Baugh C. M., Frenk C. S., 2000, MNRAS,
319, 168
Diemand J., Madau P., Moore B., 2005, MNRAS, 364, 367
Dolphin A. E., Weisz D.R., Skillman E.D., Holtzman J.A., 2005,
to appear in ed. Valls-Gabuad D. & Chavez M., Resolved Stel-
lar Populations, ASP Conf.Ser. [arXiv:astro-ph/0506430]
Fenner Y., Gibson B. K., Gallino R., Lugaro M., 2006, ApJ, 646,
184
Ferrara A., Tolstoy E., 2000, MNRAS, 313, 291
Frebel A., Christlieb N., Norris J. E., Aoki W., Asplund M., 2005,
Nat, 434, 871
Gilmore G., Wilkinson M., Kleyna J., Koch A., Evans N. W.,
Wyse R. F. G., Grebel E. K.et al., 2007, Nucl. Phys. B, in
press (astro-ph/0608528)
Gnedin N. Y., 2000, ApJ, 542, 535
Gnedin N. Y., Kravtsov A. V., 2006, ApJ, 645, 1054
Grebel E. K., Gallagher III J. S., 2004, ApJ, 610, L89
Heger A., Woosley S. E., 2002, ApJ, 567, 532
Helmi A. et al., 2006, ApJ, 651, L121
van den Hoek L. B., Groenewegen M. A. T., 1997, A&A, 123, 305
Hurley-Keller D. A., Mateo M., Grebel E. K., 1999, ApJ, 523,
25L
Keres D., Katz N., Weinberg D. H., Dav´e R., 2005, MNRAS, 363,
2
Kravtsov A. V., Gnedin O. Y., Klypin A. A., 2004, ApJ, 609, 482
Ikuta C., Arimoto N., 2002, A&A, 391, 55
Ibata R. A., Lewis G. F., Irwin M., Totten E., Quinn T., 2001,
ApJ, 551, 295
Lacey C., Cole S., 1993, MNRAS, 262, 627
Lanfranchi G., Matteucci F., 2007, A&A, 468, 927
Larson R. B., 1974, MNRAS, 169, 229
Larson R. B., 1998, MNRAS, 301, 569
Lejeune Th., Cuisinier F., Buser R., 1997, A&A, 125, 229
Mac Low M.-M., Ferrara A., 1999, ApJ, 513, 1421
Martin C., Kobulniky H., Heckman T., 2002, ApJ, 57413, 1421
Mateo M., 1998, ARA&A, 36, 435
Mayer L., Moore B., Quinn T., Governato F., Stadel J., 2002,
MNRAS, 336, 119
Mayer L., Mastropietro C., Wadsley J., Stadel J., 2006, MNRAS,
369, 1021
Mets M., Kroupa P., 2007, MNRAS, 376, 387
Mihos J. C., Hernquist L., 1996, ApJ, 464, 641
Moore B., Diemand J., Madau P., Zemp M., Stadel J, 2006, MN-
RAS, 368, 563
Mori M., Umemura M., 2006, Nat, 440, 644
Mori M., Ferrara A., Madau P., 2002, ApJ, 571, 40
Omukai K., 2000, ApJ, 534, 809
Omukai K., Tsuribe T., Schneider R., Ferrara A., 2005, ApJ, 626,
627
Pont F., Zinn R., Gallart C., Hardy E., Winnick R., 2004, ApJ,
1247, 840
Read J. I., Pontzen A. P., Viel M., 2006, MNRAS, 371, 885
Ripamonti E., Tolstoy E., Helmi A., Battaglia G., Abel T., 2007,
EAS Publications Series, 24, 15
Ricotti, M., Gnedin N. Y., 2005, ApJ, 629, 259
Salvadori S., Schneider R., Ferrara A., 2007, MNRAS, 381, 647
Schneider R., Ferrara A., Natarajan P., Omukai K., 2002, ApJ,
571, 30
Schneider R., Ferrara A., Salvaterra R., Omukai K., Bromm V.,
2003, Nat, 422, 869
Schneider R., Omukai K., Inoue A. K., Ferrara A., 2006, MNRAS,
369, 1437
Schneider R., Salvaterra R., Choudhury T. R., Ferrara A., Buri-
gana C., Popa L. A., 2007, MNRAS, submitted
Sheltrone M., Venn K., Tolstoy E., Primas F., Hill V., Kaufer A.,
2003, AJ, 125, 684
Smecker-Hane T. A., Stetson P. B., Hesser J. E., Lehnert M. D.,
1994, AJ, 108, 507
Stinson G. S., Dalcanton J. J., Quinn T., Kaufmann T., Wadsley
J., 2007, ApJ, 667, 170
Fujita A., Mac Low M. M., Ferrara A., Meiksin A., 2004, ApJ,
Stetson P. B., Hesser J. E., Smecker-Hane T. A., 1998, Astronom-
613, 159
ical Society of the Pacific, 110, 533
Ganguly R., Sembach K. R., Todd T. M., Savage B. D., 2005,
Tassis K., Abel T., Bryan G. L., Norman M. L., 2003, ApJ, 587,
ApJ, 157, 251
Geisler D., Smith V. V., Wallerstein G., Gonzales G., Charbonnel
C., 2005, AJ, 129, 1428
⋆⋆ www.arcetri.astro.it/science/cosmology/index.html
13
Tolstoy E., 2004, ApJ, 617, L119
Vader J. P., 1986, ApJ, 305, 669
Venn K. et al., 2004, AJ, 128, 1177
Volonteri M., Haardt F., Madau P., 2003, ApJ, 582, 559
Woosley S. E., Weaver T. A., 1995, ApJ, 101, 181
c(cid:13) RAS, MNRAS 000, 1–??
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.